Business Cycle Essays This collection of Tables and Essays can be accessed from http://www.albany.edu/cer/bc/

The Stock Market, Oil Price Shocks, Economic Recessions and the Business Cycle
With An Emphasis on Forecasting

Some Featured Highlights

Edward Renshaw
Professor of Economics
State University of New York at Albany

December 2002

There is a possibility that the U.S. economy may experience a double dip recession as far as non farm payroll employment is concerned.

While no two recessions are exactly alike, they all seem to have the property of being triggered by inflation and/or unsustainable increases in one or more components of GDP.

Government spending is one variable that should not be ignored. There have been economic recessions associated with all of the major wars that the U.S. has participated in during the last century. The most interesting aspect to the Gulf War of 1990-91 is that the U.S. recession began in August of 1990, before our soldiers began to fight, rather than after the war was over. This would suggest that the mere threat of another Gulf War might have a negative effect on the U.S. economy in the year 2003.

The payroll employment contraction from March 2001 to April 2002 was moderated to a considerable extent by an increase in government employment amounting to almost half a million new jobs. Budget deficits have become such a problem at the state and local level, however, that government lay-offs might have a negative impact on employment during 2003.

Higher prices for imported oil, compared to last year, could also have a negative impact on consumer spending for other purposes this winter. See Essay 6.

The industrial recession which began in October 2000 was triggered in part by an unsustainable increase in spending by business enterprises for new computers and software. The improvements in labor productivity associated with this "technological revolution" has helped to engineer the weakest recovery in payroll employment for any recession in the post 1947 period and raises the possibility that the depression in employment will last for a very long time.

Two of the most interesting components of the Federal Reserve's index of industrial production are lumber (most of which is used for residential construction) and consumer automotive products. When the U.S. was not engaged in a prolonged war, as was the case in 1951-52 and 1966-67, year-to-year declines in these two variable have either led or been associated with recessions dated by NBER.

Lower interest rates have kept residential constrution unusually strong until recently. But with little or no growth in employment, that is not likely to remain the case forever.

The sale of automobiles has also been elevated by deep discounts and "zero percent financing". Without good jobs, however, most young people are not going to be able to afford new cars. Intense competition from foreign auto producers could also have a negative impact on auto production in this country.

The unanswered question is what should the federal government do to revive economic activity? One of the most efficient approachs would be for Congress to subsidize a prompt acceleration of expenditure for public works.

The emphasis at the present time appears to be more tax cuts. Tax cuts are not very efficient at moderating recessions, however, since many businesses and persons who are most impacted by a recession do not have any taxable income.

On August 4, 1981 Congress enacted the largest tax cut bill in U.S. history up to that time. The economic recession which began in that month did not end until 16 months later. One of the problems with large tax cuts, from a recessionary perspective, is that they encourage reductions in government employment and expenditure for other purposes.

The recession of 1990-91 was aided in part by a 26.2 percent reduction in real spending by the federal government for new structures during the trough year of 1991. This may have been one of the factors that enabled a little known governor from Arkansas to oust an incumbent president. See Table 1.5.

In the early post World War II period increases in the value of new public structures put in place during years containing a recessionary trough in business activity did help to moderate recessions. Since the recession of 1960-61, however, expenditure for public works has had a negative effect on economic recoveries, more often than not. In Albany, New York and many other capital cities in the USA construction budgets have been routinely robbed or short changed during recessions to hide budget deficits which are unconstitutional in most states.

During the great depression of the 1930s John Maynard Keynes suggested that the U.S was suffering "from a bad attack of economic pessimism." One of his proposed cures was an increase in expenditures for public works. "Public works even of doubtful utility may pay for themselves over and over again at a time of severe unemployment, if only from the diminished cost of relief expenditures." This was not a new idea. Philip Wicksteed, who died in 1927, had long recommended that the state should undertake public works in "slack periods when they can be executed at least expense, and will, at the same time, have a tendency to counteract a serious evil."

The multipliers that can be derived from models of the U.S. economy would suggest, in any event, that additional expenditure for public works have a first year effect that is at least twice as large as the multiplier for most tax cuts and will keep on yielding benefits for many years in the future. See Essay 8.

Crumbling infrastructure, moreover, is now perceived to be a serious problem in many parts of the U.S. An Associated Press computer analysis of Federal Highway Administration records found 29 percent of U.S. bridges to be deficient as of August 31, 2000.

An additional person employed in the construction of highways and bridges can be expected to put at least one other person to work outside the construction industry during recessions and generate enough additional tax revenue within one year to cover more than half of the budgetary cost.

The Federal Reserve began to lower its funds rate rather aggressively in January 2001 with the hope of preventing another recession. The life cycle hypothesis, however, implies that persons who want to even out their consumption over an entire life time should save more when real interest rates are low and save less when they are high. This type of behavior makes it easier for one to understand why personal saving rates often increase during recessions and why monetary policy has acquired the reputation of only working after a long and variable lag. For retired persons who have deposited most of their assets in savings accounts with flexible rates, monetary easing can have a very harmful affect on consumption.

Earlier comments with regaard to the current recession can be found at the end of the following index.


A Tabular Index to the Essays and Tables

Essay 1: The Duration of Business Expansions and Depressions.

Essay 2: Economic Recessions and Those Sometimes Misleading Indicators.

Essay 3: The Stock Market and the Business Cycle.

Essay 4: Economic Recessions and Interest Sensitive Components of GDP.

Essay 5: Employment Recessions and Their Duration.

Essay 6: Can There Be Another Recession Without an Oil Price Shock?

Essay 7: The Accelerator Principle Revisited.

Essay 8: Rehabilitating the Keynesian Multiplier.

Essay 9: The Political Business Cycle, the Fed's War on Inflation and the Targeting of Nominal GDP.

Some Earlier Commentary Pertaining to the 2001-0? Recession

July 2002

The outlook for the U.S. economy is now worse than was anticipated in March 2002. BEA has now revised its accounts to show three quarterly dips in real GDP during 2001 instead of only one. There has been hardly any increase in payroll employment, so far. The most prolonged stock market crash in the post World War II period has ruined many stock options and lowered the demand for pricey new homes. With many state and local governments cutting expenditures and laying off workers to help balance their budgets there is a possibility of a "double dip" recession and the most prolonged depression in employment since the great depression of the 1930s.

March 2002

There is a possibility that the NBER recession which began in March 2001, with the peaking out of payroll employment, may have ended in December 2001 or January 2002. If January marks the bottom there have only been six shorter recessions since NBER began dating peaks and troughs that go back to 1854.

The recession of 2001-02, however, may be recorded in some history books as the "mildest" since BEA began calculating real GDP on a quarterly basis starting in 1947. There has so far only been one recorded quarterly dip in real GDP since 1993. The nine preceding recessions are now characterized as having experienced at least two lower quarterly values for real GDP expressed in chained 1996 dollars.

The mildness of the current recession is related in part to a very warm winter in the north eastern U.S. The savings in heating oil in conjunction with a grudging reduction in the price of oil on the part of OPEC has enabled consumers to spend more on goods and services produced in this country.

Another factor that helped to produce a mild recession was a propensity on the part of U.S. auto producers to offer new car buyers large discounts or cheap financing rather then let their sales plunge as much as was the case during previous recessions. The "zero" financing was aided to a considerable degree by an aggressive lowering of short term interest rates by the Federal Reserve beginning in January 2001 when it became clear that the industrial recession, which began in October 2000, might not be very mild. Lower interest rates also helped to keep housing and other types of construction from tumbling as much as was the case in previous recessions. The tax cuts enacted at the Federal level may have also helped to keep consumer spending from declining.

Another factor contributing the mildness of the current recession was a lot of pent up demand for workers on the part of business enterprises that were not adversely affected by the slump in demand for nonresidential structures, equipment and software. Payroll employment did not peak out until six months after the October peak in industrial production.

While more workers have lost their jobs in this recession than was the case during the recessions of 1960-61, 1969-70 and 1980, the impact has been moderated by a greater propensity to retire older workers who were eligible for social security and other types of retirement benefits.

The destruction of the World Trade Center in September 2001 may have also had a favorable effect on employment by bolstering the demand for clean-up workers and by increasing the need for security personnel, equipment and software, and weapons of various sorts to wage an on going war on terrorism.

One would hope that the employment recession is over. It should be appreciated, however, that there have sometimes been one or two upward blips in the quarterly values for real GDP during recessions that were followed by another quarterly decline. See Table 7.2.

Many states and local governments are now experiencing financial difficulties and may decide to cut spending and employment rather than raise taxes to balance their budgets. With stock prices, long term interest rates and oil prices on the rise there is also a possibility that some consumers will decide to save more by cutting back spending or use more of their income to purchase vehicles with greater fuel efficiency that are produced in other countries.

Continued growth in consumer consumption and incredibly low personal savings rates, in any event, are the primary reasons for the mild recession so far.

With many high school and college graduates finding it very difficult to obtain good jobs, however, and with the capacity utilization rate in U.S. industry at about the lowest level in recorded history, it is not likely that business investment will recover at a very rapid rate in the near future.

November 2001

On Monday, November 26, the NBER dating committee finally decided that the longest expansion in business cycle history ended with the peaking out of payroll employment during March 2001. The unanswered question is how long will the current recession last?

Since the 43 month recession from August 1929 to March 1933 the recessions dated by the National Bureau of Economic Research (NBER) have ranged in duration from only six months to sixteen months for the 1973-75 and 1981-82 recessions. See Table 1.1. If the more recent range is not exceeded the current recession should end on or before June 2002.

An analysis of some leading and coincident indicators may provide at least a little ground for greater optimism. The coincident indicator with the best track record for identifying the recessionary bottoms decided upon by the NBER dating committee is the Federal Reserves index of industrial production. In the post 1947 period this indicator has (so far) always come within one month of identifying the recessionary troughs dated by NBER.

Since 1947 industrial recessions have never lasted more than 17 months. If this relationship continues to hold, the current industrial recession, which began in October 2000, should be over on or before February 2002.

Further support for this conclusion can be obtained by examining the behavior of both industrial production and payroll employment in the vicinity of the peaks identified by NBER. See Table 1.6 and Table 5.3.

It should be noted, however, that recessions with relatively small first quarter declines in real gross private domestic investment tend to be of longer duration than those that get off to a steeper start. See . The same sort of pessimistic conclusion has been applicable to payroll employment recessions. See Table 5.7.

The leading economic indicator with the best track record at identifying recessionary troughs before they occur is the S&P composite stock price index. The stock market crash of 2000-01 now has the distinction of being the second worst dip in this index in the post World War II period. It, like the worst crash of 1973-74, is also associated with a prolonged decline in industrial production.

The data in column (4) of Table 3.4 show the recessionary lead times for the S&P composite for the nine completed industrial troughs experienced since 1945. All of these lead times fall in the comparatively narrow range of from four to seven months. The lead times have been longer, on the average, for those recessions associated with U.S. participation in wars of various kinds, than for recessions which have occurred in more peaceful times.

If the current recession turns out to have characteristics similar to the preceding three war time recessions, industrial production can be expected to bottom out in the vicinity of February to April of 2002. One would hope that the current recession will end before then. The nice thing about this prediction, however, is that it can be used as a standard for helping to measure the curative properties of recent federal tax cuts and the lower interest rates that have been engineered by the Federal Reserve this year.

Optimists can perhaps take some comfort in the fact that the following lead times to an industrial trough, after a bottoming out of the stock market, have been shorter, on the average, for cases where industrial production had already been in a state of decline for at least nine months and for cases where the Fed funds rate was down 35 percent or more from its cyclical peak.

October 2001

While lower interest rates have usually helped to bolster housing construction, they can also have harmful effects. For conservative retired persons who have kept most of their financial assets in money market funds a dramatic reduction in short term interest rates can have negative effects on income and consumption. It is not unreasonable to suppose, moreover, that sophisticated home owners and business enterprises that refinance outstanding debt will use a portion of the savings to purchase stock at bargain prices, rather than increase their consumption or investment in new goods and services. The same conclusion might very well be applicable to tax cuts.

September 2001

There was evidence in support of an old fashioned NBER type of recession even before the destruction of the World Trade Center on September 11.

Industrial production peaked out way back in September 2000, however, and has been drifting down at a rapid enough rate to leave little doubt that the U.S. has been experiencing a manufacturing recession.

Payroll employment peaked out in March 2001 and had plunged enough in August to cause most economists to substantially lower their growth forecasts for the rest of the year. From 1947-2000 the U.S. economy always experienced an NBER peak or was headed down in that type of recession after four lower closing values for this indicator.

The civilian unemployment rate is another useful recession indicator. From 1947-2000 the U.S. economy was always mired down in an NBER type of recession after a cumulative increase in this variable of .9 percentage points. See Table 5.2. During August 2001 the civilian unemployment rate was up one full percentage point from the preceding low.

The recessions dated by NBER in the post 1947 period can all be identified on the basis of a second lower quarterly total for real GDP. The second lower total must be smaller than the first but doesn't have to immediately follow a first quarterly decline in real GDP. See Table 7.2. This hasn't happened yet but might happen before the year is over.

It should be appreciated that most forecasting systems and BEA's preliminary projections of late occurring information have an upward bias in the vicinity of recessionary peaks. Real GDP has been increasing at an annualized rate of less than two percent since the second quarter of 2000. This is the most prolonged period of stagnant growth in the history of the NIPA accounts. See Table 7.2. In August 2001 BEA reduced its estimated growth rates for 1998, 1999 and 2000 and lowered the second quarter annualized growth rate for 2001 to only .2 percent. As more and better information becomes available there is a possibility that the second quarter growth rate might be lowered into negative territory.

From 1948-99 the U.S. economy was always in the midst of an NBER recession after five lower monthly closing values for industrial production. See Table 1.6. There was a recurrence of that type of signal in February, 2001.

Since the Fed began to fight inflation in the late 1950s, by raising its funds rate by three percentage points or more during economic expansions, payroll peaks have lagged industrial peaks, more often than not, with a median lag in the three to five month range.

The good news is that employment declines which began four or more months after a peak in industrial production have been both milder and of shorter duration than those with peaks of a more coincident nature. See Table 5.3.

The bad news is that employment recessions with a very weak decline in the first three months of the recession have lasted longer than those with a more severe decline. See Table 5.7. It will be interesting to see if the Fed's rather aggressive reduction in short term interest rates and the President's tax rebates alter this relationship.

Some even worse news, perhaps, is that economic depressions are not getting shorter, on the average, if viewed from the perspective of how long it takes a "coincident" indicator such as industrial production or payroll employment, to recover to another new high. See Table 1.3. There are a number of economic and financial distortions, in any event, which might keep economic activity depressed for a long time.

On September 7, 2000 President Clinton warned Crown Prince Abdullah of Saudi Arabia that high oil prices might tip the U.S. economy into another recession. The ten poorest growth years for the U.S. economy from 1948-2000 all followed a year-year increase in the average price of crude oil of five percent or more and an anemic increase in June-December industrial production amounting to .1 percent or less. See Table 6.1. All of these poor growth rates were closely related to recessions.

During the fourth quarter of 2000 U.S. industrial production experienced the sharpest decline since the recessionary trough year of 1991 and more than wiped out a modest increase during the third quarter.

The oil price shock of 1999-2000, in any event, was amplified by an over emphasis on U.S. built trucks and SUVs that don't get very good gas milage and by a shortage of natural gas that sent home heating costs and some electrical bills soaring to the highest levels in history.

Another variable that may have helped to set the stage for a recession is a very jittery stock market. From 1928 to 1999 years with 43 or more one day declines of one percent or more for the S&P composite were all closely linked to economic recessions. See Table 3.6. Since this index was extended back to 1928, all of the cumulative new high declines lasting more than 170 trading days have been linked to recessions. See Table 3.7.

It should be appreciated that the dividends and earnings associated with the S&P composite are still over valued from an historical point of view. Sagging earnings may keep this recession indicator from recovering to another new high as quickly as was the case after the last six crashes amounting to 13.9 percent or more. To the extent that investment and consumption are linked to stock holder wealth, that could have a depressing effect on economic activity.

A soaring stock market, in any event, may have helped to raise single family housing starts in 1999 to the highest levels since the baby boomer expansion of 1977-78. If the stock market doesn't recover fairly quickly there aren't going to be as many people who can afford the pricey homes that are now being constructed.

A booming stock market may have also helped to bloat consumption and lower personal saving rates to levels not experienced since the great depression of the 1930s. This, in turn, has led to a condition where many communities are now over malled and over stored.

The consumption explosion has been facilitated in part by a rapid expansion of consumer debt. The interest payments needed to finance expanding debt could depress both investment and consumption, even if there is no precipitous decline in personal income.

An almost tripling of business investment in software and equipment, since the recessionary trough of 1990, might turnout to be another distortion that will keep economic activity in the high tech industry depressed for a long time. Over estimates of consumer willingness to buy goods on line has ruined some software companies and helped to create a surplus of office structures in some cities.

In the post 1947 period economic recessions have tended to be longer the higher the percentage of real GDP devoted to gross private domestic investment during the quarter containing an NBER peak in economic activity. See Table 4.2. If improved technology still has the potential to boost productivity at a rapid rate it could mean the loss of a lot of white collar jobs in a world where there is little or no growth in other types of employment.

The accelerator relationship which is examined in Essay 7 implies that real GDP must increase by about two percent per year just to keep private investment from falling and having a deleterious feedback effect on the growth of economic activity. The instability of the GDP growth rate in the zero to two percent range has sometimes made it very difficult for the Federal Reserve to effectively fight inflation without pushing the U.S. economy into a recession.

The employment variable with one of the best forecasting records prior to the Gulf War recession of 1990-91 is the number of employees on nonagricultural payrolls in goods-producing industries. A cumulative drop in goods-producing employment amounting to 89 thousand workers is not a healthy development. This variable was signaling a near term recession in August 2000. See Table 6.2

For some other indications of a recession click on to: Table 2.1, Table 2.3, Table 3.5, and Table 5.4.

Since the Korean War build-up year of 1952 one has only needed at least three lower totals for non agricultural payroll employment spread over a three to five month period to be confident that the U.S. economy was involved in the eight recessions dated by the National Bureau of Economic research. See Table 5.5.

When the U.S. economy is viewed from this perspective it seems clear that the next few months might turnout to be a very testy time for both leading indicators and monetary and fiscal efforts to achieve a soft landing that will not be marred by another dated recession.

One reason most forecasters are not inclined to predict a recession until after it has occurred is that it might become a self fulfilling prophesy and hurt their employers. There are enough clouds on the horizon, however, so that students, consumers, government employees, business enterprises and ordinary investors might find it advantageous to keep a close eye on policy developments taking place in our nation's capital and some of the economic and financial indicators examined in this collection of essays and tables on the business cycle.

What should students who are worried about another recession do? My advice would be to consider going on to college or graduate school. Unemployment can reduce the "opportunity cost" of a more advanced education and will sometimes result in a big payoff there after.



The Rocky Road Ahead

About 80 percent of the world's oil production comes from large fields which were discovered before the Arab oil embargo of 1973 and are rapidly being depleted. There are a number of models and forecasting approaches which suggest that the world production of conventional oil will peak out in this decade and gradually decline. Further support for this hypothesis can be found in a host of production peaks which occurred before OPEC agreed to reduce its output on March 23, 1999.

U.S. oil production, for example, peaked out way back in 1970 and at the end of 1999 had declined by about 39 percent in spite of major improvements in exploration and recovery technology and expanded oil production in Alaska and the Gulf of Mexico. During the 12 year period after this peak the U.S. economy experienced four recessions. If oil production peaks of a world wide nature are not to result in more recessions, automobile producers and consumers in this country will have to appreciate that the era of very cheap gasoline may be nearing an end and do their part to improve the efficiency of our motor vehicle population.

Big increases in oil production in other countries outside of OPEC and a 24 percent increase in the fuel efficiency of U.S. passenger cars forced OPEC to reduce its oil production by about 48 percent from 1977 to 1985.

From 1988 to 1999, however, increases in oil production in other non OPEC countries was not sufficient, on the average, to offset declining production in the United States. This, in a world of modest improvements in the fuel efficiency of passenger vehicles in the U.S. and rapid increases in motor vehicles in developing countries, enabled OPEC to increase its oil production by 78 percent from 1977 to 1998, before the Asian flu lowered the demand for gasoline in some developing countries.

More than half of the world's proven oil reserves are believed to be located in the Persian Gulf region. Oil production outside this region peaked out in December 1997 and as of June 2000 was still down almost 1.4 percent in spite of higher oil prices.

Giant oil discoveries are now rare and are usually located either in deep water or hostile regions in terms of climate and/or politics. The slow recovery in exploratory drilling from the Asian flu and the steep downward trend in reserves per exploratory well since the 1960s makes it rather questionable whether conventional oil production outside the Persian Gulf will ever recover to a very dramatic new high.

The most talked about frontier outside of the Gulf is in largely unexplored portions of the Caspian Sea. Once a pipeline is constructed to that area it might help to offset declining production in say the North Sea, but not for long, judging by the short lived increase in U.S. oil production after the construction of the Alaska pipe line.

The unanswered question is whether oil producers in the Persian Gulf area will be able and willing to increase their output enough after the current recession is over to more than offset modest increases in the rest of the world and prevent another world wide recession similar to that which occurred after the first big upward spike in crude oil prices in 1973-74.

Iran has little excess capacity at the present time and is unlikely to ever restore production to the peak level that was achieved under the Shaw in 1974. The two countries in the Persian Gulf with the most unused capacity are Iraq and Saudi Arabia. It is by no means clear that the rest of the world will be better off if Iraq, with its war like history and stored away weapons, is allowed to increase its production dramatically.

During the world wide business expansion which followed the Gulf War recession of 1990-91, and was terminated with the Asian flu in 1998, world demand for oil increased at an average annual rate of more than one million barrels per day. Saudi Arabia, which is believed to possess about a quarter of the world's proven oil reserves, might be able to increase its oil output at that rate for two or three years.

Crop production in Saudi Arabia, however, is already in a state of decline as a result of an over emphasis on ground water mining. If that desert kingdom is to feed a rapidly growing population and enjoy prosperity for the rest of this millennium it will need to be rather conservative in its exploitation of remaining oil and gas reserves. Most of the world's other major oil exporting countries are also plagued with natural resource constraints which will make it unwise for them to restore oil production to new historic highs.

There is not much doubt, in any event, that the era of very cheap gasoline is nearing an end. If another recession is to be avoided in this decade, history would suggest that the United States, which now imports almost 60 percent of its oil using money borrowed from the rest of the world, will have to do its part by taking vigorous steps to improve the fuel efficiency of its motor vehicles.

Essay 1:

The Duration of Business Expansions and Depressions

Edward Renshaw
Professor of Economics
State University of New York at Albany

Economic recessions have been studied and analyzed for more than a century. Economists, however, have not had much success at identifying recessionary peaks in business activity in close proximity to their occurrence. This in turn has made it difficult for decision makers to implement measures that might prevent or at least moderate some of the damage that is caused by recessions.

While much labor has gone into the development of models and indicators that might warn decision makers of an impending recession, the success of these efforts has been disappointing. McNees (in Lahiri and Moore 1991, chapter 9) has noted that before the two most prolonged recessions in the Post World War II period (the recessions of 1973-75 and 1981-82) forecasters were not able to identify the peak until "about the time that it was occurring".

Robert Eggert has been summarizing the economic forecasts of a noted panel of economists on a monthly basis since 1976. In July 1979 his Blue Chip Economic Indicators newsletter proclaimed, "PANEL SAYS RECESSION IS HERE--WITH NO UPTURN UNTIL SECOND-HALF 1980". While the panel was right about an upturn occurring in the second half of 1980, the business cycle dating committee of the National Bureau of Economic Research (NBER) finally decided that the actual peak in business activity didn't occur until January 1980 or six months after a majority of the economists surveyed by Eggert thought it was upon us.

The headline for the July 1981 issue of the Blue Chip news letter was "ECONOMIC EXUBERANCE ENVISIONED FOR 1982". It wasn't until November--four months after the NBER peak in business activity, that a majority of the Blue Chip panelists believed that another recession had started. "However, a big majority of the 39 who said we were in a recession agreed with President Reagan that it would be mild. Most thought it would be over by the end of the first quarter of 1982." The actual trough did not occur until November 1982.

In April 1985 forecasters were queried as to how long the on going recovery in business activity would last. The average response was that the business expansion would last 49 months, which put the next peak in business activity at December 1986, more than 3 and one-half years before the actual occurrence of another recession.

In the July 10, 1990 issue of Blue Chip Economic Indicators it was reported that, "The year-ago CONSENSUS forecast for 1990 of a "soft landing" (no recession, but sluggish growth) remains intact. The economy in 1991 is projected to be a shade better, but the prognosis is still largely 'more of the same'." This was the month that ended the longest peace time expansion of the U.S. economy up to that point in time.

During the year 2001 it wasn't until the September 11 terrorist attack on New York City that a majority of economic forecasters got around to admitting that the U.S. economy was probably in the midst of another recession.

Some Reasons for the Poor Forecasting Record

One reason economists have not been very successful at identifying economic recessions before they occurred is the wide variation in the duration of business expansions. In the post World War II period business expansions have ranged in duration from only 12 months for the abortive recovery from the short lived recession of 1980 to more than 90 months for three expansions.

If economic recessions are expected to occur less frequently than was formerly the case, that should discourage some economists from "sticking their neck out" and trying to predict or identify them before they occur.

Another reason for not getting involved in predicting the end of business expansions is the turbulent behavior of real GDP growth rates in the vicinity of business peaks. Six of the ten recessions experienced in the post 1947 period were preceded by at least one quarter with a negative growth rate for chain weighted real GDP. During six of the last eight quarters followed by dated recessionary declines in real GDP, however, the growth rates for real GDP were higher than in the preceding quarter. Business expansions, it would seem, often sputter to an unpredictable end.

It should be appreciated that much of the information pertaining to real GDP is not available at the time preliminary estimates are made. The tendency for BEA to project past growth rates into the future creates an upward bias in the vicinity of recessionary peaks that makes it very difficult to know whether the economy is sinking or not. It was not until July 2002 that BEA finally decided that the U.S. economy declined in both the first and second quarter of 2001 as well as in the third quarter.

The problem of turbulence is not limited to real GDP and its components. All of the indicators that have been developed to identify cyclical turning points and provide advanced warning of a possible recession also suffer from chaotic behavior that can inspire false predictions about what will happen to the economy in the future.

One of the first lessons to be learned from the coincident indicators that are examined in the process of dating business peaks is that they seldom agree on the exact timing of the peak. None of the Conference Board's coincident indicators have consistently declined in the first month after an NBER peak. The usefulness of leading and coincident indicators in helping to identify business peaks before or shortly after their occurrence has also been tarnished in recent years by numerous dips in the indicators that did not trigger a near term recession.

The longest consistent set of forecast data available for the U.S. economy has been compiled by the Research Seminar in Quantitative Economics at the University of Michigan. Over the 35 year period from 1953-87 there were seven years when real GNP declined on a year-to-year basis. In every one of these years the forecasts which were made in the preceding November either missed the recession altogether or underestimated its severity by amounts ranging from only .1 percentage points for the short lived 1980 recession to 4.5 percentage points for the recessionary year of 1974 (McNees 1988, Table 1, p. 17). For the seven recessionary years as a whole the average bias was 1.9 percentage points.

Webb (in Lahiri and Moore 1991, pp. 113-115) has concluded that for horizons of several quarters ahead it is hard to beat a simple forecasting rule of never predicting a recession. At the four-quarter horizon he found that several forecasting services, which use econometric models, did no better than a simple no-recession forecast. "The very low number of long range forecasts of recession may be an implicit acknowledgement of the forecasting services of the low accuracy of such forecasts." The many break-downs and poor success of models that endeavor to explain the duration of business expansions is another reason for not sticking one's neck out and predicting a recession before it occurs.

Any method of trying to identify recessionary peaks before their occurrence, in any event, is highly suspect since inflationary shocks are hard to predict in advance of their occurrence and because changes in the character of the U.S. economy may have made it more recession proof. The data in Table 1.1 on the business cycle expansions and contractions dated by the National Bureau of Economic Research would strongly suggest, however, that business expansions are lasting longer than they used too and that economic contractions are shorter than they once were.

The five longest expansions in business cycle history have all occurred since the horrendous 43 month contraction in economic activity which lasted from August 1929 to March 1933.

While economists have not had much luck at developing models that do a good job of explaining the duration of business expansions there is one interesting relationship that can be used as a standard to help measure the success of monetary and government policies in preventing recessions.

In 1993 the Center for International Business Cycle Research found that the duration of post World War II expansions in economic activity have been about equal to 15 months plus 1.57 times the monthly lag for a secular rise in the prime rate after months containing an NBER trough in business activity. See Table 1.2. If this relationship had continued to hold the most recent business expansion would have ended in January 1997.

In The Longevity of Expansions Haimowitz (1998) suggests that some of the reasons for longer expansions might include: a larger government sector, the use of automatic stabilizers, financial reforms, economic policy activism on the part of both government and the Federal Reserve, structural changes in the economy (such as a reduction in the number of persons employed in the volatile goods producing sector of our economy), incomes that are more recession proof, changes in inventory behavior, increased globalization and improvements in producer and consumer confidence. To this list should be added a more forward looking monetary policy on the part of the Federal Reserve (McNees 1986) and changes in the way inflation is measured.

Rising imports have shifted some of the cost of a slump in goods demand to other countries and bolstered the demand for U.S. exports. If the economies of other countries are not fully synchronized with what is happening in the U.S. the rising share of imports and exports should have a stabilizing effect on both the U.S. economy and world wide output.

Residential construction is one of the most volatile sectors of the U.S. economy but may not destabilize economic activity as much in the future, as it has in the past. In 1950, 6.96 percent of nominal GDP was allocated to this sector. During the housing boom of 1972, when almost a quarter of a million more housing units were started than in 1950, residential construction had slipped to only 5.62 percent of GDP and during the more recent slow growth year of 2001 residential construction was only equal to 3.95 percent of GDP.

Changes in business inventories are another volatile component of GDP that have been closely linked to recessions. Just-in-time delivery of parts and computerized management of business inventories, however, have reduced some of the instability caused by wide fluctuations in inventory investment. During the economic recession of 1982 the inventory/sales ratio for manufacturing and trade peaked out at 1.67. By the end of 1999 it had sunk to 1.32.

The consumption of services is the most stable component of GDP. Since 1950 its share of nominal GDP has increased from 20.6 percent to 40.8 percent in 2001.

Gross private domestic investment (GPDI) has the distinction of being the most recession prone component of real GDP. Since the food and oil price shock recession of 1973-75, the decline in real chain weighted GPDI has drifted downward from 29.4 percent to only 17.1 percent for the more recent recession.

Some Developments that May Have Helped to Stabilize the Inflation Rate

The good news with regard to the wage-price spiral is that cost of living escalator clauses have gone out of fashion making wage inflation less responsive to food and energy price shocks. In 1990, for the first time since 1960, the growth of average hourly earnings slowed during a year containing a peak in business activity. High unemployment in Europe and many developing countries and the globalization of manufacturing have made many industrial unions more concerned about job security than a possible resumption of wage-price inflation.

While consumers are now spending a higher fraction of their income on housing some much needed stability has been imparted to the home ownership component of the CPI cost of shelter index as a result of a 1983 shift to a rental equivalent measure of home owner costs. The old procedure was to measure the price of new houses, mortgage interest rates and other cost elements that are borne by those relatively few families that are fortunate enough to be able to afford a new home during the month in question. This made the cost of home ownership extremely sensitive to changes in monetary policy. In 1970, 1975, 1978, 1979, 1980, 1981 and 1982 the official CPI increased from one to 2.5 percentage points more rapidly than an experimental rental equivalent CPI. Tight money, instead of being a cure for inflation, probably made the official inflation rate worse in some of these years.

The bad news with regard to the inflation rate for shelter is that it is more sticky and not likely to come down as much in response to short lived recessions as other prices. This component of the consumer price index, moreover, is likely to continue to increase more rapidly, on the average, than the CPI as a whole.

Medical care services are another inflation prone component of the CPI that ought, in principle, to be more controllable in the next few years with many older persons quitting smoking, drinking more moderately, getting more exercise and about to shrink in numbers as a result of the baby bust which occurred during World War II and the great depression of the 1930s. The AIDS epidemic and many new and more costly drugs and medical procedures, however, are likely to make health care cost containment one of our most enduring and frustrating problems.

Since food and energy price shocks are less likely to be amplified now by cost of living adjustment clauses in labor contracts and by changes in mortgage interest rates it is reasonable to hope that future changes in the CPI inflation rate will be smaller, on the average, than was the case from 1945-91 and that the Fed won't have to resort to draconian increases in short term interest rates to keep inflation in check.

It should be appreciated, however, that economic recessions are the only sure cure for inflation. In the post 1947 period employment recessions have reduced the CPI inflation rate from a little less than one percentage point, for the mild recession of 1960-61, to almost seven percentage points for the prolonged recession of 1981-82. The reduction in the inflation rate one year before the employment peak to one year after the employment trough has been about equal, on the average, to the percentage point increase in the civilian unemployment rate from its cyclical low to its cyclical high. The higher the 12 month inflation rate at the employment peak, the greater the drop in the inflation rate, if the increase in the unemployment rate was about the same. See Table 1.4. Including the duration of the payroll recessions as an independent variable in the regression used to identify the parameters in this table doesn't improve the adjusted R square. This would suggest that prolonged employment recessions are not very helpful at reducing the inflation rate.

Identifying Economic Recessions Before They Are Over

The recessions dated by the National Bureau of Economic Research from 1947- 2001 were all associated with at least two lower quarterly totals for real GDP. See Table 7.2. One of the problems with this "rule of thumb" for describing recessions is that the recessionary troughs have some times occurred before this type of information was available.

It should be appreciated that recessionary peaks are of a judgmental nature and are not easy to identify on the basis of economic and financial indicators. The most widely publicized "coincident" indicators are industrial production and non agricultural payroll employment.

From 1948-2001 the U.S economy was always mired down in or on the verge of an NBER type of recession after industrial production, which has often been a leading indicator, experienced five lower closing values.

Since 1960 non agricultural payroll employment has often been a lagging indicator at the recessionary peaks identified by NBER. The U.S. economy, however, has (so far) always experienced a peak or been mired down in an NBER recession after four lower closing values for this indicator. See Table 1.6.

The labor force variable with the best overall record for enabling one to identify NBER recessions before they were over is the civilian unemployment rate. From 1948-2001 the U.S. economy was always mired down in a recession after a cumulative increase in this variable of .9 percentage points or more. Since the strike prone labor force unrest of the 1950s, however, one has only needed a cumulative increase of .6 percentage points in the unemployment rate to identify NBER recessions. See Table 5.2.

The Duration of Economic Depressions

In December 1992 NBER's dating committee finally decided that the preceding recession ended in March 1991 and had a duration of only eight months. If you believe this assessment there have only been two shorter recessions in the history of business cycle analysis, the seven month recession following World War I and the six month recession from January to July of 1980.

One of the problems with NBER's dating system is that it doesn't provide any insight as to the severity of the recessions or the time that was required for the economy to recover to a new historic high. This problem can be ameliorated, to some extent, by focusing one's attention on the behavior of gross domestic product (GDP) expressed in constant dollars and other statistical series that have acquired the reputation of being "coincident" indicators of economic activity. See Table 1.3.

Strong demand for US exports caused industrial production to peak out in September 1990, two months after the official NBER peak in business activity, and to only decline by a modest 4.2 percent at the NBER Trough in March 1991. The 25 month depression in industrial production from September 1990 to October 1992, however, was of median duration for the post 1948 period.

The 34 month depression for payroll employment, on the other hand, established a new record for depressed activity that was six months longer than the 28 month depression associated with the 1981-82 recession. The prolonged period of no growth in payroll employment probably cost President Bush his job and gave rise to the immortal political slogan, "it's the economy, stupid."

The prolonged depression in employment may have been related in part to a 26.2 percent reduction in federal investment in structures during the recessionary trough year of 1991. See Table 1.5.

The more important point to note in connection with Table 1.3 is that post World War II recessions in economic activity are not getting shorter, if we view them from the perspective of how long it takes a coincident indicator, such as payroll employment or industrial production, to recover to a new historic high. The average duration of the last four completed recessions is higher for each of these indicators than the average duration for the first five recessions since 1947.

Prolonged depressions in economic activity and our inability to develop policy oriented models that do a good job of explaining the duration of business expansions would suggest, in any event, that some attention should be paid to leading economic indicators and other forecasting approaches that might be of value in identifying recessionary peaks in business activity in close proximity to their occurrence.

References

Haimowitz, Joseph (1998). "The Longevity of Expansions," Economic Review, Federal Reserve Bank of Kansas city, Fourth Quarter, pp. 13-34. This article can be accessed on the bank's Website at www.kc.frb.org.

Lahiri, Kajal and Geoffrey Moore (1991). Leading Economic Indicators: New Approaches and Forecasting Records(New York: Cambridge University Press).

McNees, Stephen (November/December 1986). "Modeling the Fed: A Forward-Looking Monetary Policy Reaction Function," New England Economic Review, pp. 3-8.

------(July/August 1988). "How Accurate Are Macroeconomic Forecasts?" New England Economic Review, pp. 15-36.

Zarnowitz, Victor (1992). Business Cycles: Theory, History, Indicators and Forecasting(University of Chicago Press).


Table 1.1

Business Cycle Expansions and Contractions Dated by NBER.

                                                                     

Business Cycle Reference Dates    ----Duration in Months----
  Trough           Peak           Contraction      Expansion
                                  Previous Peak    Trough to
                                  to Trough        Peak
Dec. 1854        June 1857           ......           30
Dec. 1858        Oct. 1860             18             22
June 1861        Apr. 1865              8             46*
Dec. 1867        June 1869             32             18
Dec. 1870        Oct. 1873             18             34

Mar. 1879        Mar. 1882             65             36
May  1885        Mar. 1887             38             22
Apr. 1888        July 1890             13             27
May  1891        Jan. 1893             10             20
June 1894        Dec. 1895             17             18

June 1897        June 1899             18             24
Dec. 1900        Sep. 1902             18             21
Aug. 1904        May  1907             23             33
June 1908        Jan. 1910             13             19
Jan. 1912        Jan. 1913             24             12

Dec. 1914        Aug. 1918             23             44*
Mar. 1919        Jan. 1920              7             10
July 1921        May  1923             18             22
July 1924        Oct. 1926             14             27
Nov. 1927        Aug. 1929             13             21

Mar. 1933        May  1937             43             50
June 1938        Feb. 1945             13             80*
Oct. 1945        Nov. 1948              8             37
Oct. 1949        July 1953             11             45*
May  1954        Aug. 1957             10             39

Apr. 1958        Apr. 1960              8             24
Feb. 1961        Dec. 1969             10            106*
Nov. 1970        Nov. 1973             11             36
Mar. 1975        Jan. 1980             16             58
July 1980        July 1981              6             12
Nov. 1982        July 1990             16             92*
Mar. 1991        Mar. 2000              8            120

Average, peacetime cycles

  1854-1982 (25 cycles)                19             27
  1854-1919 (14 cycles)                22             24
  1919-1945  (5 cycles)                20             26
  1945-1982  (6 cycles)                11             34

*Business expansions which may have been prolonged or shortened by wars (Civil War, World Wars I and II, Korean War, Vietnam War and Operation Desert Storm).

Source: National Bureau of Economic Research, Inc.


Table 1.2

Using the CIBCR Prime Rate Formula to Predict the Duration of Business Expansions.

                                                                     

       Date of                    Duration of Business Expansion   Actual
---------------------    Lag in   ------------------------------   Minus
  NBER     Prime Rate  Months for     Predicted      Actual      Predicted
 Trough     Trough     Prime Rate     --------in Months------    Duration
   (1)        (2)        (3)             (4)n          (5)         (6)n

Oct. 1945  Nov. 1947      25              54            37         -17

Oct. 1949  Aug. 1950      10              31            45          14*

May  1954  July 1955      14              37            39           2

Apr. 1958  Aug. 1958       4              21            24           3

Feb. 1961  Nov. 1965      57             104           106           2*

Nov. 1970  Mar. 1972      16              40            36         - 4

Mar. 1975  Apr. 1977      25              54            58           4

July 1980  Aug 1980        1              17            12         - 5

Nov. 1982  Mar. 1987      52              97            92         - 5

Mar. 1991  Feb. 1994      35              70           120          50

(4)n. The predicted duration of the business expansion is equal to 15 months plus 1.57 times the monthly lag for the prime rate in column (1). This formula was developed at the Columbia University's Center for International Business Cycle Research under the direction of its Director, Geoffrey Moore, and widely publicized by Lindley Clark in "A Slump Predictor Clinton Should Love," The Wall Street Journal, December 28, 1993, p. A10.

(6)n. Column (5) minus column (4).

Source of basic data: The Federal Reserve Bulletin.

*Business expansions which may have been prolonged by wars.


Table 1.3

The Duration of Recessionary Depressions in Months (or Quarters) for the Federal Reserve's Composite Index of Industrial Production (IP), the Bureau of Labor Statistics' Employees on Nonagricultural Payrolls (EMP) and BEA's Estimates of Real Gross Domestic Product in 1987 Dollars (RGDP) from Own Peaks to a Sustained New High Recovery

                                                                      

 Dates of Recessionary Peaks    Duration to a New High Recovery
----------------------------     -------------------------------
   IP       EMP      RGDP         IP       EMP       RGDP 

July 48   Sep. 48    48-4         21        22         5

July 53   June 53    53-2         20        23         6

Mar. 57   Mar. 57    57-3         23        25         5

Jan. 60   Apr. 60    60-1         21        20         5

Oct. 69   Mar. 70    69-3         26        18         6


Nov. 73   Oct. 74    73-4         37        16         8

May  79   Mar. 80    80-1         26        10         4

July 81   July 81    81-3         26        28         7

Sep. 90   June 90    90-2         25        34         7

Sep. 00   Mar. 01    00-4          ?         ?         4

Average Duration to a New High Recovery

 First Five Recessions           22.2      21.6      5.4

 Last Four Completed Recessions  28.5      22.0      6.5

Source of basic data: The Survey of Current Business, October 1994 and January/February 1996.


Table 1.4

Recessionary Declines in the CPI Inflation Rate

                                         
                                         Decline in Inflation Rate
Economic      % Point           CPI      -------------------------
Recession     Increase       Inflation      Actual     Predicted
              Civilian        Rate at
            Unemployment   Recessionary
                Rate           Peak

                (1)             (2)            (3)n        (4)n     

 1948-49        4.5             6.5            3.0         4.4

 1953-54        3.6             1.0            1.5         1.6

 1957-58        3.8             3.6            3.3         2.7

 1960-61        2.3             2.0            1.1          .7

 1969-70        2.7             6.0            2.5         2.4

 1973-75        4.4            12.0            5.9         6.1

   1980         2.2            14.7            4.0         4.8

 1981-82        3.6            10.7            7.5         4.9

 1990-91        2.7             4.7            1.5         2.0

 2001-?         1.9             3.0

Average Values  3.3             6.4            3.4         3.4

(3)n. The actual 12 month CPI inflation rate at the employment peak in column (2) minus the 12 month CPI inflation rate one year after the employment trough.

(4)n. The predicted decline in the CPI inflation rate is equal to the increase in the civilian unemployment rate from its cyclical low to its recessionary peak in column (1) plus one-third of the peak CPI inflation rate in column (2) minus 2.3 percentage points.


Table 1.5

Percentage Changes in Chain Type Quantity Indexes for Government Investment in Structures During Years Containing a Recessionary Trough in Business Activity.


Trough    Federal   State and
 Year                 Local

1949      - 2.0       38.6

1954      -15.6       20.2

1958       11.4       10.8

1961        6.5        8.3

1970      -18.0      - 7.5

1975      - 3.2      -  .8

1980       15.1      -  .1

1982       17.7      -  .9

1991      -26.2         .6

2002?

Source of basic data: National Income and Product Accounts, 1929-94, Volume 2, Table 7.11.


Table 1.6

Using Industrial Production (IP), Non Agricultural Payroll Employment (EMP) and the Civilain Unemployment Rate (U) to Identify NBER Recessions Before They Are Over.


------------------Dates----------------------   --Duration in Months--
   NBER     5th Lower   4th Lower   U up .9 %   Shortest     NBER 
   Peak     Close IP    Close EMP   Points      Lag          Recession

Nov. 1948   Jan. 1949   Feb. 1949   Jan. 1949       2           11

July 1953   Dec. 1953   Oct. 1953   Nov. 1953       3           10*

Aug. 1957   Dec. 1957   Aug. 1957   Nov. 1957       0            8*

Apr. 1960   June 1960   Aug. 1960   Oct. 1960       2           10

Dec. 1969   Apr. 1970   Aug. 1970   Mar. 1970       3           11

Nov. 1973   Dec. 1974   Feb. 1975   July 1974       8           16#

Jan. 1980   May  1980   July 1980   Apr. 1980       3            6

July 1981   Dec. 1981   Dec. 1981   Nov. 1981       4           16#

July 1990   Dec. 1990   Oct. 1990   Nov. 1990       3            8*

Mar. 2001   Feb. 2001   Aug. 2001   Aug. 2001      -1            ?

*The asterisks identify recessionary declines associated with cases where the fourth lower closing value for payroll employment occurred before the fifth lower closing value for industrial production. These recessions have been shorter on the average than the other recessions dated by NBER.

#The hatch marks identify recessionary declines when the lower closing value dates for both industrial production and payroll employment lagged NBER peak dates by five or more months.

Source of basic data: The Survey of Current Business, October 1994.



Essay 2:

Economic Recessions and Those Sometimes Misleading Indicators

Edward Renshaw
Professor of Economics
State University of New York at Albany

In the last decade there was a revival of interest in leading economic indicators on the part of academics (Lahiri and Moore 1991) and some improvements in indicator construction and interpretation that might aid forecasters and policy analysts to make better use of the cyclical indicators that were pioneered by the National Bureau of Economic Research (NBER) in the 1920s and 30s.

In 1961 the Commerce Department's Bureau of the Census began to publish Business Cycle Developments which featured some of these indicators. In 1968 this publication was expanded to include previously unpublished composite indexes and its title was changed to Business Conditions Digest. In 1972 responsibility for the cyclical indicators was turned over to the Bureau of Economic Analysis. In April 1990 the indicators were integrated into the Survey of Current Business. In December 1995 the responsibility for maintaining and publicizing the composite indexes of leading, coincident and lagging indicators was turned over to the Conference Board.

In this essay we will provide a brief history of important changes in the construction of the index of leading economic indicators and then address such issues as how to reduce the number of false or highly premature indications of an impending recession. Persons who are worried about a possible recession should keep a close eye on the prime rate charged by banks and one of the index of leading indicators' most interesting components, new private housing units authorized by local building permits.

An Historical Perspective

The Commerce Department's first index of leading economic indicators was seriously distorted by a failure to deflate some of its components which were measured in current dollars. At the December 1969 peak in business activity the preliminary lead time for the index was only three months and at the November 1973 peak in economic activity the index registered a new all time high. The composite index was substantially overhauled in 1975 to correct this problem and at the January 1980 peak in business activity the new index had been depressed for a total of 15 months.

The success of the revised index after 1980, however, was impaired by other objectives. Zarnowitz and Boschan noted in (1975) that there are six criteria which are often used in selecting and assessing cyclical indicators: (1) Economic significance--how well understood and how important is the role in business cycles of the variable represented by the data? (2) statistical adequacy--how well does the given series measure the economic variable or process in question? (3) timing at revivals and recessions--how consistently has the series led (or coincided, or lagged) at the successive business cycle turns? (4) conformity to historical business cycles--how regularly have the movements in the specific indicator reflected the expansions and contractions in the economy at large? (5) smoothness--how promptly can a cyclical turn in the series be distinguished from directional change associated with shorter (mainly irregular) movements? and (6) currency or timeliness--how promptly available are the statistics and how frequently are they reported?

Many of the components in the composite index of leading indicators are not as trendy as real GNP or GDP. In order to satisfy objective (4) and make the index of leading economic indicators do a better job of conforming to the upward trend in what many economists would consider our best overall measure of economic activity, the Commerce Department added .142 percent to the index of leading economic indicators each month to achieve an overall targeted trend of .261 percent per month. This trend adjustment sometimes delayed the down turn in the index of leading indicators enough to obscure the risk of a recession. At the July 1981 business peak the new index of leading economic indicators had been depressed for only three months and at the July 1990 peak in economic activity it registered another historic high at a time when six of its eleven components had been depressed for 25 months or more.

In November 1993 the trend adjustment was eliminated. See the discussion by Green and Beckman in the October (1993) Survey of Current Business. This adjustment and a reweighting of the components to make the unadjusted lead times for the composite index conform more closely with the lead times for the median component of the revised index can be considered a major improvement. The index of leading economic indicators which was taken over by the Conference Board, however, was still plagued by a problem of false signals--which led Nobel Laureate Paul Samuelson to once quip that the index of leading economic indicators has correctly predicted 9 of the last five recessions. A further revision of the composite index by the Conference Board in December 1996 eliminated some of the misleading signals.

In Table 2.1 we show some adjusted lead times at business peaks for the Conference Board's first revised index of eleven leading economic indicators. The adjusted lead times are associated with cumulative declines amounting to one percent or more from the index's own peak value to those peaks identified by the National Bureau of Economic Research. A cumulative decline of this magnitude does help to reduce the number of false signals. Two false or rather premature signals, that are associated with tight monetary policy and stock market "corrections" in the midst of prosperity, can still be identified in the 1959-96 period, however.

Coping with the Problem of False Signals

One way to cope with the possibility of false signals is to ignore declines in the index of leading economic indicators until after the Fed has ceased to worry about a recession and has begun to raise short term interest rates aggressively in an effort to prevent an acceleration in the inflation rate. In the post 1947 period the U.S. has (so far) never experienced an NBER peak in business activity until three months after the average prime rate charged by banks has increased by one third or more from its cyclical low.

In the post 1961 period there has (so far) never been a recession until at least five months after the prime rate was up more than 50 percent from its cyclical low. See Table 5.4. This threshold would have allowed one to have eliminated both of the false or very early warning signals for the Conference Board's first revised index of leading indicators. It was not until May 2000 that this indicator began to signal the possiblity of another recession.

Once this amount of financial stringency has been imposed on the economy one should pay closer attention to one of the leading index's most interest sensitive components: the number of new private housing units authorized by local building permits. The collected data are related to the issuance of permits and not the start of construction, which frequently occurs several months later and sometimes not at all.

In 1989 this analyst noted that residential building permits is the only component of the index of leading economic indicators that has consistently declined by at least twenty percent before each of the economic recessions in the post 1947 period. For the five cases where the number of housing units authorized by local building permits exceeded 1800 thousand units there was at least a 35 percent decline in the number of authorized units before the next recessionary peak identified by NBER. This set of relationships turned out to provide the only model or rule of thumb in Lahiri and Moore's collected compendium on Leading Economic Indicators that did a good job of identifying the July 1990 peak in business activity shortly before it occurred. See the last column of Table 2.2 for an updated version of this rule of thumb.

In May 1995, after the Conference Board's revised index of leading economic indicators declined by a little over one percent, the number of authorized building permits was only down about 17 percent and was heading upward in response to lower mortgage rates.

Authorized housing units peaked out at 1,762 thousand units in January 2000 and slipped about 16 percent to only 1,486 thousand units during August 2000. While comparatively modest investment in this sector and a downward drift in mortgage rates since February 2000 prevented this variable from being a reliable indicator of the more recent recession, economy watchers should probably continue to keep a close eye on this component of the Conference Board's index of leading indicators. The unadjusted lead time for this indicator was 14 months, making it the sixth longest lead time for the last ten recessions.

Inverted Yield Curves

An inverted yield curve for government securities implies a prediction on the part of participants in the fixed income securities market that long term interest rates are likely to decline. This in turn is only likely to happen if the economic growth rate is slowing down and the economy is in danger of slipping into another recession. There is now an extensive literature on the use of interest rate spreads to forecast economic activity and identify business peaks before they occur. See Estrella and Mishkin (1996) for a review of this literature.

Since 1955 the US economy has never experienced a dated recession until at least seven months after the yield on one year Treasury notes first exceeded the monthly yield on ten year Treasury bonds. See Table 2.3. Except for the Vietnam War build up credit crunch inversion of 1965-66 all of the lead times for this recession indicator were in the seven to 17 month range. And except for the two years (1965 and 1967) surrounding the Vietnam War build-up, all December yields for one year Treasury notes in excess of the yield on ten year Treasury bonds have been followed by below average growth rates for real GDP expressed in 1987 dollars. During March 2000 the yield curve for U.S. Treasury securities began to show signs of an inversion and during August it became decidedly inverted.

Modern portfolio theory assumes that risk will be reward in the market place. Since the prices of long term bonds fluctuate more widely than the prices of notes, bills and marketable commercial paper, there is a presumption that the slope of the yield to maturity curve should normally slope upward. If risk is not being reward in the bond market one should probably be cautious about investing in the stock market.

The data in Table 2.3, in any event, support the hypothesis that it has been rather risky to have owned stock after a yield curve inversion. The February 1989 inversion is the only case where one couldn't have sold an index fund tracking the S&P 500 at the end of the month after the first inversion and then been able to repurchase the index at a lower value during the stock market crashes associated with recessionary peaks in business activity.

It makes sense, in any event, for investors and people who are concerned about the future of economic activity to keep a close eye on the shape of the yield to maturity curve which has now become a component of the Conference Board's revised index of leading economic indicators.

One of the more important lessons to be learned from the current recession, however, is that an aggressive effort on the part of the Federal Reserve to lower its funds rate can sometimes be more successful at eliminating inverted yield curves and bolstering both the housing market and indexes of leading indicators to some extent than in preventing or shortening the life of economic recessions. See Table 9.4.

Persons who are concerned about the possibility of another recession, in any event, should pay close attention to other types of recession indicators. And that is what the remaining chapters in this compendium are all about.

References

Estrella, Arturo and Frederic Mishkin (1996). "The Yield Curve as a Predictor of U.S. Recessions," FRBNY Current Issues, June.

Green, George and Barry Beckman (1993). "Business Cycle Indicators: Upcoming Revision of the Composite Indexes," Survey of Current Business, October, 44-51.

Lahiri, Kajal and Geoffrey Moore (1991). Leading Economic Indicators: New Approaches and Forecasting Records(New York: Cambridge University Press).

Moore, Geoffrey (1961). Business Cycle Indicators(Princeton: Princeton University Press), Vol. 1, Table 3.2, p. 56.

Renshaw, Edward (1992). The Practical Forecasters' Almanac(Burr Ridge, Illinois: Irwin Professional Publishing).

Zarnowitz, Victor and Charlotte Boschan (1975). "Cyclical Indicators: An Evaluation and New Leading Indexes," Business Conditions Digest, May, v-xiv.


Table 2.1

Fluctuations in the Conference Board's First Revised Index of Leading Economic Indicators Amounting to 1.0 Percent or More.

                                                                     

              Dates of             Index of Leading Indicators
----------------------------------  -------------------------    Adjusted
 Index  Index Down  NBER  Index      Index    NBER    Index     Lead Time
 Peak      1.0%     Peak  Trough     Peak     Peak    Trough    in Months

  (1)      (2)      (3)     (4)       (5)      (6)     (7)        (8)n

May  59  Oct. 59  Apr. 60  Mar. 60    75.1     73.8    73.6       - 6

Mar. 66  June 66   -----   Feb. 67    85.1     ----    83.2        ---

Apr. 69  July 69  Dec. 69  Apr. 70    87.4     85.7    83.9       - 5

Feb. 73  June 73  Nov. 73  Feb. 75    92.4     90.6    83.9       - 5

Oct. 78  Dec. 78  Jan. 80  Apr. 80    93.0     90.1    87.0       -13

Apr. 81  June 81  July 81  Mar. 82    90.9     89.8    88.7       - 1

June 88  Mar. 89  July 90  Jan. 91   100.8     99.6    97.8       -16

Dec. 94  May  95   -----   May  95   101.6     ----   100.4        ---

(8)n. Months from the NBER business peak in Column (3) to the month when the revised index of leading economic indicators was down one percent in Column (2).


Table 2.2

Cyclical Peaks for New Private Housing Units Authorized by Local Building Permits and Recessionary Peaks in Business Activity.

                                                                     

                   Thousands of Housing
 Peak Months for         Units
-----------------  --------------------           Lead Time in Months
Housing  Economic   Permit    Economic  Percent   -------------------
Permits  Activity   Peak       Peak     Decline   Unadjusted  Adjusted

                      (1)       (2)        (3)       (4)        (5)n
Oct. 47  Nov. 48     1476      1076       27.1       -13        - 1

July 50  July 53     2282      1248       45.3       -36        -28

Feb. 55  Aug. 57     1946      1187       39.0       -30        -11

Nov. 58  Apr. 60     1729      1232       28.7       -17        - 5

Feb. 69  Dec. 69     1689      1306       22.7       -10          0

Dec. 72  Nov. 73     2688      1557       42.1       -11          0

Jun. 78  Jan. 80     2065      1333       35.4       -19        - 1

Sep. 80  July 81     1545       974       37.0       -10        - 1

Feb. 84  July 90     2043      1108       45.8       -77        - 3

Jan. 00  Mar. 01     1762      1627        7.7       -14     No Signal

(5)n. Months from the next economic peak identified by the NBER after the building permit index has been down 20 percent for two months in a row if the permit peak was less than 1800 units and down 35 percent for two months in a row if the permit peak was over 1800 units.

Source of basic data: An index (1987 = 100) in the October 1995 issue of the Survey of Current Business is multiplied by 15.348 to produce unit values for the 1947-53 period. The basic data are collected by the Bureau of the Census, Manufacturing and Construction Division, Building Permits Branch, Washington, DC 20233.


Table 2.3

Using an Inverted Yield Curve for the One Year Treasury Note Rate and the Ten Year Treasury Bond Rate to Identify Business Peaks Before They Occur and Risky Times to Have Owned Common Stock.


--------Date of---------      Lead   -----Values for the S&P 500-----
   Yield          NBER        Time     After      After   Recessionary
   Curve        Business       in      Yield      NBER    Closing Low
Inversion         Peak       Months  Inversion    Peak    S&P 500

   (1)n           (2)          (3)       (4)       (5)       (6)

Dec. 1956      Aug. 1957      - 8      46.67     45.22     38.98

Sep. 1959      Apr. 1960      - 7      56.88     54.37     52.30

Dec. 1965      Dec. 1969      -48      92.43     92.06     69.29

Mar. 1973      Nov. 1973      - 8     111.52     95.96     62.28

Sep. 1978      Jan. 1980      -16     102.54    114.16     98.22

Sep. 1980      July 1981      -10     125.46    130.92    102.42

Feb. 1989      July 1990      -17     288.86    365.15    295.46

Apr. 2000      Mar. 2001      -11    1452.43   1160.13    776.76

(1)n. The first month in a business expansion when the yield on one year Treasury notes rises above the yield on ten year Treasury bonds.

Source of basic Data: The Federal Reserve Bulletin.



Essay 3:

The Stock Market and the Business Cycle

Edward Renshaw
Professor of Economics
State University of New York at Albany

Some of the most persuasive evidence in support of a longer run pattern to stock prices has been provided by the National Bureau of Economic Research. In a massive compendium which was published in 1961, Geoffrey Moore found that common stocks were excelled as a leading indicator of business cycles by only the net change in the number of operating businesses. Stock prices were classified as being a leading indicator 31 times, roughly coincident 14 times and a lagging indicator only five times.

The most publicized weakness of stock prices as an indicator of economic activity is a propensity for them to have predicted more economic recessions than have actually occurred. This point is illustrated in Table 3.1 where cumulative reversals amounting to 5.5 percent or more in the monthly average daily closing prices for the S&P composite 500 stock price index are used to define peaks and troughs in the index.

This reversal percentage was the smallest percentage that one could use and still eliminate all of the abortive stock market rallies which occurred in the midst of economic recessions from 1947-91. Once the economy was in a recession and the stock market has recovered 5.5 percent on a monthly average basis the S&P index continued up for an additional gain of at least 25 percentage points before the index reversed itself on the downside by 5.5 percent. (See the gains in column 5 of Table 3.1 which are marked with a double asterisk.)

Rules of thumb, which would have worked well in the past sometimes break down, however. During the industrial recession which began in October of 2000 the S&P composite experienced a monthly average rally of 7.1 percent from March 2001 to May of that year. This rally was probably related to a rather aggressive effort on the part of the Federal Reserve to lower its funds rate.

After declining another 17.8 percent between May and September the S&P composite then had another false rally amounting to 10.4 percent before declining 25.9 percent and hopefully bottoming out during October 2002. This crash may have been related to the publicity associated with numerous accounting scandals and a better appreciation that the stock market was still overvalued from an historical perspective.

It will be noted that there have been 29 downside reversals of 5.5 percent since May 1946 and only ten recessionary peaks. Seventeen of these downside reversals occurred in the midst of a business expansion and were followed by an upside reversal of at least 5.5 percent before the economy became mired down in a recession. There are eight cases where a downside reversal of 5.5 percent was followed by a recession and three cases where a reversal of this magnitude did not occur until after the business peaks of January 1980, July 1991 and March 2001.

From January to April of 1994 the monthly values for the S&P composite stock price index declined 5.446 percent and came very close to adding another contraction of 5.5 percent to the downside reversals in Table 3.1. The disqualification of this contraction enabled the S&P index to exceed the old record established in January 1966 of 43 months of sustained rise in stock prices (without a dip of 5.5 percent or more) when the monthly values for the S&P index recovered to a new historic high of 481.92 in February 1995.

The staying power of the 1990-2000 bull market can also be illustrated by examining fluctuations of ten percent or more in the monthly values for the S&P composite and the Cowles Commission indexes--which have been extended back to the 1870s. In October 1996 the S&P index exceeded the previous record of 71 months of sustained increase without a correction of ten percent or more which was established from October 1923 to September 1929.

Economy watchers and investors who want to distinguish between those cumulative declines that have turned out to be the prelude to a recession and those crashes which have occurred in the midst of business expansions are probably well advised to ignore monthly average values for the S&P index and focus their attention on cumulative new high declines amounting to 13.9 percent or more.

In the 1958-2001 period cumulative dips of 13.9 percent or more which occurred after a cyclical increase in the discount rate on new issues of 91 day Treasury bills of 3.15 percentage points or more were all associated with economic recessions. See the down dates identified with hatch marks in Table 3.2. This type of tight monetary policy was predicting another recession when the S&P dipped 13.9% on November 30, 2000.

From 1928 to 2001 there were only three recessions that began more than one month after a cumulative dip of 13.9 percent. The first exception is associated with the Vietnam War wind down recession of 1970. Very tight labor markets and a problem of accelerating inflation encouraged the Board of Governors of the Federal Reserve to increase the Federal funds rate by more than three percentage points during the first eight months of 1968. The stock market reacted negatively and was down more than 15 percent from its November 1968 high by the end of July, but the U.S. economy didn't peak out until December 1969 when the Federal Government really began to contract its defense expenditures. Pent up demand for motor vehicles and other consumer goods by returning veterans of an ill fated war made the recession of 1970 one of the mildest on record in terms of the decline in payroll employment.

The second case when the S&P composite experienced a cumulative dip of 13.9 percent several months in advance of a recessionary peak occurred during the first five months of 1973 when the Fed was also raising short term interest rates at a very rapid rate in response to accelerating inflation.

In the Economic Report of the President which was transmitted to Congress in February 1973, the Council of Economic Advisers noted that price and wage controls had worked better than many had expected. The good news with regard to CPI inflation, however, was about to end as a result of uncontrolled prices for basis food stuffs. The world wide food shortage which followed crop failures in the Soviet Union and a number of other countries caused the price of wheat at the farm level to increase 124.4 percent and the price of corn to increase 62.4 percent on a year-year basis during 1973 after increases of 31.3 and 45.4 percent respectively in 1972.

The fourth and biggest Arab-Israeli War in 25 years erupted in the afternoon of Yom Kippur on October 6. On October 19 a total ban on oil exports to the U.S. was imposed by Arab oil producing nations. Food inflation in conjunction with long waits at gasoline pumps and a sizable increase in the price of household energy caused the CPI inflation rate to more than double from 3.4 percent in 1972 to 8.7 percent for 1973 and helped to tip the U.S. economy into another recession by December 1973.

Both of these exceptions, however, can easily be identified as the prelude to a recessionary bear market by examining changes in the Fed funds rate in the 12 months terminating with a cumulative dip of 13.9 percent for the S&P composite. The four cases (7/69, 5/73, 3/80 and 9/82) where the Fed funds rate had increased by 2.00 percentage points or more on a 12 month basis are all associated with recessionary bear markets.

The third case of a recessionary peak occurring several months after a dip of 13.9 percent in the S&P composite is the more recent recession beginning in March 2001. In this case the Fed funds rate was only up 1.09 percentage points on a 12 month basis. Part of the reason for the delay, however, might have been a stock market that was still over valued from an historical perspective.

If the S&P composite has experienced a cumulative decline of 13.9 percent in the midst of a business expansion and the U.S. economy has not yet experienced a recessionary peak or any indication of a near term peak based on the recession indicators in Essays 2, 4 and 6, history would suggest that the stock market may be experiencing a crash in the midst of prosperity.

There is not much doubt that the U.S. economy is more recession proof than it used to be. It should be appreciated, however, that most of the completed stock market crashes in the midst of prolonged business expansions since 1928 have occurred fairly early in the expansion or in the midst of a war time build up like the crash of 1966.

The first exception is the celebrated crash of 1987. A big difference between that crash and the more recent late occurring, 1998 crash in the midst of prosperity, is the state of U.S. exports. Our exports increased 16.1 percent in 1988 in response to a one third reduction in the real multilateral trade- weighted value of the U.S. dollar from 1985-88.

During the third quarter of 1998 U.S. exports in chain weighted dollars were down by about 1.8 percent from the preceding 4th quarter peak in response to financial turmoil and economic recessions in South Asia, Japan, Russia and numerous developing nations. Consumer demand for goods and services in the U.S., on the other hand, was up by about 3.9 percent in response to declining inflation and improved employment.

In the 1929-2001 period all of the false recessionary signals occurring in a business expansion 32 months or more after a recovery from an NBER recessionary trough were eventually followed by a recession associated with the next dip of 13.9 percent or more for the S&P composite.

In the post 1946 period there has usually been a full recovery from a crash in the midst of prosperity in less than two years. The only exception, so far, is the crash which began in September 1976 when the all item CPI was about to zoom upward from only 4.9 percent in 1976 to 13.3 percent in 1979. Respectable earnings growth and increases in the Federal funds rate which were not sufficient to offset increases in the inflation rate, however, did enable the S&P composite to finally recover the ground that was lost from 9/21/76 to 3/6/78 by August 15, 1979.

Did The Stock Market End the Longest Expansion in Business Cycle History?

January and December are the most interesting months to study the stock market. By the mid-1970s, statisticians were beginning to conclude that there might be some seasonality to the monthly rates of return from holding common stock due primarily to large January returns and by the early 1980s it was discovered that most of the exceptional returns in this month were associated with small firms.

While much has been written about the January size effect, it is less well appreciated that this month often sets a tone for the market that will carry forward to the end of the year. From 1947 to 2000 there were 30 years when the S&P composite increased by one percent or more in January and in every one of those years the financial return for the entire year was positive.

Over the somewhat longer period from 1942 to 2000 there were 19 years when the S&P composite declined in January. See Table 3.5. The January 1974 decline occurred in the middle of a prolonged recession and the January 1970 and 1982 declines in years containing a recessionary trough. Five of the other January declines occurred in the first two years of recovery after a year containing a recessionary trough. See those years in this table that are identified with an "R". All of the other January declines were followed by a recessionary peak in from three to 24 months.

Another way to eliminate some of the longer lead times associated with January declines in the S&P composite is to pay close attention to the behavior of short term interest rates. When the prime rate charged by banks increased by .25 percentage points on either a January to January basis or on a preceding year to year average basis in the 1942-2001 period, there were only two January declines in 1956 and 1978 when the U.S. economy didn't experienced a recessionary peak in the remainder of the decline year or early in the following year.

The 19 month lead following the January 1956 decline occurred in the second year after the recessionary trough of 1954. The 24 month lead following the January 1978 decline occurred under very unusual circumstances in the third year after the recessionary trough of 1975.

The 1977-79 period is very unique in that it represented a protracted period when the inflation rate for the all item consumer price index was not only increasing faster than the Fed funds rate but was also in excess of most short term interest rates. The high inflation rates relative to the cost of borrowing provided consumers and business enterprises with a strong incentive to rush out and purchase durable goods on credit until Paul Volcker took over the helm of the Federal Reserve in 1979 and raised short term interest rates fast enough to tip the U.S. economy into two recessions in less than two years.

Under more ordinary inflation fighting conditions in the post World War II period, prolonged business expansions have ended during the year starting out with a January decline in the S&P composite or shortly after as was the case for the January decline in the year 2000.

History would suggest, in any event, that investors should be cautious after a January dip in the S&P composite. Further declines to the lowest following closing value for this index have ranged from a modest 2.6 percent after the January 1978 case to a more monstrous 46.3 percent after the January 1973 decline. The 1.6 percent decline during January of 2002 was immediately followed by five down days in a row. Being able to distinguish between crashes in the midst of prosperity and those associated with recessions, however, can be of some value to investors.

Economic Recessions Associated with a Jittery Stock Market

The parentheses associated with the January declines in Table 3.5 show the number of days that the S&P composite declined by one percent or more during the year in question. All of the years with 25 or more one percent down days from 1947-2000 were closely associated with economic recessions.

Five of these years contained a peak in economic activity. One year (1974) was a year of intervening decline in economic activity. Two of the years contained a recessionary trough and one year (1962) was a first year of recovery from the recession of 1960-62.

The year 2000 now has the distinction of being the second most jittery year associated with a January decline in the S&P composite since the great depression of the 1930s.

Since the S&P composite was extended back to 1928 all of the years with 43 or more one percent down days have been closely associated with economic recessions. See Table 3.6.

Three of these years contained recessionary peaks. One of these years, 2000, was soon followed by a recession. There were four cases of intervening decline in economic activity, one year containing a recessionary trough and three cases of 43 or more one day declines in the S&P composite which occurred during the first two years of recovery from an economic recession.

The year 2000 now has the distinction of being the most jittery year for the NASDAQ composite in its entire thirty year history and the tenth most jittery year in the history of the S&P composite.

Prolonged New High Declines in the S&P Composite

From 1928-2001 prolonged new high declines in the S&P composite lasting more than 200 trading days were all associated with economic recessions. See Table 3.7.

Recessionary crashes lasting more than 400 trading days were all of prolonged duration.

The unanswered question as of November 2002 is whether the very prolonged crash from March 24, 2000 to October 9, 2002 will also be associated with a prolonged recession.

Stock Market Declines After Business Peaks

One of the best times to have sold common stock in the post 1947 period is about eight months before a peak in business activity as defined by the National Bureau of Economic Research. An investor with the clairvoyance to have followed this strategy could have repurchased a portfolio similar to the S&P index two months after the business peak at an average discount of about nine percent.

The problem with this method of refuting the random walk hypothesis is that economists haven't been very successful at identifying business peaks before their occurrence. The many failures of economic models, in this regard, should make one a bit suspicious of market timing news letters. While it is easy to devise switching models that would have generated impressive profits in the past, the out of sample performance of these models has been rather dismal, especially in the 1990s.

Suppose one was able to identify business peaks at the end of the month in which NBER finally decided that they occurred. Would it then be too early to purchase common stock? The data which are presented in Table 3.3 show that the S&P index has then proceeded to lose from 3.5 to 35.1 percent of its value on a daily closing value basis before bottoming out during the ten dated recessions since 1947.

Monthly Lead Times for Stock Prices at Industrial Troughs

While much attention has been paid to the unreliability of stock prices as a predictor of economic recessions it is less well appreciated that this market use to be a very superior predictor of economic recoveries. If we ignore the money supply, M-2 expressed in constant dollars, which is so trendy as to have not exhibited a distinct trough during some recessions, stock prices can claim the distinction of being the only component of the index of leading economic indicators that consistently led economic recoveries from 1947-99.

In Table 3.4 we show the recessionary lead times for the lowest daily recessionary closes for the S&P composite that have been associated with industrial production troughs. Daily lows for the S&P are more interesting from a speculative point of view than the monthly average lows associated with most leading indicator indexes. Industrial troughs have (so far) coincided with most NBER trough dates in the post 1947 period with two misses of only one month each.

All of the S&P lead times to industrial troughs from 1948-91 fell in the comparatively narrow range of from four to seven months. The lead times were longer, on the average, for those recessions associated with U.S. participation in international wars of various kinds, than for recessions which have occurred in more peaceful times.

There is a possibility, however, that corporate accounting scandals may have destroyed the usefulness of stock prices as a leading indicator of the current recovery. One would hope, in any event, that industrial production won't retreat to a new low during 2003.

If the current recession turns out to be an exception, it follows that if investors are to maximize the benefits from stock market crashes associated with economic recessions, they will usually have to possess the courage to buy stocks before the economy and most statistics with leading indicator properties have turned up.

One statistic that ought to be kept in mind by investors who want to take advantage of economic recessions is the civilian unemployment rate. From 1947- 2001 the U.S. economy was always bogged down in another recession after a cumulative increase in the unemployment rate amounting .9 percentage points.

For those recessions from 1947-90 the S&P composite always appreciated at least 14.8 percent one year after the end of the month following a cumulative increase in the unemployment rate of this magnitude. See Table 3.8.

During five of the nine completed recessions in this table the S&P composite had experienced a bear market low before the end of the month following a cumulative increase in the unemployment rate of .9 percent.

The behavior of the stock market in 2002 would suggest, however, that economic and financial relationships that would have been useful predictors in the past can break down and not be as reliable in the future.


Table 3.1

Duration and Amplitude of Monthly Average Stock Market Fluctuations Amounting to 5.5 Percent or More.

                                                                     

    Date of Cycle      S&P Stock Index  Months Duration  % Change in Index
--------------------   ---------------  ---------------  -----------------
   Peak     Trough     Peak     Trough   Rise   Decline   Rise    Decline
                        (1)      (2)     (3)      (4)     (5)       (6)

May  1946  Nov. 1946   18.70    14.69     30        6     65.0     21.4
Feb. 1947  May  1947   15.80    14.34      3        3      7.6      9.2
July 1947  Feb. 1948   15.77    14.10      2        7     10.0     10.6
June 1948  June 1949   16.82    13.97      4       12     19.3     16.9*

June 1950  July 1950   18.74    17.38     12        1     34.1**    7.3
Jan. 1953  Sep. 1953   26.18    23.27     30        8     50.6     11.1*
July 1956  Feb. 1957   48.78    43.47     34        7    109.6**   10.9
July 1957  Dec. 1957   48.51    40.33      5        5     11.6     16.9*
July 1959  Oct. 1960   59.74    53.73     19       15     48.1**   10.1*

Dec. 1961  June 1962   71.74    55.63     14        6     33.5**   22.5
Jan. 1966  Oct. 1966   93.32    77.13     43        9     67.8     17.3
Sep. 1967  Mar. 1968   95.81    89.09     11        6     24.2      7.0
Dec. 1968  June 1970  106.48    75.59      9       18     19.5     29.0*

Apr. 1971  Nov. 1971  103.04    92.78     10        7     36.3**   10.0
Jan. 1973  Aug. 1973  118.42   103.80     14        7     27.6     12.3
Oct. 1973  Dec. 1974  109.84    67.07      2       14      5.8     38.9*
July 1975  Sep. 1975   92.49    84.67      7        2     37.9**    8.5
Sep. 1976  Mar. 1978  105.45    88.82     12       18     24.5     15.8
Aug. 1978  Nov. 1978  103.92    94.71      5        3     17.0      8.9

Feb. 1980  Apr. 1980  115.34   102.97     15        2     21.8     10.7*
Nov. 1980  July 1982  135.65   109.38      7       20     31.7**   19.4*
Oct. 1983  July 1984  167.65   151.08     15        9     53.5**    9.9
Aug. 1987  Dec. 1987  329.36   240.96     37        4    118.0     26.8

Jun. 1990  Oct. 1990  360.39   307.12     30        4     49.6     14.8*
July 1998  Sep. 1998 1156.58  1020.64     93        2    276.6**   11.8
July 1999  Oct. 1999 1380.99  1300.01     10        3     35.3      5.9
Aug. 2000  Mar. 2001 1485.46  1185.85      9        7     14.3     20.2*
May. 2001  Sep. 2001 1270.37  1044.64      2        4      7.1     17.8*
Mar. 2002  Oct. 2002 1153.79   854.63      2        7     10.4**?  25.9?

*Stock market declines that were associated with economic recessions.

**Rises following economic recessions.

Source of Basic Data: Economic Report of the President.


Table 3.2

Recessionary Bear Markets and Stock Market Crashes in the Midst of Prosperity, 1928-

                                                                     

                                             % Peak-   Months from Previous 
   S&P      S&P      S&P      NBER    S&P    Trough    Recession Trough to
   Peak    Down     Trough    Peak    Peak   Decline   S&P  S&P Down    NBER
   Date    13.9 %    Date     Date    Value    S&P     Peak  13.9 %     Peak
                              (1)      (2)    (3)     (4)    (5)n      (6)

 9/ 7/29 10/23/29   6/ 1/32   8/29*    31.92  -86.2     22    23         21*

 7/18/33  7/21/33   3/14/35   ----     12.20  -33.9      4     4F        --

 3/ 6/37  4/28/37   3/31/38   5/37*    18.68  -54.5     48    49         50*

 8/ 6/38  9/10/38   9/27/38   ----     12.81  -14.2      2     3F        --

11/ 9/38  1/26/39   4/28/42   ----     13.79  -45.8      5     7F        --

 5/29/46  9/ 3/46   5/17/47   ----     19.25  -28.8      7    11F        --

 6/15/48  2/10/49   6/13/49  11/48*    17.06  -20.6     32    40(-1.45)  37*

 6/12/50  7/17/50   7/17/50   ----     19.40  -14.0      8     9F( 2.49) --

 1/ 5/53  9/14/53   9/14/53   7/53*    26.66  -14.8     39    47(- .32)  45*

 8/ 2/56  2/11/57  10/22/57   8/57*    49.74  -21.6     27    33(  .48)  39*

 8/ 3/59 10/25/60# 10/25/60   4/60*    60.71  -13.9     16    30(- .43)  24*

12/12/61  5/22/62   6/26/62   ----     72.64  -28.0     10    15(  .90)  --

 2/ 9/66  8/18/66  10/ 7/66   ----     94.06  -22.2     60    66( 1.23)  --L

11/29/68  7/23/69#  5/26/70  12/69    108.37  -36.1     93   101( 1.03) 106

 1/11/73  5/21/73# 10/ 3/74  11/73    120.24  -48.2     26    30(  .80)  36

 7/15/75  9/16/75   9/16/75   ----     95.61  -14.1      4     6F( 1.05) --

 9/21/76 12/19/77   3/ 6/78   ----    107.83  -19.4     18    32( 1.01)  --L

 2/13/80  3/24/80#  3/27/80   1/80*   118.44  -17.1     59    60(  .42)  58*

11/28/80  9/ 4/81#  8/12/82   7/81*   140.52  -27.1      4    14(- .10)  12*

10/10/83  7/24/84   7/24/84   ----    172.65  -14.4     11    20( 1.06)  --

 8/25/87 10/16/87  12/04/87   ----    336.77  -33.5     57    59(  .91)  --L

 7/16/90  8/22/90# 10/11/90   7/90*   368.95  -19.9     92    93(- .19)  92*

 7/17/98  8/31/98   8/31/98   ----   1186.75  -19.3     88    89(  .50)  --L

 3/24/00 11/30/00# 10/09/02   3/01   1527.46  -49.1    108   116(  .28) 120

# The hatch marks identify bear market declines of 13.9 Percent or more which occurred after a cyclical increase in the discount rate on new issues of 91 day Treasury bills of 3.15 percentage points or more. From 1958-2001 all of these cases were associated with recessionary bear markets.

* The asterisks identify recessionary bear markets with an NBER peak before or not more than one month after the S&P composite stock price index had decline by 13.9 percent or more.

F identifies false signals that are associated with cumulative declines of 13.9% that occurred less than a year after the previous recessionary trough identified by NBER.

L identifies false signals that are associated with cumulative declines of 13.9% that occurred at least 32 months after the previous recessionary trough identified by NBER. In the 1929-2001 period all of these false signals were eventually followed by dips of 13.9% that are associated with economic recessions.

(5)n. The numbers in parentheses are the percentage changes in non agricultural payroll employment during the three months ending in the month when the S&P composite was down 13.9 percent from the peak dates. Payroll employment was already down at this juncture during five of the last ten recessionary bear markets.


Table 3.3

Stock Market Declines Associated with Business Peaks in the Post World War II Period

                                                                     

           Date of            Closing Values S&P Index     % Decline
----------------------------  ------------------------  -----------------
 Market   Business   Market   Market  Business  Market  M. Peak   B. Peak
  Peak      Peak       Low     Peak     Peak      Low   B. Peak   M. Low

 6/15/48  11/30/48   6/13/49   17.06    14.75    13.55     13.5      8.1

 1/05/53   7/31/53   9/14/53   26.66    24.75    22.71      7.2      8.2

 8/02/56   8/30/57  10/22/57   49.74    45.22    38.98      9.1     13.8

 8/03/59   4/29/60  10/25/60   60.71    54.37    52.30     10.4      3.8

11/29/68  12/31/69   5/26/70  108.37    92.06    69.29     15.1     24.7

 1/11/73  11/30/73  10/03/74  120.24    95.96    62.28     20.2     35.1

 2/13/80*  1/31/80   3/27/80  118.44*  114.16    98.22      3.6     14.0

11/28/80   7/31/81   8/12/82  140.52   130.92   102.42      6.8     21.8

 7/16/90   7/31/90  10/11/90  368.95   356.15   295.46      3.5     17.0

 3/24/00   3/30/01  10/09/02 1527.46  1160.13   776.76     24.0     33.0

*Only case where the stock market peak occurred after the business peak.

Source of Basic Data: Standard and Poor's Security Price Index Record.


Table 3.4

Some Monthly Lead Times for the S&P Composite at Industrial Troughs

                                                                     

Date of        Closing     Industrial   S&P Lead   --Percentage Change S&P--
Recessionary   Value       Low Month    Time In    Preceding   Next    Year
Low Value S&P  S&P                      Months     Peak to     Three   After
Composite      Composite                           Low Value   Months  Low

   (1)            (2)         (3)        (4)           (5)      (6)     (7)

 6/13/49         13.55       10/49#      - 4         -20.6     16.2    42.1

 9/14/53         22.71        4/54       - 7*        -14.8      8.7    37.1 

10/22/57         38.98        4/58       - 6         -21.6      5.7    31.0

10/25/60         52.30        2/61#      - 4         -13.9     15.7    30.7

 5/26/70         69.29       11/70       - 6*        -36.1     17.2    43.7

10/03/74         62.28        3/75#      - 5         -48.2     13.5    38.0

 3/27/80         98.22        7/80       - 4         -17.1     18.1    37.1

 8/12/82        102.42       12/82#      - 4         -27.1     36.2    58.3

10/11/90        295.46        3/91       - 5*        -19.9      6.7    29.1

10/09/02        776.76       12/01#      +10         -49.1        ?      ?

#The hatch marks identify cases where industrial production peaked out at least nine months before the S&P low. The following lead times for these cases to an industrial trough have been shorter, on the average, than for the other cases.

*The asterisks identify lead times that may have been lengthened as a result of U.S. involvement in international wars of one sort or another.

Source of Industrial Trough Data: BCI series 47, the Survey of Current Business, October 1994.


Table 3.5

Recessionary Peaks Following January Declines in the S&P Composite Stock Price Index.


                    Preceding Change Prime
Year    January     Rate Charged by Banks     Lead Time in Months
        % Decline   Jan.-Jan     Yr.-Yr.      from Economic Peak to
        S&P Comp.                             January Dip in S&P  

1948P    - 4.0(25)     .25*       ----             - 10

1953P    -  .7(15)     .00         .44*            -  6

1956R    - 3.6(21)     .50*        .11             - 19
1957P    - 4.2(25)     .50*        .61*            -  7

1960P    - 7.1(17)    1.00*        .65*            -  3

1962R    - 3.8(34)     .00      -  .32

1968     - 4.4( 9)     .04         .00
1969P    -  .8(18)     .95*        .65*            - 11
1970T    - 7.6(33)    1.55*       1.67*

1973P    - 1.7(43)     .82*     -  .45             - 10
1974D    - 1.0(68)    3.73*     - 2.77

1977R    - 5.1(12)  -  .75      - 1.02
1978     - 6.2(24)    1.68*     -  .02             - 24

1981P    - 4.6(30)    4.91*       2.60*            -  6
1982T    - 1.8(37)  - 4.41        3.60*

1984R    -  .9(15)  -  .16      - 4.07

1990P    - 6.9(42)  -  .39        1.55*            -  6

1992R    - 2.0(12)  - 3.02      - 1.55

2000     - 5.1(53)     .75*     -  .35             - 14
2002     - 1.6      - 4.30      - 2.32

"D" identifies a year of intervening decline in economic activity. "P" identifies a year containing a recessionary peak in activity. "R" identifies a first or second year of recovery after a trough year. "T" identifies a year containing a recessionary trough in economic activity.

The parentheses associated with the January declines show the number of days that the S&P composite was down one percent or more during the year in question.

*The asterisks identify changes in the prime rate amounting to .25 percentage points or more. If the U.S. economy was not in the midst of a recession these cases were always followed by a recessionary peak in from three to 24 months during the 1947-2001 period. Non recessionary cases with an increase of .25 percentage points or more for the prime rate on both a January-to-January basis and a preceding year-to-year basis were followed by recessionary peaks in from three to only eleven months. "R" type of recovery years were all associated with lead times of 19 or more months.


Table 3.6

Jittery Years When the S&P Composite Experienced 43 or More One Day Declines of One Percent or More.


Year     Number of One
         Percent Declines

1929P        52
1930D        79
1931D       106
1932D       107
1933T        84
1934R        64
1935R        45

1937P        64
1938T        83
1939R        56

1973P        43
1974D        68

2000F        53
2001P        54

"D" identifies a year of intervening decline in economic activity.

"F" identifies a year which was followed by a recessionary peak.

"P" identifies a year containing a recessionary peak.

"R" identifies first and second years of recovery following a trough year.

"T" identifies a year containing a trough in economic activity.


Table 3.7

Prolonged New High Declines in the S&P Composite Lasting More than 167 Trading Days and Economic Recessions.


   Dates of S&P      Closing Values S&P   Percent  Trading Days  Duration NBER
  Peaks   Troughs      Peaks   Troughs    Decline  to Trough     Recessions
                                                                 in Months

09/07/29  06/01/32     31.92      4.40     -86.2       811           43#

08/02/56  10/22/57     49.74     38.98     -21.6       307            8

08/03/59  10/25/60     60.71     52.30     -13.9       311           10

11/29/68  05/26/70    108.37     69.29     -36.1       369           11

01/11/73  10/03/74    120.24     62.28     -48.2       436           16#

11/28/80  08/12/82    140.52    102.42     -27.1       430           16#
10/10/83  07/24/84    172.65    147.82     -14.4       199   Recovery Relapse

03/24/00  10/09/02   1527.46    776/76     -49.1       637            ?#

The hatch marks identify recessions associated with stock market crashes with trading day declines lasting more than 400 days. All of these dated recessions have prolonged duration records.


Table 3.8

Buying the S&P Composite at the End of the Month Following a Cumulative Increase in the Civilian Unemployment Rate of .9 Percentage Points.

Purchase    Closing   Percentage Change S&P to
Month       Value      Interim   One Year
            S&P        Low       Later

Feb. 1949     14.62    - 7.3     17.8

Dec. 1953     24.81    -  .0*    45.0

Dec. 1957     39.99    +  .9*    38.1

Nov. 1960     55.54    -  .4*    28.4

Apr. 1970     81.52    -15.0     27.5

Aug. 1974     72.15    -13.7     20.4

May  1980    111.24    -  .7*    19.2

Dec. 1981    122.55    -16.4     14.8

Dec. 1990    330.22    - 5.7*    26.3

Sep. 2001   1040.94    - 25.4   -21.7

*The asterisks identify modest interim declines (or gains) where the bear market low occurred before the end of the purchase month.



Essay 4:

Economic Recessions and Interest Sensitive Components of GDP

Edward Renshaw
Professor of Economics
State University of New York at Albany

While none of the individual components of real GDP are very reliable recessionary indicators, history would suggest that a noteworthy loss of investment confidence in more than one interest sensitive sector of the U.S. economy will often spread to other sectors and help to tip the economy into a recession.

This point is illustrated in Table 4.1, which shows the lead times from NBER peaks during business expansions for three or more annualized quarterly growth rates for real GDP, consumer durables, non residential equipment and software, and residential investment amounting to 1.9 percent or less during business expansions--which have occurred after or are associated with a sustained increase in the prime rate charged by banks of a third or more from its cyclical low.

Other investment categories with a propensity to "collapse" during economic recessions (such as changes in business inventories and non residential structures) are ignored on the grounds that they tend to be lagging indicators at recessionary peaks.

The accelerator model, which is discussed in Essay 7 provides a rational for being especially concerned about real GDP growth rates in the unstable range of from zero to two percent.

All of the recessionary peaks experienced from 1947 to 2001 were preceded by at least one quarter with three poor growth rates for the economic variables in Table 4.1. The longest lead time, following a one-third increase in the prime rate from its cyclical low, is 19 months. This lead time is associated with the Korean War build-up year of 1951 and can easily be eliminated on the basis of no dramatic declines in either real GDP or its components with leading indicator properties.

For the eight quarters where the growth rates for real GDP and its components were all under 1.9 percent the lead times from the business peaks identified by the National Bureau of Economic Research are all in the comparatively narrow range of from seven to zero months.

For the nine quarters when the annualized decline in residential investment was equal to 9.4 percent or more the lead times are all in the even narrower range of from 5 to zero months.

Preliminary estimates of gross domestic product are usually available about one month after a quarter is over. The more serious problem with using poor growth rates for real GDP and some of its more interest sensitive components to help identify business peaks is that some of the lead times may have been too short to enable the Fed to take actions that might have prevented another recession.

At the end of 2000 all four of our interest sensitive variables were signaling a possible recession and this may have encouraged Fed Chairman Alan Greenspan to promptly initiate a series of Fed funds rate cuts.

Once the economy has peaked out recession watchers should keep a close eye on the ratio of real gross private domestic investment (GPDI) to real GDP. The higher this ratio in a quarter containing an NBER peak, other things equal, the longer the recession is likely to be. Rapid declines in real GPDI in the following quarter, however, have tended to shorten recessions. See Table 4.2.


Table 4.1

Recessionary Lead Times for Cases of Three or More Quarterly Growth Rates for Real GDP, Consumer Durables, Non Residential Equipment and Software, and Residential Investment Equal to 1.9 Percent or Less During Business Expansions that Are Associated with or Have Occurred after a Sustained Increase in the Prime Rate of a Third or More from Its Cyclical Low.

                          

                  ------Annualized Percentage Changes-----
Quarter  Prime    Real  Consumer   Equipment  Residential  Lead Time in
         Rate Up  GDP   Durables   Software   Investment   Months from
         Ratio                                             Bus. Peak
          (1)n    (2)      (3)          (4)        (5)        (6)

1948-3   1.33     1.7      ---        - 5.2*     -11.9*       - 2

1951-4   1.42      .9    - 3.0        - 3.7*       ---        -19
1952-3   1.50     ---    -23.6*       -47.7*     - 4.1        -10
1953-2   1.62     ---    - 2.3        -  .9        1.3        - 1

1956-3   1.33    - .5    - 7.6*         ---      - 7.5        -11
1957-2   1.33    - .9    - 9.8*       - 1.2      - 9.3*       - 2

1959-4   1.43     1.3    -18.7*       -  .8      -11.2*       - 4

1969-2   1.83     1.0    - 1.3          ---      - 4.2        - 6
1969-4   1.89    -1.9    - 2.1        - 2.0      -26.2*         0

1973-2#  1.58     ---    - 6.3*         ---      -18.8*       - 5
1973-3   2.08    -1.6    - 3.1          ---      -15.9*       - 2

1979-1   1.88     1.0    - 4.1          ---      - 9.0*       -10
1979-2   1.86      .3    - 8.9*       - 4.4*     - 6.9        - 7
1979-4   2.45     1.3    -10.2*       - 5.4*     -14.8*       - 1

1981-2   1.80    -2.8    -15.2*         ---      -14.6*       - 1

1989-4   1.40     1.4    -12.6*       - 7.2*     - 6.5        - 7
1990-2   1.33      .9    -13.1*       - 8.5*     -17.2*       - 1

2000-4   1.58     1.1    - 5.3*       - 5.4*        .0        - 3  

# The hatch mark identifies a special quarter when only two of the GDP components declined but they were both relatively large asterisks declines.

(1)n. The monthly average prime rate at the end of the quarter expressed as a ratio to the monthly cyclical low value for the prime rate in the preceding business expansion.

* The asterisks identify consumer durable declines amounting to five percent or more, declines in non residential equipment and software amounting to three percent or more, and declines in residential investment amounting to nine percent or more. Since the Korean War wind down recession of 1953-54, the lead from the next business peak identified by the National Bureau of Economic Research in the last column have all been equal to seven months or less for quarters with two or more asterisks.

Source of basic data: The annualized growth rates for real GDP and its components can be obtained from the National Income and Product Account Table 8.1.


Table 4.2

Gross Private Domestic Investment (GPDI) as a Percent of GDP Expressed in Chained 1996 Dollars During Quarters Containing an NBER Recessionary Peak in Economic Activity and the Duration of the Recessions Identified by NBER.


Peak       GPDI %   Monthly Duration   % Decline GPDI
Quarter    of GDP   of Recession       Following Quarter

2001-1     18.00          ?              - 4.7

1973-4     14.82         16              - 5.9

1981-3     14.51         16              - 3.4

1980-1     14.35          6*             - 9.0

1990-3     13.58          8*             - 6.9

1948-4     13.37         11*             -15.1

1969-4     12.67         11              - 3.2

1957-3     11.62          8*             - 8.4

1960-2     11.52         10              -  .6

1953-3     11.26         10*             - 8.0

*The asterisks identify monthly durations associated with following quarter declines in real GPDI amounting to 6.9 percent or more in the last column. Recessions with rapid first quarter declines in GPDI tend to be of shorter duration than those that get off to a slow start. This has been especially the case since the Korean War wind down recession of 1953-54.

Source of basic data: The Survey of Current Business, August 2000.



Essay 5:

Employment Recessions and Their Duration

Edward Renshaw
Professor of Economics
State University of New York at Albany

Of the twelve components in the Commerce Department's original index of leading economic indicators, which was first released in November 1968, only three (the average workweek in manufacturing, residential building permits, and stock prices) have survived without substantial modification.

The average weekly hours of nonsupervisory workers in manufacturing provides a good example of an enduring component of the index of leading indicators that has always peaked out in advance of an economic recession but has sometimes not declined enough to provide a very reliable warning of its occurrence (Renshaw 1992, Table 1.20).

Evidence in support of a more stable economy that may have helped to prolong the most recent business expansion, however, is especially apparent when one examines cyclical fluctuations in nonagricultural payroll employment, and its major components: employment in goods producing industries and service industries. Since the mild recession of 1960-61 employment recessions have been largely confined to layoffs in goods producing industries.

The good news with regard to employment in manufacturing is that automation and other improvements in productivity have reduced the share of total employment in goods producing industries by more than 50 percent since the post World War II peak in 1953. From March 1980 to December 1999 there was actually a net reduction in the number of persons classified as being employed in goods production at a time when real GDP is estimated to have increased by more than 80 percent.

The shrinking share of nonsupervisory workers in recession prone, goods producing industries has helped to moderate recent employment recessions in comparison to those which occurred from 1948-58. See Table 5.5.

The expense of automating has also made it more advantageous for some industries, such as the automobile industry, to offer rebates rather than cut production and lay off skilled workers when consumer demand is weak. This should help to stabilize the demand for motor vehicles and lengthen business expansions.

Since the mild and rather unexpected recession of 1960-61 employment in goods producing industries has usually peaked at least four months in advance of the recessionary peaks identified by NBER. The only exception, so far, is a one month lag after the recessionary peak of November 1973 when peace time wage and price controls were still in effect and the Fed was apparently hoping that the inflationary shock resulting from crop failures in the Soviet Union and other parts of the world would soon moderate. See Table 5.1. A failure on the part of the Board of Governors to raise short term interest rates as rapidly as the consumer price index during the last half of 1973 encouraged many people to invest in new houses and other types of durable goods rather than have the purchasing power of their savings diminished by inflation while on deposit in commercial banks and other types of thrift institutions.

The 18 month lead time for goods producing employment before the July 1990 peak in business activity was probably one of the factors that encouraged the Board of Governors to lower the federal funds rate by 1.70 percentage points between March 1989 and July 1990. While the reduction in short term interest rates wasn't sufficient to prevent the employment recession of 1990-92, there is a possibility that this recession wouldn't have occurred if it weren't for the oil price shock and economic uncertainty associated with Iraq's invasion of Kuwait in August 1990.

The civilian unemployment rate, is one of the most widely publicized cyclical indicators. It is classified as being a leading economic indicator at recessionary peaks and a lagging indicator at recessionary troughs. From 1947- 2001 the U.S. economy was always mired down in an NBER type of recession after a cumulative increase in the civilian unemployment rate amounting to .9 percentage points or more. See Table 5.2. During August 2001 the unemployment rate was up one full percentage point from its 2000 monthly low of 3.9 percent.

Increases in the civilian unemployment rate during the nine recessions from 1947 to 1994 ranged from a low of 2.2 percentage points for the short-lived recession of 1980 to a high of 4.5 percentage points for the 1948-49 recession. The highest unemployment rate in the post World War II period was 10.8 percent for December 1982 and can be compared to an average unemployment rate of 24.9 percent for all of 1933.

The Duration of Payroll Employment Recessions

While economists have not had much success at developing models that do a good job of explaining the duration of business expansions, some progress has been made in explaining the duration of employment recessions.

Since much of the reduction in real GDP during an economic recession can be explained by a reduction in inventory investment, it is not unreasonable to suppose that the sooner the peak in industrial production, other things equal, the sooner an inventory recession will be over.

In the post 1947 period the duration of employment recessions has been about equal to 15 months minus the monthly lead time for industrial production at its peak relative to the peak in payroll employment. See Table 5.3. The implication would seem to be that a year or more may be required to turn an industrial recession around and that lower interest rates may not be very effective at halting a cyclical decline in payroll employment that begins before or at about the same time as the decline in industrial production. This formula was predicting that the recemt payroll recession might bottom out in December of 2001.

That did not happen, however. There was a bottom in April 2002 but the recovery has been so weak as of November 2002 that the employment recession may not be over.

Another way to estimate the duration of payroll employment recessions is to calculate the percent decline during the first three months of the recession. The steeper the decline, other things equal, the shorter the recession. See Table 5.7. The gain in employment during the three months prior to the payroll recession is not nearly as helpful. As more information becomes available, however, it becomes increasingly clear that modest declines in payroll employment aren't always associated with prolonged recessions. See the last column of this table related to six month declines and the last column of Table 5.3.

A Sequential Approach to Recession Forecasting

From 1946-62 the U.S. economy never experienced an economic recession until at least three months after a sustained increase in the prime rate charged by banks of a third or more from its preceding cyclical low. The data in Table 5.4 would suggest, however, that the U.S. economy has become more resistant to tighter monetary policy in recent decades.

In the period from 1963-2001 our economy never experienced a recession until at least five months after the prime rate was up more than fifty percent from its cyclical low. During two of the last six recessions the U.S. economy did not peak out until 16 months after a cumulative increase in the prime rate of more than 50 percent from its cyclical low.

Once the prime rate has signaled a near term recession it makes sense for economy watchers to pay close attention to the behavior of the average weekly hours of production or nonsupervisory workers in manufacturing which has a long history of being included in the index of leading economic indicators which was pioneered by the National Bureau of Economic Research and eventually turned over to the Conference Board. Cumulative peak related declines of .3 hours for this variable following the prime rate signals in the 1946-2001 period were always followed by a recessionary peak in from one to only eleven months if the economy was not already suffering from a recession, as was the case during February 1974. The August 2000 signal for this variable implied that the longest expansion in business cycle history should end either in the year 2000 or the first six months of 2001.

Once the prime rate signal has been followed by a shorter work week signal the historical data would suggest that economy watchers should pay close attention to associated or following back-to-back increases in the civilian unemployment rate. Since 1959 the U.S. economy has always experienced a recessionary peak after this type of triple whammy.

Another way to try to determine whether the U.S economy is in the midst of a recession is to examine monthly declines in nonagricultural payroll employment. Since the Korean War build-up year of 1952 our economy has always been involved in a recession once payroll employment experienced at least a third lower closing total spread over a three to five month period.

The Dating of Recessionary Peaks

The leads and lags associated with various coincident indicators in Table 5.6 would strongly suggest that NBER's dating of recessionary peaks use to be heavily influenced by the behavior of personal income less transfer payments expressed in deflated dollars. More of the lead- lag times for this indicator are in the minus one to plus one range than for any other component of BEA's index of coincident indicators prior to the current recession.

It should be appreciated, however, that this indicator has experienced numerous ups and downs on a month-to-month basis during business expansions and has been subject to a lot of revisions spread over five or more months which may have provided NBER with a nice excuse for not getting involved in dating business peaks until after some recessions were over.

There is also a question as to whether personal income less transfer payments will continue to be a fairly reliable predictor of recessionary peaks in a new era where many corporate executives and some of the employees in start up companies have been rewarded for a large portion of their services with stock options (which can't be exercised immediately) rather than ordinary wages and salaries. Another reason for ignoring this indicator is that it is no longer computed on a regular basis.

While persons concerned about the possibility of another recession should pay close attention to industrial production and manufacturing and trade sales, which have often been leading indicators prior to NBER peaks, there is not much doubt that most people (including NBER's dating committee) are now more concerned about jobs than other indicators of a recessionary peak.

One of the problems with recessionary peaks and troughs of a judgmental nature is that they don't provide one with a reliable indication of how severe the recessions were and may have discouraged economic analysts from investing more time and effort in developing indexes of leading economic indicators that do a good job of tracking specific indicators such as payroll employment or industrial production.

If future dips in most coincident indicators were to become shallower and less prolonged there might be some difficult problems of a controversial nature in deciding whether there has been a recession.

These considerations and the recent propensity for actual and prospective recessions to become highly politicized suggest that it might not be such a bad idea if the National Bureau of Economic Research simply abandoned the task of trying to date business peaks.

Reference

Renshaw, Edward (1992). The Practical Forecasters' Almanac(Burr Ridge, Illinois: Irwin Professional Publishing).


Table 5.1

Recessionary Lead Times for the Number of Employees on Nonagricultural Payrolls in Goods Producing Industries, Business Cycle Indicator Series 40.

                                                                     

         Date of              Goods Employment     Lead Times From NBER
-------------------------    ------------------      Peaks in Months
Goods    Goods     NBER       Own         NBER     --------------------
Employ.  Employ.   Peak       Peak        Peak     Unadjusted  Adjusted
Peak     Down 1%             ----Thousands-----
  (1)      (2)      (3)        (4)         (5)        (6)        (7)n

Sep. 48  Dec. 48  Nov. 48    18,916      18,794       - 2        + 1

Apr. 53  Sep. 53  July 53    21,304      21,266       - 3        + 2

Dec. 56  July 57  Aug. 57    21,291      20,942       - 8        - 1

Feb. 60  Mar. 60  Apr. 60    20,903      20,716       - 2        - 1

July 69  Jan. 70  Dec. 69    24,495      24,354       - 5        + 1

Dec. 73  July 74  Nov. 73    25,264      25,214       + 1        + 8

July 79  Mar. 80  Jan. 80    26,606      26,448       - 6        + 2

Mar. 81  Nov. 81  July 81    25,654      25,630       - 4        + 4

Jan. 89  May  90  July 90    25,361      24,949       -18        - 2

July 00  Apr. 01  Mar. 01    25,774      25,602       - 8          1

Footnotes for Table 5.1

(7)n. Months from the NBER peak date in column (3) to the date in column (2) when goods employment was down one percent or more from its own peak.

Source of basic data: The Survey of Current Business, January 1995 and more recent issues.


Table 5.2

Fluctuations in the Civilian Unemployment Rate of .9 Percentage Points or More and U.S. Business Cycles

                                                                     

                                          Unemployment Rate   Lead Time
Unemploy.   U up .9%  Unemploy.  Business  ----------------  (in Months)
  Trough    Points      Peak       Peak    Trough    Peak  Total Adjusted

                                              (1)     (2)    (3)    (4)n

Jan. 1948  Jan. 1949  Oct. 1949  Nov. 1948    3.4     7.9    -10     +2

Jun. 1953  Nov. 1953  Sep. 1954  July 1953    2.5     6.1    - 1     +4

Mar. 1957  Nov. 1957  July 1958  Aug. 1957    3.7     7.5    - 5     +3

Feb. 1960  Oct. 1960  May  1961  Apr. 1960    4.8     7.1    - 2     +6

May  1969  Mar. 1970  Aug. 1971  Dec. 1969    3.4     6.1    - 7     +3

Oct. 1973  July 1974  May  1975  Nov. 1973    4.6     9.0    - 1     +8

May  1979  Apr. 1980  July 1980  Jan. 1980    5.6     7.8    - 8     +3

July 1981  Nov. 1981  Dec. 1982  July 1981    7.2    10.8      0     +4

Mar. 1989  Nov. 1990  June 1992  July 1990    5.0     7.7    -16     +4

Oct. 2000  Aug. 2001  Dec. 2001  Mar. 2001    3.9     5.8    - 5     +5

(4)n. Months from the business peaks identified by NBER after the civilian unemployment rate has reversed itself on the upside by at least .9 percentage points after experiencing a trough.

Source of Data: Economic Report of the President.


Table 5.3

The Duration of Employment Recessions in Months Ranked in Order of the Monthly Lead time for Industrial Production

                                                                     

  Date of Peak for                                                 Peak-Trough
----------------------  Monthly Lead Time  Duration of Recession   % Decline
Industrial  Payroll       Industrial        Actual    Predicted    Payroll
Production  Employment    Production        -----in months-----    Employment

                             (1)              (2)        (3)n         (4)

 Nov. 73    Oct. 74          -11                6          4         -2.9

 May  79    Mar. 80          -10                4          5         -1.4

 Sep. 00    Mar. 01          - 6               13?         9         -1.5?

 Oct. 69    Mar. 70          - 5                8         10         -1.5


 Jan. 60    Apr. 60          - 3               10         12         -2.3

 July 48    Sep. 48          - 2               13         13         -5.2

 Mar. 57    Mar. 57          - 0               14         15         -4.3

 July 81    July 81          - 0               16         15         -3.0

 July 53    Jun. 53          + 1               14         16         -3.5

 Sep. 90    Jun. 90          + 3               20         18         -1.6

(3)n. The predicted duration is equal to 15 months minus the monthly lead time for the industrial production peak in column (1).

Source of data: The old cyclical indicator section of the Survey of Current Business, October 1994.


Table 5.4

A Sequential Approach to Recession Forecasting Based on the Behavior of the Prime Rate Charged by Banks and Some Employment Indicators.


      Recessionary Signals          Lead Times in Months from NBER
----------Monthly Dates---------    -------Recessionary Peaks-----
Prime     Weekly    Unemployment    Prime   Weekly   Unemployment
Rate      Hours     Rate            Rate    Hours    Rate

    Cases Where the Prime Rate Was Up a Third or More from its Cyclical Low

Aug. 48   Sep. 48   Dec. 48         -  3     -  2     +  1

Nov. 51   June 53   Aug. 53         - 20     -  1     +  1

Sep. 56   Mar. 57   June 57         - 11     -  5     -  2

Sep. 59   Feb. 60   July 60         -  7     -  2     +  3

   Cases Where the Prime Rate Was Up More than 50 Percent from Cyclical Low

Jan. 69   Nov. 69   Feb. 70         - 11     -  1     +  2

June 73   Feb. 74   Feb. 74         -  5     +  3     +  3

Sep. 78   Apr. 79   Jan. 80         - 16     -  9        0

Dec. 80   Feb. 81   Sep. 81         -  7     -  5     +  2

Mar. 89   Aug. 89   Aug. 90         - 16     - 11     +  1

May  00   Aug. 00   Apr. 01         - 10     -  7     +  1

The recessionary signals for the average weekly hours of production or nonsupervisory workers in manufacturing are based on following cumulative peak declines of .3 hours after the prime rate was up a third or more from 1946-61 and over 50 percent from 1962-2000.

The unemployment signals are based on back-to-back monthly increases of .1 percent or more that are associated with or follow the average weekly hours signals.

Another interesting way to begin the task of identifying recessionary peaks since 1962 is to look for cyclical increases in the prime rate equal to 3.50 percentage points or more. The lead times for this signal are about the same, on the average, as an increase in the prime rate of over 50 percent.


Table 5.5

Nonagricultural Payroll Recessions and How They Compare to the Recessionary Peaks and Troughs Dated by the National Bureau of Economic Research.


Peak Employment    Employment Trough    Percentage Declines In Employment
   Months              Months             Peak to     Peak to Third
                                          Trough      Lower Monthly
                                                      Total

Sep. 1948(- 2)    Oct. 1949(  0)          - 5.2       -  .3

June 1953(- 1)    Aug. 1954(+ 3)          - 3.5       -  .3

Mar. 1957(- 5)    May  1958(+ 1)          - 4.3       -  .2

Apr. 1960(  0)    Feb. 1961(  0)          - 2.3       -  .9

Mar. 1970(+ 3)    Nov. 1970(  0)          - 1.5       -  .6

Oct. 1974(+11)    Apr. 1975(+ 1)          - 2.9       - 1.8

Mar. 1989(+ 2)    July 1980(  0)          - 1.4       - 1.1

July 1981(  0)    Nov. 1982(  0)          - 3.0       -  .4

June 1990(- 1)    Feb. 1992(+11)          - 1.7       -  .4

Mar. 2001(  0)    Apr. 2002               - 1.1?      -  .2?

The figures in parentheses are the lead times in months from the recessionary peaks and troughs identified by the National Bureau of Economic Research.

Source of basic data: Survey of Current Business, October 1994 and more recent issues.


Table 5.6

Cyclical Leads (-) and Lags (+) for Some Coincident Indicators at the Business Peaks Identified by the National Bureau of Economic Research.


NBER Peak   Employees   Personal       Industrial   Manufacturing
Dates       Non Ag.     Income Less    Production    and Trade
            Payrolls    Transfer                     Sales
                        Payments

Nov. 1948     - 2         - 1            - 4           + 1

July 1953     - 1         - 1              0           - 3

Aug. 1957     - 5           0            - 5           - 6

Apr. 1960       0         + 1            - 3           - 3

Dec. 1969     + 3         NST            - 2           - 2

Nov. 1973     +11           0              0             0

Jan. 1980     + 2           0            + 2           -10

July 1981       0         + 1              0           - 6

July 1990     - 1         - 3            + 2           - 4

Mar. 2001       0           ?            - 6           - 6

NST identifies a case with no specific turn. Personal Income less transfer payments was estimated to have experienced a double peak expressed in 1987 dollars on November 1969 and April 1970. It should be noted that this indicator is no longer computed on a regular basis by BEA.

Source of basic data: Survey of Current Business, October 1994.


Table 5.7

The Duration of Employment Recessions in Months Ranked in Order of the Percentage Declines in Nonagricultural Payroll Employment During the First Three Months of the Recession.

                                                              Employment
  Date of     Three Month % Change Employment   Duration of   Six Month
Employment         Before         After         Recession     % Change
   Peak             Peak          Peak          in Months     After Peak

Oct. 1974           .07           -1.81             6           -2.89

Mar. 1980           .42           -1.12             4           -1.04

Apr. 1960           .83           - .77            10           -1.19

Mar. 1970           .31           - .62             8           - .69


June 1990           .39           - .39            20           - .84

June 1953           .01           - .32            14           -1.52

Sep. 1948           .63           - .31            13           -2.27

Mar. 1957           .29           - .23            14           - .46

July 1981           .21           - .21            16           -1.05

Mar. 2001           .22           - .12            13?          - .63

Source of basic data: The Survey of Current Business, October 1994.



Essay 6:

Can There Be Another Recession
Without an Oil Price Shock?

Edward Renshaw
Professor of Economics
State University of New York at Albany

In 1999 total U.S. crude oil production was 38 percent less than in 1970, in spite of large oil discoveries in Alaska and improved offshore drilling technology which has boosted output in the Gulf of Mexico. Since the U.S. Department of Energy began tracking production and imports in 1973, oil imports have almost tripled, increasing from about 12 percent of total supply to more than 57 percent in 1998.

Rapid increases in oil prices in conjunction with weak economic indicators are of particular concern since they may indicate that the economy has already slipped into another recession. The eleven poorest growth years for the U.S. economy from 1948-2001, were all preceded by a year-to-year increase in the motor fuel component of the consumer price index amounting to three percent or more and a June-December increase in industrial production of .1 percent or less. See Table 6.1.

Since 1946 the U.S. economy was always mired down in an industrial recession after back-to-back increases in the average annual values for the motor fuels component of the all item consumer price index amounting to nine percent or more. See Table 6.3.

While higher oil prices may not be the cause of economic recessions they can lead to restrictive monetary policies that will help to terminate a business expansion. Except for 1953 (when the consumer price index was still increasing at a modest .7 percent annual rate), 1971 and 1976 (when the U.S. economy was beginning to recover from economic recessions), and 1999 (when oil prices were recovering from the Asian Flu slump), the Federal Reserve has not allowed the conventional money supply M1 to increase as rapidly as the consumer price index when crude oil prices were increasing at an average rate of five percent or more.

If the economy was not already in an economic recession, the Fed has often resisted the inflationary effect of large increases in crude oil prices by allowing short-term interest rates to rise at a rapid rate. The 1990-91 recession was different, however. Between March 1989 and July 1990 the Fed allowed the yield on new issues of 91 day Treasury bills to decline by 1.2 percentage points, or more than 13 percent, in an unsuccessful effort to prevent an economic recession.

How an Increase in Oil Prices Can Cause a Recession

It is easy to understand how a large increase in gasoline prices might trigger a recession in the United States. It will reduce the demand for large, gas guzzling motor vehicles produced by domestic auto makers and siphon off purchasing power from consumers to oil companies and petroleum exporting countries that might not be able to recycle the proceeds in a manner that will quickly generate an offsetting demand for goods and services produced in the USA.

U.S. oil companies will use some of the extra revenue to look for petroleum in other countries, reduce their indebtedness or repurchase some of their own shares and petroleum exporting countries will use some of their proceeds to purchase securities issued by the U.S. Treasury, that help to finance unemployment benefits, but don't generate jobs in this country. An oil induced recession that begins at the gas pump and cripples the auto industry, can then be expected to spread to other industries as consumers cut back expenditures for other goods and services to keep their motor vehicles operating.

Once an economic recession has begun in the United States it can easily spread to other countries as a result of a reduction in imports for other types of goods and services besides petroleum products. After the Arab oil embargo of October 1973, for example, U.S. imports in constant dollars declined by 13.3 percent from the fourth quarter of 1973 to the first quarter of 1975. During the milder recession of 1990-91 the reduction in real imports was 5.1 percent.

The 1990-91 oil related recession in the United States was unique in that the decline in industrial production in Canada and the United Kingdom began before the October 1990 drop in U.S. industrial output. Japan and Germany were able to weather the storm for a while but eventually slipped into recessions that were partly the result of a very anemic recovery from the 1990-91 recession in the United States.

The most disturbing aspect to the 1990-91 recession in the United States is that it occurred during a period of monetary easing without a very precipitous increase in the price of motor fuels. The implication would seem to be that the U.S. economy may be even more vulnerable to oil price shocks in the future as it becomes more dependent on imported oil. Most adults now own their own cars and in recent years many colleges and universities have had to expand their parking lots to accommodate commuting students.

From 1973 to 1991 there was a 63 percent increase in passenger car miles per gallon in the USA. This helped to lower the real price of motor fuels to levels not experienced since the great depression of the 1930s. Since the collapse of world oil prices in the mid 1980s and the relapse which occurred after the successful completion of Operation Desert Storm, however, American motorists have become infatuated with minivans, light duty trucks and sport utility vehicles that do not get very good gas mileage. This can also be expected to make the U.S. more vulnerable to another oil price shock recession.

Increases in oil production in the North Sea and other areas outside OPEC, and a recessionary drop in oil demand in some Asian countries led to a dramatic drop in crude oil prices in 1998. Production cut-backs by Saudi Arabia and other petroleum exporting countries caused the price of imported oil to rise above the 1998 level in March of 1999, however. During 2000 the price of crude oil was almost twice as high as the 1999 average price of $15.56 per barrel.

One of the problems with oil prices and most of the other recessionary indicators in this compendium is that they do not do a very good job of enabling one to identify recessionary peaks close to their occurrence. In Table 6.2 we provide readers with some adjusted lead times for five widely publicized variables with both leading and coincident properties that have been helpful at identifying the last nine recessionary peak months in close proximity to their occurrence. The indicator thresholds are based on the largest cumulative changes with the property of enabling one to identify all of these peaks with a lag of not more than two months.

Our first indicator of a possible recession is a cumulative increase in the civilian unemployment rate of .4 percentage points or more after its cyclical low. This is an important variable to keep an eye on since some economists at the Fed believe that we will be stuck with a problem of accelerating inflation if the unemployment rate is allowed to remain below 4.5 percent.

The employment variable with one of the best forecasting records is the number of employees on nonagricultural payrolls in goods-producing industries. A cumulative drop in goods-producing employment amounting to 89 thousand workers is not a healthy development and a cumulative increase of 70 thousand in the average weekly number of initial claims for unemployment insurance has often meant that a business expansion was over.

Manufacturing and trade sales expressed in constant dollars have a history of being perhaps the best "coincident" indicator with some leading indicator properties. Cumulative declines in this variable amounting to 1.0 percent or more have been closely linked to recessions. Our last indicator of a recessionary peak is a cumulative decline in the monthly average values for the S&P composite stock price index amounting to four percent or more.

It should be appreciated that there have been some false or rather premature signals associated with some of these recessionary thresholds. When three or more of these variables are pointing in the direction of a possible peak in business activity, however, investors should be rather cautious.

When four of these indicators were pointing in the direction of a recession six of the last dated dated business expansions were over. When all of these indicators were pointing in the direction of a recession every expansion was over.

As of March 2001 all of these indicators were pointing in the direction of a recession.


Table 6.1

Using Year-to-Year Increases of Three Percent or More for the Motor Fuel Component of the Consumer Price Index and Anemic Increases in Industrial Production Amounting to .1 Percent or Less to Forecast Poor Growth Years for the U.S. Economy.

                          
                               
Year     ---------Growth Rates for----------  Following Year
         Motor    Industrial     Following    Financial Return
         Fuel     Production     Year Real    for the S&P 
         Price    June-Dec.         GDP       Composite
           (1)         (2)          (3)           (4)

1948       13.2       - 2.5         - .6          17.8**

1953        6.0       - 6.9         - .7          51.2**

1957        4.4       - 5.9         -1.0          42.4**

1960        3.0       - 5.0          2.3          26.6**

1969        3.0       -  .5           .2           3.5**

1973        9.9          .1         - .6         -26.0
1974       35.3       - 8.0         - .4          36.9**

1979       35.3       -  .6         - .2          31.5**

1981       11.4       - 2.3         -2.0          20.4**

1989        9.4       -  .3          1.8         - 3.1
1990       14.4       - 2.1         - .5          30.0**

2000       28.4       -  .4           .3         -11.8

Footnotes for Table 6.1

**The financial return associated with the S&P composite stock price index the year after an increase in the price of motor fuels of 3 percent or more and a June-December decline in industrial production of .5 percent or more.

Source of basic data: and Economic Report of the President. For an interesting discussion of major oil price shocks see: Philip Verleger, "Third Oil Price Shock: Real or Imaginary?" Oil and Gas Journal June 12, 2000, pp. 76-88.


Table 6.2

Lead Times in Months from the Business Peaks Identified by the National Bureau of Economic Research for Five Economic Indicator Thresholds with Both Leading and Coincident Properties.


  NBER   Civilian   Goods       Initial   Manufact   S&P          Median
  Peak   Unemploy   Producing   Unemploy  Trade      Composite    Lead
  Date   Rate       Employment  Claims    Sales                   Time

            (1)        (2)        (3)        (4)        (5)        (6)

Nov. 48    - 9        - 1        - 6          2        - 3        - 3

July 53      2          1        -24        - 1        - 3        - 1

Aug. 57    - 3        - 3          1        - 4        -10        - 3

Apr. 60    - 1        - 1          1          1        - 7        - 1

Dec. 69      1        - 1          2        - 1        -11        - 1

Nov. 73      2          2          2          1        - 8          2

Jan. 80    - 5        - 5        - 9        - 9          2        - 5

July 81      2        - 2          2        - 2        - 5        - 2

July 90    - 8        -13        - 9        - 3          1        - 8

Mar. 01      0        - 2        - 4          0        - 5        - 2

(1) Civilian unemployment rate up .4 percentage points from cyclical low.

(2) Goods-producing employment down 89 thousand from cyclical peak.

(3) Average weekly initial unemployment insurance claims up 70 thousand from cyclical low.

(4) Manufacturing and trade sales expressed in constant dollars down 1.0 percent or more from cyclical high.

(5) Monthly average value for the S&P composite down 4 percent or more from cyclical high.

Source of basic data: Survey of Current Business, October 1994 to January 1995.


Table 6.3

Yearly Increases in the Average Annual Values of the Motor Fuels Component of the All Item Consumer Price Index of Nine Percent or More and Their Relation to Recessionary Peaks in Industrial Production.


Year   Growth Rate      Date of Recessionary
       CPI Index for    Peaks in Industrial
       Motor Fuels      Production

1947      13.3                 
1948      13.2          July

1973       9.9          November
1974      35.3                  

1979      35.3
1980      38.9          March
1981      11.4          July

1989       9.4          April
1990      14.4          September

1999       9.2
2000      28.2          September

Since 1946 the U.S. economy has always been in the midst of an industrial recession after back-to-back increases in the average price of motor fuels of nine percent or more.



Essay 7:

The Accelerator Principle Revisited

Edward Renshaw
Professor of Economics
State University of New York at Albany

In 1990 three economists received the Nobel Prize for their contributions to a relatively new branch of the dismal science that can perhaps best be described as financial economics. The contributions of Harry Markowitz, William Sharpe and Merton Miller have revolutionized the teaching of finance in business schools and have helped to crowd the IS-LM framework, which was made famous by another Nobel Laureate, to the back of most text books on money and banking.

The blurry nature of what used to be separate disciplines has led the National Association of Colleges and Employers, which tracks campus recruiting activity, to make no distinction between economics and finance majors when reporting salary offers that are made to college graduates.

The ideas of Markowitz, Sharpe and Miller, in any event, were well received by Nobel Laureates Paul Samuelson, James Tobin and Franco Modigliani, and a wider spectrum of economists who believe that financial markets are relatively efficient. While the efficient market hypothesis has been challenged in recent years, the economics profession has not had a lot of success at developing models that refute the random walk hypothesis and allow one to take advantage of exceptional circumstances.

There is one model, however, which has withstood a test of both time, and a major change in the way GDP and its components are now deflated, that may be of some value in helping one to understand the bull market of 1995-99.

During the last seven recessions the cumulative decline in gross private domestic investment has always been greater than the decline in real GDP. The volatility of investment and its sensitivity to declines in real GDP can be partially explained on the basis of an accelerator principle.

This principle is firmly embedded in pre-Keynesian literature. Michael Evans (1969) has noted that its origins can be traced back to publications of Aftalian, Bickerdike and Hawtrey in the early 1900s. Probably the best-known early study is J. M. Clark's (1917) article on "Business Acceleration and the Law of Demand".

An updated accelerator model which was publicized by Renshaw in (1992, Table 1.32) is illustrated in Table 7.1 where we show that the year-to-year growth in real gross private fixed domestic investment has often been about equal to three times the growth rate for real GDP minus six percentage points. This relationship can be derived from a Cobb Douglas production function if we assume that fixed investment is proportional to the flow of productive services from the capital stock.

Where Q is real GDP, K is the flow of productive services from the capital stock and T is a time trend that represents technology, labor and other resources that have been left out of the production function we have:

                   2 1/3
              Q = T K                                           (1)

Since depreciation charges and the income from capital services are a primary source of funds for investment purposes it is not unreasonable to suppose that the aggregate amount of real fixed investment might be roughly proportional to the flow of services from the capital stock:

             I = vK                                             (2)

Substituting equation (2) into (1) and noting that small changes in variables that are related to each other in a multiplicative way are approximately equal to percentage changes we can conclude that: the percentage change in real GDP should be about equal to 2.0 percentage points plus one-third of the percentage change in real fixed private domestic investment. One of the implications of this relationship (when the equation is solved for the growth of fixed investment) is that real GDP must increase by about two percent per year just to keep private investment from falling and having a deleterious feedback effect on the growth of economic activity.

The instability of the GDP growth rate in the zero to two percent range has sometimes made it very difficult for the Federal Reserve to effectively fight inflation without pushing the U.S. economy into a recession.

When the actual growth of real fixed investment expressed in chained 1996 dollars has been at least 2.2 percentage points greater than the expected increase that is predicted by the accelerator model in Table 7.1, optimism on the part of business enterprises and developers has tended to spread to the stock market and produce following year returns for the S&P index (dividends plus price appreciation expressed as a percent of price at the beginning of the year) that have been positive. The behavior of the stock market in the year 2000 would strongly suggest, however, that a very prolonged period of super optimism can help to get some investors in trouble.

The following year financial returns the S&P composite have also been positive after a decline in real GDP expressed in 1996 dollars of minus .2 percentage points or more.

It should be emphasized that investment optimism on the part of business enterprises and developers hasn't done a very good job of protecting investors from some nerve racking "corrections" in the midst of more pronounced bull markets. Persons who purchased an index fund tracking the S&P 500 at the end of 1946, 1947, 1969, 1977 and 1993 would have had to wait a full year to be sure of getting their money back. All of these corrections, however, were followed by at least one year of double digit returns before the S&P index crashed for a long enough period of time so that its dividends were unable to compensate for lower prices.

It should be noted that positive error terms for the accelerator model in column (4) of Table 7.1 have a propensity to flock together or occur in "runs" and can be considered a good example of the Keynesian notion of "animal spirits". The seven error terms in excess of 2.2 percent from 1992-98 have now exceeded the prolonged period of super investment optimism for the U.S. economy which began toward the end of World War II and persisted from 1944-48 before ushering in the great bull market of the 1950s.

Using Quarterly Percentage Changes in Real GDP to Distinguish Between Growth Recessions and the Real Thing

To identify NBER recessions from 1948-2000 one needs at least two lower quarterly totals for real GDP in chain weighed 1996 dollars. See Table 7.2.

Quarterly declines of 1.8 percent or more for real GDP have all been closely linked to NBER recessions and are helpful at enabling one to identify recessions before they were over.

One quarterly decline of .3 percent or more in a sequence of two or more growth rates under 2.0 percent help to illustrate the instability of real GDP in the zero to two percent range and will sometimes allow one to identify an NBER recession sooner than two negative growth rates.

From 1948-2001 there was only one "growth recession" involving a sequence of three quarterly growth rates for real GDP under 2.5 percent. They were all positive growth rates beginning in the second quarter of 1989 and were soon followed by an NBER recession beginning in the third quarter of 1990.

References

Clark, J. M. (1917). "Business Acceleration and the Law of Demand," Journal of Political Economy, Vol. 25, No. 1 (March), 217-235.

Evans, Michael (1969). Macroeconomic Activity: Theory, Forecasting, and Control(New York: Harper & Row),


Table 7.1

Stock Returns and the Accelerator Relationship Between the Year to Year Growth Rates for Chain Weighted Real GDP and Gross Private Fixed Domestic Investment.

    
                                                                  
Year    Real    Actual    Predicted   Actual Minus   Following Year
        GDP   Investment  Investment   Predicted     S&P Financial
                                       Investment       Return
         (1)      (2)       (3)n          (4)            (5)

1960     2.5       .9        1.5        -  .6           26.6
1961     2.3T   -  .3         .9        - 1.2          - 8.8
1962     6.0      9.0       12.0        - 3.0           22.5
1963     4.3      7.7        6.9           .8           16.3
1964     5.8      9.7       11.4        - 1.7           12.3
1965     6.4     10.2       13.2        - 3.0          -10.0
1966     6.6      5.7       13.8        - 8.1           23.7
1967     2.5    - 1.9        1.5        - 4.0           10.8
1968     4.8      6.9        8.4        - 1.5          - 8.3
1969     3.0      6.2        3.0          3.2*           3.5

1970      .2T   - 2.1#     - 5.4          3.3*          14.1
1971     3.3      7.5        3.9          3.6*          18.7
1972     5.4     12.0       10.2          1.8          -14.5
1973     5.8      9.1       11.4        - 2.3          -26.0
1974   -  .6    - 6.3#     - 7.8          1.5           36.9**
1975   -  .4T   -10.7#     - 7.2        - 3.5           23.6**
1976     5.6      9.8       10.8        - 1.0          - 7.2
1977     4.6     14.4        7.8          6.6*           6.4
1978     5.5     11.5       10.5          1.0           18.4
1979     3.2      5.6        3.6          2.0           31.5

1980   -  .2T   - 6.4#     - 6.6           .2          - 4.8
1981     2.5      2.2        1.5           .7           20.4
1982   - 2.0T   - 7.0#     -12.0          5.0*          22.3**
1983     4.3      7.5        6.9           .6            6.0
1984     7.3     16.8       15.9           .9           31.1
1985     3.8      5.3        5.4         - .1           18.5
1986     3.4      1.2        4.2        - 3.0            5.7
1987     3.4    -  .0        4.2        - 4.2           16.3
1988     4.2      3.6        6.6        - 3.0           31.2
1989     3.5      2.7        4.5        - 1.8          - 3.1

1990     1.8    - 1.8      -  .6        - 1.2           30.0
1991   -  .5T   - 6.9#     - 7.5           .6            7.4**
1992     3.0      6.5        3.0          3.5*           9.9
1993     2.7      8.1        2.1          6.0*           1.3
1994     4.0      9.1        6.0          3.1*          37.1              
1995     2.7      6.0        2.1          3.9*          22.7
1996     3.6      9.3        4.8          4.5*          33.1
1997     4.4      9.6        7.2          2.4*          28.3
1998     4.3     11.4        6.9          4.5*          20.9
1999     4.1      7.8        6.3          1.5          - 9.0
2000     3.8      6.1        5.4           .7          -11.8
2001      .3     -3.8#      -5.1          1.3             ?

Footnotes for Table 7.1

(3)n. The predicted growth rate for fixed investment is equal to three times the growth rate for real GDP in column (1) minus six percentage points.

T denotes years containing a trough in business activity.

# identifies year to year declines in fixed investment in excess of two percent. Since the Korean War wind down recession of 1953-54 all of these declines have been associated with economic recessions.

* denotes years when the difference between the actual and the predicted growth rate for fixed investment was equal to 2.2 percentage points or more. From 1946-2000 all of these years were followed by a positive financial return for the S&P index.

** denotes financial returns following a year to year decline in real GDP in column (1) of minus .3 percentage points or more. All of these financial returns have been positive.

Source: This is a partially up-dated table that was first published in The Practical Forecasters' Almanac(Burr Ridge, Illinois: Irwin Professional Publishing, 1992), Table 1.32, pp. 131-32. The growth rates for real GDP and gross private fixed domestic investment have been revised to correspond with the chain weighted values published in The Survey of Current Business, December 1999.


Table 7.2

Quarterly Dips and Back to Back Changes in Real GDP of Less Than 2.5 Percent, 1948-2000.

Date First     --Quarterly Percentage Changes in Real GDP--
Slow Quarter   1st   2nd   3rd   4th  Fifth  Sixth  Seventh

1948-3R        1.7    .8  -5.5  -1.1*  4.6  -4.0

1953-3R       -2.5  -6.3* -2.0    .6

1955-4G        2.1  -1.7

1956-3        - .5

1957-1R        2.3  - .9   4.0  -4.1*-10.3   2.4

1959-3G       - .2   1.3

1960-2R       -2.0    .7  -5.0*  2.3

1967-2        - .3

1969-2R        1.0   2.3  -1.9  - .6*   .8   3.6  -4.2

1973-3R       -1.6   3.4  -3.0   1.1  -4.4* -2.2  -5.0

1977-4G         .5   1.1

1979-1G        1.0    .3

1979-4R        1.3   1.3  -7.9  - .6*

1981-2R       -2.8   4.9  -4.6  -6.5*  1.7  -1.9    .3

1989-2G        2.2   1.9   1.4

1990-2R         .9  - .7  -3.2* -2.0   2.3   1.0   2.2

1993-1        - .1

1995-1G        1.5    .8

2000-3R         .6   1.1  - .6  -1.6* - .3

"G" identifies growth recessions and "R" cases which were eventually dated as real recessions by The National Bureau of Economic Research. To identify NBER recessions one needs at least two lower quarterly values for real GDP. Quarterly declines of 1.8 percent or more have all been closely linked to NBER recessions and are helpful at enabling one to identify some recessions before they were over.

*The asterisks identify declines associated with a second lower annualized quarterly total for real GDP.

Source of basic data: The Survey of Survey of Current Business.



Essay 8:

Rehabilitating the Keynesian Multiplier

Edward Renshaw
Professor of Economics
State University of New York at Albany

While most large scale econometric forecasting models are basically Keynesian in character, there has been a notable reluctance on the part of textbook writers to confront economic theory with fact. Part of the over-reliance on pure theory may be related to the difficulty that has been encountered in establishing plausible direct multiplier relationships between aggregate income and the two most important fiscal policy variables, government spending and taxes. The absence of easily verified multipliers in the early post World War II period led Milton Friedman to conclude in the words of Paul Samuelson (1976), "that fiscal policy per se has essentially no predictable effect of any significance on the prospects for inflation or deflation, for high employment or mass unemployment."

Keynesian economics is basically disequilibrium or depressionary economics. During periods of reasonably full employment one can use production possibility curves involving a tradeoff between guns and butter to seriously question the notion of sizable multiplier effects for government spending. Three economic recessions in about one decade, the monetary revolution of 1979, and the Economic Recovery Tax Act of 1981, however, helped to produce a "window of opportunity" where it was easier to explain changes in the national income and product accounts on the basis of a rather simple Keynesian model which focuses one's attention on the problem of large budget and trade deficits in the U.S. (Renshaw 1990). A large decline in the personal saving rate and some major changes in the way real GDP is computed, though, have made it desirable to up-date that version of the Keynesian multiplier.

In Chapter 8 of The General Theory of Employment, Interest and Money it is stated that: "The fundamental psychological law, upon which we are entitled to depend with great confidence both a priori and from our knowledge of human nature and from the detailed facts of experience, is that men are disposed, as a rule and on the average, to increase their consumption as their income increases, but not by as much as the increase in their income."

The simplest way to describe this type of consumption function is with the linear equation, C = a + bY. When this equation is fit to cross-sectional data for a particular time period one obtains a value for "a" that is positive and a value for "b" that supports the Keynesian hypothesis that the marginal propensity to consume "mpc" is less than one.

In the analysis which follows we will assume that the marginal propensity to consume in relation to current income is in the vicinity of .5. This number was first obtained from empirical data by Paul Samuelson in 1941. In an appendix to Chapter 11 of A. H. Hansen's Fiscal Policy and Business Cycles he found that "In the period between the two world wars, the marginal propensity to consume relative to income produced was approximately .5."

While there may have been numerous years and periods when the mpc was not equal to .5, there are a lot of advantages in featuring this number when confronting theory with fact and trying to reconcile textbook multipliers with those derived from complex models of the US economy.

In deriving multipliers, however, we will ignore the Commerce Department's estimates of disposable income and follow the usual textbook practice of utilizing a broader and less conventional definition of disposable income. Our starting point is the accounting identity, income equals consumption plus investment plus government purchases of goods and services plus net exports:

       Y = C + I + G + NX                                     (1)

Consumption is now regarded to be a function not only of current income but also wealth which can be approximated by consumption or the value of income in the preceding period (Duesenberry 1949, Modigliani 1947 and Friedman 1957). Since positive amounts of net exports cannot be consumed in the home country and are unlikely to have as big an effect on local output as domestic investment, if they are used to finance investment in other countries, this variable will be subtracted from GDP along with net taxes to arrive at a useful definition of disposable income that can be simply linked to both consumption and aggregate income:

       C = aY   + b(Y - NX - T)                                 (2)
             -1
Setting Y equal to nominal GNP in billions of current dollars and fitting this type of consumption function to U.S. data for the 1952-86 period allowed this analyst to obtain the following least squares regression (Renshaw 1990):

       C = 1.59 + .25Y   + .5(Y - NX -T)                       (3)
                      -1

Disregarding the constant term for equation (3), which is not very significant from a statistical point of view, and substituting rounded values for the remaining coefficients into accounting identity (1) enables one to obtain the following multiplier equation:

       Y = .5Y   + 2I + G + (G - T) + NX                       (4)
              -1

In equation (4) we have separated the government expenditure multiplier into two G variables to allow one to more easily assess the economic implications of large trade and federal budget deficits in the national income and product accounts.

In Table 8.1 all of the errors for our featured version of the Keynesian multiplier model implied by equation (4) are assumed to change the propensity to spend out of the previous year's income. The coefficient for the lagged income (or GDP) variable in column (6) is expressed in percentage points and is referred to as the "shifting propensity to spend".

While some of the error terms for this multiplier model may indeed be related to shifts in the marginal propensity to consume out of current income, rather than the previous year's income, a consolidation of the error terms in conjunction with lagged GDP can serve a useful purpose in enabling one to more easily appreciate, in relative terms, how well this version of Keynesian economics has explained the data during different periods and at various points in the business cycle.

The smaller the year-year changes in the shifting propensity to spend out of the previous year's income in column (7), the more confident one can be that the marginal propensity to consume is in the vicinity of .5 and that our fiscal policy multipliers are in the right ball park.

The large differences in the shifting propensity to spend that emerged in the late 1960s and continued to occur through most of the 1980s, help to explain why Keynesian economics has fallen out of favor, in some quarters, and is now confronted with a number of competing schools of thought (Snowdon, Vane and Wynarczyk 1994).

During the 1990s, however, the year-year differences in column (7) have been less on the average than was formerly the case. The implication is that our multiplier equation now fits the data better than was the case in the 1960s, when Keynesian economics was the most highly featured school of macroeconomic thought.

In a world where inflation is no longer as serious a problem as was the case in the late 1960s through portions of the 1980s it can be instructive to use the data in Table 8.1 to better assess the success and failure of the fiscal policy implications of Keynesian economics.

The data in column (2) on the real value of the federal deficit in the national income and product accounts help to dramatize the importance of automatic "built in" stabilizers in helping to offset the deleterious effect of recessionary declines in gross private domestic investment on real GDP. From 1963-93 the real value of the federal deficit more than doubled from its preceding cyclical low to the first year of recovery from an economic recession.

The peaks and troughs in economic activity compiled by the National Bureau of Economic Research (NBER) suggest that the emergence of numerous automatic stabilizers since the great depression of the 1930s may have shortened the duration of economic recessions. From June 1857 to June 1938 the United States experienced 21 economic recessions with an average duration of 21.2 months. In the post World War II period no recession has persisted that long. The average duration for the ten recessions from 1945-95 is only 10.4 months.

The longest recession recorded by NBER lasted 65 months from October 1873 to March 1879. The next longest persisted for 43 months from August 1929 to March 1933. In the post-Keynesian era, the two longest recessions were the 16 month contractions from November 1973 to March 1975 and from July 1981 to November 1982.

The sad thing about this success story is that it has depended a lot more on an expansion of consumption type transfer payments rather than a contra cyclical expansion of government investment in infrastructure which can improve economic efficiency and yield long term benefits. One way to begin the task of validating this point is to examine changes in the inflation adjusted value of government consumption and investment in column (3) of Table 8.1.

During the recessionary trough year of 1961 there was a thirty billion dollar increase in total government consumption and investment and only a lower surplus for the federal government in the national income and product accounts.

From the recessionary peak in business activity in 1990 to the anemic recovery year of 1992 there was a 133.6 billion dollar increase in the real value of the federal deficit and only a 25.2 billion dollar increase in total government consumption and investment.

Since the mild recession of 1960-61 government investment in structures has declined more often than not during economic recessions. In Albany, New York, Washington, D.C. and many other capital cities in the USA construction budgets are routinely robbed or short changed during economic recessions to hide deficits. This is a very inefficient way to cope with the problem of recurring recessions.

Equation (4) implies a first year multiplier of only one for an increase in the deficit that is associated with a cut in taxes. This can be compared to a multiplier of two for an increase in government purchases that is not offset by an increase in taxes.

The problem with a tax cut is that businesses and consumers are left with a choice as to whether they use the cut to bolster their savings (and help finance the deficit) or to increase their consumption of goods and services. An increase in government purchases, other things equal, ensures an extra round of expenditure that will help to get the economy moving again. If the increase in expenditure is used to enhance the nation's infrastructure it can be expected to yield even more benefits with the passage of time.

The spectacular deficit for 1992 and the slow decline in its size in the wake of persistent cut backs in the real value of federal spending for goods and services since 1992 inspired Chairman Alan Greenspan and economists associated with the Federal Reserve Bank of Kansas City to host a major symposium on Budget Deficits and Debt: Issues and Options at Jackson Hole, Wyoming from August 31-September 2, 1995. In his summary of the symposium's conclusions Stuart Weiner noted:

"Chronic government budget deficits and escalating government debt have become major concerns in both developed and developing countries. Concern arises because fiscal imbalances siphon funds from private sector investment, retarding growth and ultimately reducing standards of living. Fiscal imbalances also create potentially large burdens on future generations, as workers may be forced to finance unfunded social programs for rapidly expanding elderly populations. And, fiscal imbalances can trigger disruptive movements in interest rates and exchange rates, as highly indebted countries become increasingly vulnerable to global market forces. Few economic issues have such far-ranging implications as excessive deficits and debt."

Stanley Fischer (1994) examined data for a sample of 94 countries between 1962 and 1988 and found that large budget deficits are negatively associated with economic growth.

Large budget deficits, moreover, provide no assurance that a country's unemployment rate will be low. Belgium, Italy, Canada, Spain and Austria were projected to have more net governmental financial liabilities outstanding as a percentage of nominal GDP during 1996 than the United States and were all suffering from higher unemployment rates according to statistics in the OECD Economic Outlook for June 1996.

Economists in recent decades, however, have tended to favor tax cuts over increases in government purchases when the U.S. economy slips into a recession on the theory that they can be implemented more quickly. There is little evidence to support this conclusion so far. During the 1990-01 recession the U.S. Congress was inundated with all sorts of tax cut proposals but none of them were implemented.

It now seems clear that the tax cuts and rebates enacted during the early stage of the second Bush administrtion were also unsuccessful at preventing another recession.

One way to facilitate a contra cyclical investment policy for public infrastructure would be to develop aid formulas for roads, sewage plants, and other structures that are partly funded by the federal government which are more contra cyclical in character. In the 1947-2001 period the civilian unemployment rate increased from one to 16 months in advance of the ten economic recessions which have been identified by the National Bureau of Economic Research. By letting federal support for public infrastructure automatically increase in response to rising unemployment it might be possible to prevent some recessions and moderate those that do occur.

If Keynes were alive today I am sure that he would also be intrigued by proposals to allow first time home, and perhaps new car buyers, to make penalty free withdrawals from their IRAs. If these withdrawals were limited to periods of economic weakness when real GDP is increasing at a below average rate, or when unemployment is excessively high, it would help to stabilize the U.S. economy and might even reduce the federal budget deficit since most first time home buyers would have to borrow far more than they withdraw from an IRA to finance the construction of a new home.

Fiscal Policy: The Possibility of Indirect Effects

Does a government expenditure multiplier in the vicinity of two or less mean that Robert Lucas was correct in suggesting in 1982 that Keynesian economics is dead. I don't think so. There can be crowding in as well as crowding out. His policy ineffectiveness proposition should be considered a special case that may or may not be true depending on the circumstances. If a tax cut, or a timely increase in government expenditure, keeps private investment from falling further during an economic recession, that will help to moderate the recession and might very well imply an indirect effect upon the economy that is greater than the direct effect.

The most notable difference between the great depression of the 1930s and recent recessions is the extent to which an increase in the government deficit was able to offset the adverse effect of a decrease in investment spending. During the 1929-32 recession the increase in the budget deficit only offset 9.6 percent of the assumed direct and indirect effect of the slump in investment (Renshaw 1990). In the post 1947 period increases in the government deficit have typically offset from 21 to more than 80 percent of the decline in investment times its assumed multiplier of two.

When the indirect effect of a slump in investment spending is largely offset by increases in the other variables in equation (4), a steep decline in investment spending will apparently reverse itself in a year or so as excess inventory is liquidated and worn out equipment breaks down and needs to be replaced.

Who Leads the Economy Into a Recession? Consumers, More Often Than Not.

While the Keynesian multiplier model focuses one's attention on fluctuations in gross private domestic investment as a primary cause of recessions it is by no means clear that the blame for economic recessions should be placed on business enterprises. The Commerce Department's revised data for real GDP and its components indicate that consumer investment in automobiles and other durable goods peaked out before producer's investment in new equipment during five of the seven business expansions from 1958-2001.

It should also be noted that the shifting propensity to spend out of the previous year's income in column (6) of Table 8.1 has declined during economic recessions. At least part of this unexplained collapse in spending should probably be attributed to consumers.

Some macroeconomic principles textbooks contain a page or two illustrating the Keynesian notion that "by trying to save more consumers may end up saving less". Validating the "paradox of thrift" in its strongest form, however, was so difficult in the early post World War II period that the basic idea is seldom mentioned in more advanced textbooks concerned with macroeconomics.

Unemployment insurance and other transfer payment programs to aid families with workers that have lost their jobs, however, have made it more feasible for consumers in general to actually increase their savings during an economic recession by purchasing government securities (directly or indirectly) and helping to finance an automatic increase in the government deficit.

While the peak in the Commerce Department's personal saving rate usually comes fairly early in a business expansion, the data in Table 8.2 would suggest that there may be some truth to the idea that the saving behavior of consumers can precipitate a recession. During the six business expansions from 1958-90 the personal saving rate consistently increased during the quarter containing a peak in economic activity.

The four cases from 1948-2000 with an increase of .7 percentage points or more in the saving rate during peak quarters have longer duration records than those cases with no increase or an increase of less than .7 percentage points.

It used to be the case that consumers could be counted on to lower their saving rate during an economic recession and in so doing help to moderate and shorten the recession. During four of the last five completed recessions, however, the personal saving rate was also higher during the trough quarter for real GDP than before the quarter containing a peak in economic activity.

References

Bryant, Ralph C., Dale W. Henderson, Gerald Holthham, Peter Hooper and Steven A. Symansky (1988). Empirical Macroeconomics for Interdependent Economies(Washington: The Brookings Institution), 63-72.

Case, Karl and Ray Fair (1989). Principles of Economics(New Jersey: Prentice Hall), 831-32.

Campbell, John and N. Gregory Mankiw (1991). "The Response of Consumption to Income," European Economic Review.

Duesenberry, James (1949). Income, Saving and the Theory of Consumption(Cambridge, Mass.: Harvard University Press).

-----, (1965). The Brookings Quarterly Econometric Model of the US(Rand McNally), Chapter 7.

Evans, Michael (1969). Macroeconomic Activity(Harper & Row), Chapter 3.

Fair, Ray (1984). Specification, Estimation, and Analysis of Macroeconometric Models(Harvard University Press).

Federal Reserve Bank of Kansas City (1995). Budget Deficits and Debt: Issues and Options.

Fischer, Stanley (1994). "The Role of Macroeconomic Factors in Growth," NBER Working Paper No. 4565.

Friedman, Milton (1957). A Theory of the Consumption Function(Princeton: Princeton University Press).

Gallagher, G. and Associates (1989). GRE Economics Test(Piscataway, New Jersey: Research and Education Association).

Lucas, Robert. (1982). "The Death of Keynes," in Viewpoints on Supply Side Economics, ed. Thomas Hailstones (Richmond, Virginia: Robert Dane), 3.

Modigliani, Franco (1947). "Fluctuations in the Savings-Income Ratio,"Studies in Income and Wealth(New York: National Bureau of Economic Research), 371-441.

Renshaw, Edward (1967). "The Future Income Hypothesis," The Southern Economic Journal, 34(July), 40-52.

-----, (1990). "A Keynesian View of the US Budget and Trade Deficits,"Public Finance, 45(No. 3), 440-48.

Samuelson, Paul (1941). "A Statistical Analysis of the Consumption Function," in the appendix to chapter 11 of A. H. Hansen, Fiscal Policy and Business Cycles.

Snowdon, Brian, Howard Vane and Peter Wynarczyk (1994). A Modern Guide to Macroeconomics(Brookfield Vermont: Edward Elgar Publishing Company).


Table 8.1

The Shifting Propensity to Spend Out of Previous GDP When the Government Expenditure Multiplier Is Assumed to be Equal to Two and the Basic Data Are In Billions of Chained 1996 Dollars, 1960-2000.


       Gross                                                Change    Following
      Private    Federal  Gov.                   Shifting   Shifting   Return
      Domestic   Gov.     Cons.     Net   Actual Propensity Propensity S&P
Year Investment  Deficit  Invest. Exports  GDP   to Spend   to Spend   Index

         (1)      (2)      (3)      (4)     (5)     (6)n       (7)      (8)

1960   272.8(-) -51.0L   659.5    -21.2   2357.2    53.2       ---P     26.6
1961T  271.0(-) -30.1*   691.3    -19.1   2412.1    52.1     - 1.1**T  - 8.8
1962   305.3    -33.1    732.9    -26.5   2557.6    52.8        .7      22.5
1963   325.7    -47.0    750.2    -22.7   2668.2    52.2     -  .6      16.3
1964   352.6    -30.6    764.8    -15.9   2822.7    52.4        .2      12.3
1965   402.0    -40.5    788.6    -27.3   3002.8    52.4        .0     -10.0
1966   437.3    -42.2    859.3    -40.9   3199.5    51.6     -  .8      23.7
1967   417.2(-)   5.1H   924.1    -50.2   3279.5    48.9     - 2.7**    10.8 
1968   441.3    -23.4    953.4    -67.2   3435.6    51.5       2.6**   - 8.3
1969   466.9    -63.3L   950.0    -71.2   3543.2    52.2        .7P      3.5

1970T  436.2(-)  24.9*   928.6    -65.0   3549.4    50.5     - 1.7T**   14.1 
1971   485.8     66.2H   909.7    -76.1   3660.2    50.4     -  .1      18.7
1972   543.0     21.4    909.8    -89.6   3854.2    52.6       2.2**   -14.5
1973   606.5    -13.2L   902.6    -64.3   4073.1    52.8        .2P    -26.0
1974   561.7(-)  12.4    921.3    -37.6   4061.7    50.1     - 2.7**    36.9
1975T  462.2(-) 165.7H*  939.3     -9.2   4050.3    50.0     -  .1T     23.6
1976   555.5    106.8    938.6    -43.0   4262.6    53.1       3.1**   - 7.2
1977   639.4     70.4    947.4    -68.1   4455.7    52.2     -  .9       6.4
1978   713.0     16.8    977.6    -69.2   4709.9    52.9        .7      18.4
1979   735.4    - 3.0L   997.6    -48.0   4870.1    52.1     -  .8      31.5

1980T  655.3(-)  78.4H* 1018.6      8.0   4872.3    50.4     - 1.7PT** - 4.8
1981   715.6     73.7L  1027.9      3.3   4993.9    50.4        .0P     20.4
1982T  615.2(-) 202.8*  1044.5    -16.0   4900.3    48.8     - 1.6T**   22.3
1983   673.7    244.2H  1078.9    -65.5   5105.6    51.0       2.2**     6.0
1984   871.5    201.0   1116.3   -130.3   5477.4    49.9     - 1.1**    31.1
1985   863.4(-) 209.0   1188.4   -150.9   5689.8    49.6     -  .3      18.5
1986   857.7(-) 226.4   1253.2   -166.9   5885.7    50.2        .6       5.7
1987   879.3    174.3   1290.9   -157.6   6092.6    51.4       1.2**    16.3
1988   902.8    151.4   1306.1   -113.5   6349.1    52.5       1.1**    31.2
1989   936.5    132.5L  1341.8    -81.2   6568.7    52.0     -  .5     - 3.1

1990   907.3(-) 196.2   1385.5    -58.6   6683.5    50.9     - 1.1P**   30.0
1991T  829.5(-) 248.6*  1402.8    -16.4   6669.2    50.5     -  .4T      7.4
1992   899.8    329.8H  1410.7    -18.7   6891.1    50.5        .0       9.9
1993   977.9    289.5   1398.1    -59.9   7054.1    50.4     -  .1       1.3
1994  1107.0    211.9   1399.4    -87.6   7337.8    51.0        .6      37.1
1995  1140.6    180.0   1405.9    -79.2   7537.1    51.1        .1      22.7
1996  1242.7    136.3   1421.9    -89.0   7813.2    51.2        .1      33.1
1997  1393.3     52.3   1455.4   -113.3   8159.5    50.9     -  .3      28.3
1998  1558.0    -42.4   1483.3   -221.1   8508.9    51.1        .2      20.9
1999  1660.5   -106.9   1540.6   -320.5   8859.0    52.0        .9     - 9.0
2000  1762.9   -193.6L  1582.5   -398.8   9191.4    52.8        .8     -11.8
2001  1574.6    -65.8   1640.4   -415.9   9214.5    53.4        .6P       ?    
                                                                               
                                                  

Footnotes for Table 8.1.

(6)n. The shifting propensity to spend out of the previous year's income is equal to column (5) minus two times column (1) minus the sum of columns (2), (3) and (4) when this Keynesian type residual is expressed as a percent of chain weighted real GDP in the preceding year.

(-) The minus signs in parentheses following gross private domestic investment in column (1) identify years when there was a decline in real investment. These years are mostly associated with economic recessions. The only exceptions are 1967, 1985 and 1986 when large increases in government consumption and investment in column (3) more than offset the reduction in real investment.

*The federal deficit in the national income and product accounts deflated by the chain-type price index for gross domestic product in those years containing a recessionary trough in economic activity. Built-in stabilizers have resulted in substantial increases in the real value of the federal deficit in trough years.

**Identifies years when the absolute first difference in the shifting propensity to spend in column (6)n was equal to 1.0 percentage points or more. Only one of the shifts in the 1990s have been that large.

L identifies the low value for the business cycle.

H identifies the high value for the business cycle.

P identifies a year containing a recessionary peak in economic activity.

T identifies a year containing a recessionary trough in economic activity.

Source of basic data: Economic Report of the President.


Table 8.2

The Cyclical Behavior of the Personal Saving Rate

                                                                     

         Date of                     Personal Saving Rate at          Duration 
------------------------       ------------------------------------   of NBER
Saving    Real    Real         Saving    Quarter     Real    Real     Recession
Rate      GDP     GDP          Rate      Before      GDP     GDP      in Months
Peak      Peak    Trough       Peak      GDP Peak    Peak    Trough
                                 (1)       (2)       (3)n     (4)     (5)

46-1      48-4    49-2          12.8       8.6       8.1      5.0      11

51-2      53-2    54-1           9.7       7.8       8.5(+)   8.5      10#

57-2      57-3    58-1           8.7       8.7       8.5      8.3       8

58-4      60-1    60-4           8.9       7.4       7.6(+)   7.1      10

67-1      69-3    70-4           9.6       7.2       8.5(+)   9.9*     11#

73-4      73-4    75-1          11.7      10.5      11.7(+)   9.9      16#

75-2      80-1    80-3          12.5       9.2       9.8(+)  10.0*      6

81-4      81-3    82-3          11.9       9.8      11.3(+)  11.1*     16#

84-3      90-2    91-1          11.3       7.8       7.9(+)   8.3*      8

92-2      00-4    01-3           8.9       2.9       2.9      4.0*      ?

(3)n. The (+) symbols identify cases of acceleration in the personal saving rate during quarters containing a recessionary peak in real GDP.

*Identifies cases where the personal saving rate was higher during the trough quarter than before the quarter containing a peak in real GDP.

#The hatch marks identify durations associated with an increase in the personal saving rate amounting to .7 percentage points or more in the peak quarter. These recessions have been longer, on the average, than those cases with no increase or a smaller increase.

Source of basic data: The Survey of Current Business.


Essay 9:

The Political Business Cycle,
and
The Fed's War on Inflation

Edward Renshaw
Professor of Economics
State University of New York at Albany

Section 2A of the Federal Reserve Act as amended, requires the Board of Governors "to promote effectively the goals of maximum employment, stable prices and moderate long-term interest rates". Moreover, in carrying out monetary policy the Fed is to take "account of past and prospective developments in employment, unemployment, production, investment, real income, productivity, international trade and payments and prices".

Robert Black, President of the Federal Reserve Bank of Richmond, criticized this mandate in 1984 as being too sweeping and unrealistically general, particularly if viewed as a set of objectives to be achieved in the short run. He believes that price stability is the only sensible objective for the Fed to try to achieve with monetary policy.

In 1989 Representative Stephen Neal sponsored a Congressional bill which would have required the Fed to pursue policies aimed at eliminating inflation within five years. The economic recession of 1990-91 had the effect of moth balling that idea.

In 1995, however, Senator Connie Mack of Florida introduced legislation known formally as the Economic Growth and Price Stability Act of 1995 which would make price stability the primary goal of monetary policy. This bill would require the Fed to establish an explicit numerical definition of price stability and specify a time needed to achieve stability taking into account any output and employment effects associated with the move to stability.

The Mack bill cites several reasons for making price stability the sole objective of monetary policy. It argues that inflation distorts the price mechanism, interrupting efficient allocation of resources and thereby diminishing growth and living standards. Having a single objective would reduce uncertainty about monetary policy and perhaps reduce some of the uncertainty surrounding a firm's investment.

Finally, the Mack legislation argues that making price stability the primary objective of the Federal Reserve would lead to a more stable economy. The bill's authors are of the opinion that attempts of monetary policy to stabilize the economy in the short run are often destabilizing since policy changes affect the economy with long, variable, and highly unpredictable lags.

While most of the Governors of the Federal Reserve would probably support the enactment of a Mack type bill it doesn't seem likely to happen in the near future. The U.S. electorate has a long history of being less forgiving with regard to an increase in the unemployment rate than with regard to a moderate increase in the consumer price index.

Presidential Elections and The Possibility of a Political Business Cycle

Economics, if it is a science, is the science of tradeoffs. No one has expressed this idea more colorfully than Daniel Seligman (1991): "Rule No. 1 in thinking about environmental problems is that there are no 'solutions,' only tradeoffs. We want a maximum of economic growth and a minimum of environmental damage."

The recession of 1990-91 reduced the CPI inflation rate to the lowest level in a presidential election year in over two decades and may have told us something about the nature of the electorate's willingness to sacrifice economic growth for a lower inflation rate. Is there a real tradeoff between inflation and the unemployment associated with recent recessions or are they separate problems in the minds of today's electorate?

In 1814 Broughman noted that, "A government is not supported a hundredth part so much by the constant, uniform, quiet prosperity of the country as by those damned spurts which Pitt used to have just in the nick of time." (Cited by Tufte in 1978, p. 3)

The notion that economic conditions influence voters so disturbed Nobel Laureate George Stigler in (1973) that he engaged in some econometrics to reaffirm Kramer's finding "on the electoral unimportance of ordinary fluctuations in unemployment." He then presented an argument based on rational voter behavior for "the unimportance of general economic conditions in national elections."

A more careful analysis of the historical data by Fair in (1978 and an update in 1996), however, concluded that, "Economic events as measured by the change in real economic activity in the year of the election do appear to have an important effect on votes for president. It does not matter much whether this change is measured by the growth rate of real per capita GNP or by the change in the unemployment rate, although the former gives slightly better results."

The belief that the US electorate does hold incumbent political parties accountable for their economic performance encouraged Nordhaus (1975) and MacRae (1977) to articulated theories of a political business cycle. Central to their theories is a preference function or iso-vote loss curve which reflects the willingness of the electorate to endure extra unemployment for the sake of a lower inflation rate. The authors assume that this tradeoff is predictable and can be approximated by a parabola or half circles. They also assume that the electorate is myopic and that optimum partisan policy will produce a condition of high unemployment and deflation in the early years after an election which will be followed by an inflationary boom as the next election approaches.

Keller and May (1984) have compiled some evidence which suggests that President Nixon understood the theory of the political business cycle and may have been successful at influencing monetary and fiscal policy to improve his reelection chances. Other studies, however, do not provide convincing evidence that governments have systematically manipulated economic policy in an effort to influence the outcome of presidential elections. It remains to be seen whether this will continue to be the case in the future. Bush's suggestion in 2000 that the U.S. economy might be on the verge of a recession may have helped him get elected and revived interest in the notion of a political business cycle.

When parabolas representing a non "turnover contour" curve are fit to the election year inflation and unemployment rates which encompass all those cases where the incumbent political party did not lose the presidential election, the conclusion is that the tradeoff between inflation and unemployment is not very stable. For the elections from 1920-88 Trahan and I in (1991) obtained a parabola which implies a preferred inflation rate of 1.66 percent and a maximum endurable unemployment rate of about 17.6 percent. When the parabola was fit to data for the post World War II period, however, the preferred inflation rate was 2.84 percent and the maximum endurable unemployment rate was only 9.5 percent.

Richards (1993) has examined the percentage of respondents answering "yes" to the Gallup Poll question, "Do you approve of the way President--_____is handling his job?" Using quarterly data on this variable for the period 1961:1 through 1992:3 Richards has concluded that voters consider an inflation rate equal to about three percent to be optimal in the long-run.

One of the problems with these and most other models pertaining to the possibility of a political business cycle is that they didn't do a very good job of predicting the outcome of the presidential elections in 1992 and 2000.

The recessionary indicator with the best track record for predicting the outcome of presidential elections (since Woodrow Wilson was elected in 1912 with less than 42 percent of the popular vote) is changes in the unemployment rate. Hoover in 1928 represented the last political party to get reelected after an increase in the civilian unemployment rate during the last two years of a presidential term. It should be noted that the average unemployment rate was only 4.4 percent in 1928 and had been declining since shortly after the recessionary trough of November 1927. In 1929 the unemployment rate only averaged 3.2 percent in spite of an August peak in business activity.

All of the other presidential aspirants in the 88 year period from 1912-2000 have been defeated if the unemployment rate wasn't reduced in the last two years before the election. (See the changes in the unemployment rate in parentheses identified with an asterisk in column 1 of Table 9.1.)

Since 1964 there have been two cases (1968 and 2000) when the incumbent political party lost the election when the decline in the unemployment rate in the last two years was less than - .5 percentage points. This would suggest that the U.S. electorate is somewhat myopic and must be reassured that the economy is getting better before reelecting the incumbent party.

The data in Table 9.1 are not inconsistent with the idea that the U.S. electorate dislikes inflation, however. Since President Wilson kept the U.S. out of World War I in 1916 no political part has been able to get reelected when the annual inflation rate for the CPI in the election year was in excess of 4.6 percent. Harding in 1920, Nixon in 1968, Carter in 1976 and Reagan in 1980 were all successful at ousting the incumbent party from the White House when the CPI inflation rates were in the 4.7 to 15.8 percent range.

Some caution should be exercised in interpreting this finding, however. Three of the four cases of high inflation in column (2) of Table 9.1 are associated with recessionary increases in unemployment in the last two years of the presidential term. The only recent case of a high inflation rate that was not associated with a recession is 1968, when George Wallace and his American Independent party split the Democratic vote and allowed Richard Nixon to defeat Hubert Humphrey with only 43.4 percent of the popular vote. Voters in many other countries have learned to live with inflation rates that are greater than 4.7 percent. One wonders, however, whether higher oil and natural gas prices may have helped Bush defeat Gore in the year 2000.

The more important risk that an incumbent political party faces in failing to control inflation is the possibility that the Fed, which dislikes inflation and has a long history of "leaning against the wind", so-to-speak, will tighten credit to the point of tipping the economy into an "ill timed" recession. In the post 1947 period there has only been one year (1977) when the Fed allowed the growth of M1 in constant dollars to increase when the CPI inflation rate was over five percent.

Since the reelection of President McKinley in 1900 no political party has been able to retain control of the White House if the economy was experiencing a recessionary decline in economic activity at the time of the election. Harding in 1920, Roosevelt in 1932 and Kennedy in 1960 all seem to have benefitted from recessions that did not end until after the election was over.) A very slow growth rate for real GDP in the second half of 2000 and the fear that Bush raised about a near term recession may have helped him defeat the incumbent party by a small margin.

The electorate's distaste for economic recessions appears to have increased with the passage of time. Since the election of Herbert Hoover in 1928, no president has been reelected if there was a recessionary trough in economic activity during the last two years of the presidential term. Carter in 1976, Reagan in 1980 and Clinton in 1992 all benefitted from a reluctance on the part of voters to quickly forget the most recent recession.

Another variable of concern to economists which may help to explain the outcome of presidential elections is the economic growth rate. In this century no political party has been reelected after a four-year increase in civilian employment of less than 3.5 percent. This is the only economic variable, that this analyst has discovered, which can explain the presidential turnover of 1952, when a very popular general from World War II promised to get the U.S. out of an unpopular war.

Is Zero Inflation Worth the Cost?

In constructing their voter preference curves involving a tradeoff between inflation and unemployment Nordhaus and MacRae both assumed that price stability, other things equal, is preferred by the electorate to either inflation or deflation. In 1989 Alan Greenspan, indicated that he and other governors of the Federal Reserve shared this view and were willing to support a bill sponsored by Representative Stephen Neal (D-NC) which would have required the Fed to pursue polices aimed at eliminating inflation within five years.

The attitude at the Fed, however, appears to have changed as a result of the recent slow down in the inflation rate and the weak recovery in employment from the recession which begain in March of 2001.

Seignorage

Seignorage is the revenue that governments receive from inflation when they issue currency and do not pay interest on mandated bank reserves at the Fed. Printing money can be a very rewarding business. If it only costs a dime to make a ten dollar bill the profit to the U.S. Treasury will be 9900 percent. In 1988 the Federal Reserve acquired free and clear, at the stroke of a pen, an additional $16 billion of U.S. government securities in the process of expanding the money supply. If it were not for seignorage governments would have to raise other taxes or borrow more money from the public.

It has been estimated that about two thirds of the more than $516 billion of U.S. currency in circulation is circulating outside the United States. Much of this currency has been "exported" to pay for the "importation" of illegal drugs. And a sizable portion of the currency in use in the United States is also held by people who are either engaged in illegal activities or involved in underground activities that would otherwise go untaxed.

Rao Aiyagari, a research officer at the Federal Reserve Bank of Minneapolis, has suggested that the elimination of inflation might encourage more wasteful transactions that are designed to avoid taxation. He has concluded that the benefits to be expected from a zero inflation rate would be small and might even be negative (1990).

Now that the United States has ceased to be a creditor nation and has become the world's largest debtor nation there may be another good reason for not reducing the inflation rate to zero since it would burden U.S. tax payers and make U.S. companies less competitive in the world market place by having to pay off outstanding bonds with uninflated dollars.

Fairness is another consideration. In a world where there are differential gains in productivity and where unions in the more progressive sectors have a strong preference for obtaining their share of such gains in the form of higher wages there may be a need for some positive inflation, on the average, to fairly distribute higher income and wages to persons employed in those sectors where productivity improvements are harder to achieve. The old fashioned notion of money illusion may also have some validity, when productivity is low and unable to support an accustomed increase in wages.

It should be noted that the CPI is not a perfect measure of inflation. To the extent that consumers respond to individual price increases by switching to comparable items that cost less, it may overstate inflation. The CPI can also misrepresent inflation by not being fully adjusted for improvements in the quality of goods and services. Some studies have suggested that these factors may have caused the CPI to overstate the true inflation rate by as much as two percentage points (Kahn 1994). Hershey (1994), however, has reported that researchers at the Bureau of Labor Statistics (BLS) believe that the CPI only overstates annual inflation by up to six-tenths of a percentage point. The December 1994 issue of The Monthly Labor Review is a good starting point for learning more about the various biases associated with the CPI. Since this overview was published BLS has taken a number of steps to reduce the inflationary bias in the CPI (Garner 1999).

Liquidity Preference and the Duration of Economic Recessions

The two longest recessions in the history of US business cycle analysis--the 65 month recession from October 1873 to March 1879 and the 43 month recession from August 1929 to November 1927--both followed inflation rates that were either zero or negative. The great depression of the 1930s led John Maynard Keynes to invent the notion of a liquidity trap which implies that a zero inflation rate might make it very difficult, if not impossible, for the Fed to cope with an unwanted recession. Low inflation may have also helped to keep the Japanese economy in a recession throughout much of the 1990s.

In seven of the ten U.S. employment recessions from 1945-92 the Fed allowed the yield on new issues of 91 day Treasury bills to fall below the preceding 12 month inflation rate for the CPI for one or more months. It doesn't seem likely that this method of encouraging a revival of consumer spending would be available to policy makers if the inflation rate were zero. Why would consumers ever pay a premium for Treasury bills (which are sold at a discount since they don't pay interest) when they have the alternative of holding demand deposits that are fully insured by the federal government.

While the Fed will no doubt continue to lower short term interest rates in response to recessions, it should be appreciated that a prompt decline in rates may have scary implications and encourage some consumers and producers to wait for even lower rates before reversing their purchasing and production plans. The three shortest recessions dated by NBER in the post 1947 period are all associated with increases in the Fed funds rates in the two months after the peak. See Table 9.4.

The Tinbergen Principle of Public Policy Analysis

In 1969 Jan Tinbergen of the Netherlands was awarded a Nobel prize for his work in econometrics and public policy analysis. In the absence of wonder drugs which are effective against more than one disease or problem and very special interrelationships between variables which can sometimes exhibit the wonder drug property, Tinbergen was able to show that there is a need for at least as many instruments of control as there are goals or important dimensions to a problem.

The validity of this conclusion can easily be illustrated in connection with demand and supply growth equations. To achieve the twin objective of a zero or stable inflation rate and a growth rate for real GDP that is equal to the economy's longer run potential growth rate one must have policy instruments that are effective at not only controlling the growth of aggregate demand but also the growth of aggregate supply.

Monetary and fiscal policy can be used in an effort to help control the growth of aggregate demand. It is less clear whether free market economies have any instruments that are very effective or efficient at ameliorating supply shocks. The dearth of acceptable instruments for controlling the growth of aggregate supply in the short run can lead to a situation where the nation's monetary authorities are confronted with conflicting objectives.

An increase in the federal funds rate, for example, might help to slow the growth of aggregate demand and lower the inflation rate but at the expense of rising unemployment and a slower growth of output. If the primary goal is to boost the growth of output, on the other hand, the funds rate should be lowered to accelerate the growth of aggregate demand but that might aggravate the problem of inflation.

One of the more interesting ways to resolve the problem of conflicting objectives is to assume that increases in the inflation rate and declines in the growth rate for real GDP are equally undesirable. Acceptance of the proposition that the goals of stable growth and inflation should be treated in a balanced way, leads one inevitably to the conclusion that the Fed should endeavor to stabilize the growth of nominal GDP, which in turn will be about equal to the growth of real GDP plus the growth of its implicit price deflator.

The Targeting of Nominal GDP

Nobel laureate James Tobin (1980) was an early advocate of nominal GDP targeting. He noted that over short periods of time changes in nominal GDP growth give rise to similar changes in real GDP growth with relatively little impact on the inflation rate.

After the Volcker inspired monetary revolution of 1979 the Board of Governors of the Federal Reserve often favored a policy of leaning against the wind in a manner that might have helped to stabilize the year to year growth rates for nominal GDP by letting short term interest rates rise when the GDP growth rate was accelerating and decline when the growth of nominal GDP was falling.

One of the more encouraging aspects to the data in Table 9.2 is that we seem to be moving in the direction of more stable and less radical changes in the growth rate for nominal GDP. From 1961-92 there were 18 years when the four quarter growth rate for nominal GDP accelerated or decelerated by 2.0 percentage points or more. from 1993 to 2001 there was only been one year (2001) when the change in the 4th quarter to 4th quarter growth rate for nominal GDP was more than 1.9 percentage points. See column (6) of Table 9.2.

The unanswered question is whether the improved stability of the nominal GDP growth rate is the result of structural changes in the U.S. economy or a more concerted effort on the part of the Federal Reserve to stabilize this rate.

Using Monetary Variables to Help Forecast the Growth of Nominal GDP

One development which has changed the character of business forecasting in recent years is the discovery that a consensus prediction obtained by polling a number of different professional forecasters is cheaper and theoretically superior to the majority of forecasts. A median forecast has to be equal to or better than all forecasts if it turns out to be correct. And if it is way off the mark, the median will still be at least equal to or better than half of the forecasts.

In a world where careful evaluation of past forecasts is seldom under-taken, where much of the data is subject to revision, where models and forecasting procedures are also being revised and where better than average forecasts may have simply been the result of dumb luck, rather than a superior forecasting technique, busy decision makers are probably well advised to buy into the consensus rather than incur the expense of maintaining their own forecasting unit or purchasing the reports of many different forecasting organizations which no one has time to read.

If everyone subscribed to the consensus and no one invested any effort in trying to improve the art of forecasting, however, the value of the consensus itself would surely deteriorate and might even end up being less reliable than the sometimes hard to beat assumption that this year's growth or inflation rate will be equal to last year's preliminary four quarter growth rate.

While the flowering of consensus forecasting has reduced the need for economists, it has also opened up some new and relatively simple, yet reasonably sophisticated procedures, for quickly checking-up on the predictions of professional forecasters by polling statistical indicators rather than the people who are in the business of making forecasts.

Since the invention of electronic computers the art of economic forecasting has been heavily influenced by equations which assign a set of fixed weights to different indicators. If the tradeoffs and relationships between the various indicators were well behaved and quite stable this would be an unbeatable approach. In a world where most time series with leading indicator properties are rather volatile and where economic and financial innovation, bad weather, fits of speculative enthusiasm and political disturbances can distort statistical relationships, however, it will sometimes be better to examine pertinent indicators on an individual basis and simply count the number of indicators that seem to be pointing in a particular direction. Disturbances that are peculiar to one or just a few of the relevant indicators will drop by the wayside and not distort the entire forecast.

In some cases there may be a dominantly superior indicator. There are many situations, however, where one is more likely to be right about the future if a majority of the more reliable indicators are pointing in the same direction.

One of the advantages in polling the indicators is that it does allow one to focus on a very specific objective such as the identification of a peak in business activity before it occurs. In pursuing such a narrow objective it may be possible to use some indicators that are too volatile, specialized or otherwise misbehaved to be included in a composite index of leading indicators or a general purpose forecasting model.

Interest rates, for example, are generally classified as being a lagging indicator but show some promise of being a useful predictor of business peaks if they increase at a very rapid rate or rise to too high a level in relation to previous lows.

In a presentation before the House Banking Committee on February 6, 1975, Chairman Burns noted that the Board of Governors and its open market committee do pay close attention to monetary aggregates but do not confine their attention to one particular definition of the money supply, namely demand deposits plus currency outside banks: "The reason is that this concept no longer captures adequately the forms in which liquid balances--or even just transaction balances--are currently held. Financial technology in our country has been changing rapidly. Corporate treasurers have learned how to get along with a minimum of demand deposits, and to achieve the liquidity they need by acquiring interest-earning assets. For the public at large, saving deposits at commercial banks, shares in savings and loan associations, certificates of deposit, Treasury bills and other liquid instruments have become very close substitutes for demand deposits".

In the same year (1975) this analyst examined the three most common definitions of the money supply (at that time) and found that they were all of some value in forecasting nominal GNP. Even better predictions were obtained, however, by selecting a median forecast rather than by narrowing the forecasting process down to the one monetary aggregate with the best forecasting record.

While financial innovations have distorted the growth rates for the more conventional monetary aggregates and led the Board of Governors to downgrade them as reliable indicators of what might happen to the economy it might still be a mistake to ignore their behavior altogether.

The quantity theory of money assumes that the velocity of money will remain fairly constant. If this assumption is correct the growth of nominal GDP will be about equal to the growth rate for the money supply. In Table 9.2 the December-December growth rates for currency, M1 and M2 can be compared to the following year growth rates for nominal GDP in column (6). The average value for the Federal funds rate during December is also included in this table.

In macroeconomic modeling and forecasting it has become standard practice to include the lagged dependent variable in most structural equations. In keeping with this tradition we also show the fourth quarter to fourth quarter growth rate for nominal GDP. The median value for this growth rate and the four monetary variables is then used to explain the following year to year growth rate for nominal GDP in column (6) of Table 9.2. The actual minus the predicted error terms for this type of consensus forecast are shown in parentheses.

The consensus forecasts have a mean absolute error (MAE) of only 1.22 percentage points for the 1979-99 period. This can be contrasted to MAE's in excess of two percent for each of the monetary variables and 1.51 percentage points for the preceding four quarter growth rate for nominal GDP. For the 13 cases where the median forecast is associated with the four monetary variables, the MAE is only .94 percentage points.

Four of the largest forecasting errors are associated with the economic recessions of 1981-82, 1990-91 and 2000-2? In all of these cases our consensus over estimated the growth of GDP. One of the more interesting ways to try to identify this type of bias since 1978 is to examine yearly changes in the S&P composite stock price index which is included in most indexes of leading economic indicators.

There have been six years (including 2002) since 1978, when the S&P declined. See those error terms identified with a question mark in column 6 of Table 9.2. For the five more or less resolved cases we can conclude that our consensus approach always over estimated the actual growth of GDP. It will be interesting to see if this conclusion turns out to be applicable to what happens to GDP during 2003.

Economic forecasters have not had a lot of success at predicting changes in inflation rates. Over the 23 year period from 1978-99 there were only six years (1978, 1982, 1987, 1988, 1999 and 2000) when the President's Council of Economic Advisers' fourth quarter to fourth quarter forecasts of the inflation rates for GNP or GDP turned out to be superior to a no change in the inflation rate assumption. See Table 9.3.

The CEA has a better track record with regard to forecasts of real GDP. The superior forecasting record would suggest that leading indicators are more useful at explaining changes in the real economy than the inflation rate.

In the parentheses associated with the following year growth rates for real GDP in column (7) of Table 9.2 we compute error terms for this variable based on a subtraction of the preceding fourth quarter to fourth quarter growth rate for the implicit price deflator for GDP from the median or consensus forecast for nominal GDP in columns (1) to (5).

The mean absolute error associated with this approach to real GDP forecasting is 1.12 percentage points for the 1979-99 period. The predicted growth rate of 4.5 percent for the year 2000 was so high, however, as to have probably encouraged the Board of Governors to continue a process of raising its funds rate.

References

Aiyagari, Rao (1990). "Deflating the Case for Zero Inflation," Federal Reserve Bank of Minneapolis Quarterly Review, Summer, 2-11.

Alesina, Alberto (1988). "Macroeconomics and Politics," NBER Macroeconomics Annual, 3: 13-52.

Clark, Todd (1994). "Nominal GDP Targeting Rules: Can They Stabilize the Economy," Economic Review, Federal Reserve Bank of Kansas City, 3rd Quarter, 11-25.

Fair, Ray (1978). "The Effect of Economic Events on Votes for President," The Review of Economics and Statistics, 60(May), 159-73.

Fair, Ray (1996). "The Effects of Economic Events on Votes for President, 1992 Update," Political Behavior, June, 119-39.

Fischer, Stanley (1994). "The Role of Macroeconomic Factors in Growth," NBER Working Paper No. 4565.

Garner, Alan (1999). "Progress Toward Price Stability: A 1998 Inflation Report," Federal Reserve Bank of Kansas City Economic Review, First Quarter, (84), 5-20.

Hershey, Robert (1994). "An Inflation Index Is Said to Overstate the Case," The New York Times, January 11, D1 and D15.

Hotelling, Harold (1929). "Stability in Competition," Economic Journal, (39), 41-57.

Kahn, George (1994). "Achieving Price Stability," Federal Reserve Bank of Kansas City Economic Review, First Quarter, 5-6.

Keller, R. and A. May (1984). The Presidential Political Business Cycle of 1972," Journal of Economic History, (44), 265-70.

Kramer, Gerald (1971). "Short-Term Fluctuations in U.S. Voting Behavior, 1896-64," The American Political Science Review, 65(March), 131-43.

MacRae, Duncan (1977). "A Political Model of the Business Cycle," Journal of Political Economy, 85(April), 239-63.

McNees, Stephen (1988). "How Accurate Are Macroeconomic Forecasts?" New England Economic Review, July/August, 19.

Nordhaus, William (1975). "The Political Business Cycle," Review of Economic Studies, 42(January), 169-90.

Renshaw, Edward (1975). "Economic Activity and Alternative Definitions of the Money Supply," Journal of Money, Credit, and Banking, November, 507-19.

Richards, Daniel (1993). "What Inflation Policy Do American Voters Want, and Do They Get It?" New England Economic Review, September-October, 33-44.

Seligman, Daniel (1991). "Cuddling Up to Mother Earth," Fortune, 123(July 29), 177.

Stigler, George (1973). "General Economic Conditions and National Elections," American Economic Review, 63(May), 160-67.

Tinbergen, Jan (1996). Economic Policy: Principles and Design(North Holland).

Tobin, James (1980). "Stabilization Policy Ten Years After," Brookings Papers on Economic Activity, No. 1, 19-71.

Trahan, Emery and Edward Renshaw (1990). "Presidential Elections and the Federal Reserve's Interest Rate Reaction Function," Journal of Policy Modeling, 12(1), 29-34.

-----, (1991). "A Note on the Electorate's Most Preferred Inflation Rate," Public Choice, 70, 95-97.

Tufte, Edward (1978). Political Control of the Economy(Princeton: New Jersey: Princeton University Press).


Table 9.1

Using Economic Variables to Explain the Outcome of Presidential Elections in the US, 1916-96.

                                                                     

Presidential   Civilian       CPI       Change   Civilian   President Elec.
  Election    Unemployment  Inflation   Misery   Employment   and Political
    Year         Rate         Rate       Index   Increase     Affiliation

               Cases Where the Incumbent Political Party Was Re-elected
                  (1)n       (2)n       (3)n       (4)n          (5)
1916         4.8(-3.2)       7.4*        ---        6.1      Wilson      (D)
1924*        5.5(-2.1)        .3       -14.0        7.2      Coolidge    (R)
1928         4.4( 2.5)*    - 1.2       - 2.6        7.3      Hoover      (R)
1936        16.9(-4.8)       1.0         4.5*      15.6      Roosevelt   (D)
1940        14.6(-4.4)        .8       - 2.5        8.0      Roosevelt   (D)
1944         1.2(-3.5)       1.6       -12.6       13.6      Roosevelt   (D)
1948         4.0(- .5)       2.7         3.9*      10.0      Truman      (D)
1956         4.2(- .8)       2.9         3.5*       5.9      Eisenhower  (R)
1964         5.0(- .5)       1.2       - 1.9        5.4      Johnson     (D)
1972         5.2(- .9)       3.4          .5*       8.2      Nixon       (R)
1984         7.2(-3.6)       4.0       - 8.5        5.7      Reagan      (R)
1988         5.3(-1.3)       4.4       - 1.5        9.5      Bush        (R)

1996         5.3(- .1)       3.3        -1.6        6.9      Clinton     (D)

               Cases Where the President Was Elected From an Opposing Party

1920         4.0( 2.6)*     15.8*        7.6*       3.1*     Harding     (R)
1932        23.6(14.9)*    -10.2        10.2*     -15.7*     Roosevelt   (D)
1952         2.7(-1.6)        .9       - 3.1        3.3*     Eisenhower  (R)
1960         6.6(  .4)*      1.5         1.0*       3.1*     Kennedy     (D)
1968*        3.4(- .4)*      4.7*        1.9*       9.5      Nixon       (R)
1976         7.8(  .6)*      4.8*        4.0*       8.0      Carter      (D)
1980         7.3( 1.3)*     12.4*        7.1*      11.9      Reagan      (R)
1992*        7.3( 1.2)*      2.9          .5*       2.3*     Clinton     (D)
2000         4.0(- .3)       3.4       - 1.2        6.7      Bush        (R)

Footnotes for Table 9.1

(1)n. Average rate for the years 1916-44 and the December rate for the years 1948-96. The figures in parentheses are the changes in the unemployment rate during the last two years or 24 months of the presidential term. An increase in the unemployment rate is used to predict an election turnover.

(2)n. Average annual rates 1916-44 and annual rates 1948-92. An inflation rate in excess of 4.5 percent is used to predict an election turnover.

(3)n. The misery index is equal to the sum of the unemployment and inflation rates in columns (1) and (2). The change in the misery index is from one presidential election year to the next. An increase in the index is used to predict an election turnover. The all item CPI has not been extended back to 1912. From 1911-12, however, the Bureau of Labor Statistic's food at home only index increased 6 percent. This would suggest that there may not have been an increase in the misery index from 1912-16.

(4)n. Thousands of employed civilians 14 years and older 1912-48 and employed persons 16 and over 1948-92. A four year percentage increase in civilian employment of less than 3.5 percent is used to predict an election turnover.

*Cases where a presidential election turnover is predicted. The asterisks associated with 1924, 1968 and 1992 represent years of electoral unhappiness when third party candidates were able to corral ten percent or more of the popular vote for president. Since James Buchanan was able to get the Democratic party reelected in 1856 with only 45.3 percent of the popular vote, the incumbent political party has usually lost the election after this strong a showing on the part of third party candidates. The only exception is Calvin Coolidge in 1924 when there were no economic variables pointing in the direction of a presidential turnover. Lincoln in 1860, Cleveland in 1892, Wilson in 1912, Nixon in 1968 and Clinton in 1992 may have all been the beneficiaries of third party candidates who split the incumbent party vote and allowed themselves to win with only 39.8 to 46.1 percent of the popular vote. In 1968 George Wallace, running as a third party candidate split the democratic vote and allowed Richard Nixon to defeat Hubert Humphrey with only 43.4 percent of the popular vote. In 1992 Ross Perot ran as a third party candidate and allowed Bill Clinton to defeat George Bush with only 43 percent of the popular vote. In 1980 an independent candidate by the name of John Anderson was able to capture 6.6 percent of the popular vote. We have ignored this marginal case in assigning asterisks to the outcome of presidential elections on the grounds that Reagan was able to capture 50.7 percent of the popular vote and that there were enough economic variables pointing in the direction of a turnover so that Carter might have lost the election anyway.


Table 9.2

Using Monetary Indicators to Help Forecast Following Year Growth Rates for GDP.


       Annual Growth Rates   December   4 Quarter   Following Year Growth Rate
Year  ---------------------  Fed funds  GDP Growth  --------------------------
      Currency   M1      M2    Rate       Rate      Nominal GDP     Real GDP
        (1)     (2)     (3)     (4)        (5)          (6)           (7)

1978    9.8*    8.0     7.5    10.0       14.5      11.8( 2.0)     3.4(  .6)
1979    9.2*    6.9     7.9    13.8       10.2       8.9(- .3)      .0(- .8)
1980   10.0     6.9     8.5    18.9        9.6*     12.0( 2.4)     2.5( 2.4)
1981    6.2     6.9     9.7    12.4        9.7*      4.1(-5.6)?   -1.9(-3.2)
1982    8.2     8.7*    8.8     9.0        3.5       8.5(- .2)     4.2(  .4)
1983   10.3*    9.8    11.3     9.5       11.3      11.3( 1.0)     7.3(  .6)
1984    6.8     5.8     8.6     8.4*       9.3       7.1(-1.3)     3.9(-1.0)
1985    7.6    12.4     8.0*    8.3        7.1       5.7(-2.3)     3.4(-1.7)
1986    7.6*   16.9     9.5     6.9        5.1       6.5(-1.1)     3.5(-2.0)
1987    9.0     3.5     3.6     6.8*       7.8       7.7(  .9)     4.2(- .5)
1988    7.8     4.9     5.8     8.8        7.5*      7.5(  .0)     3.5(- .3)
1989    4.9      .8     5.5*    8.4        6.3       5.7(  .2)     1.7(- .2)
1990   10.9     4.0     3.8     7.3        4.6*      3.2(-1.4)?   - .2(- .7)
1991    8.3     8.7     3.1     4.4*       4.0       5.6( 1.2)     3.3( 1.7)
1992    9.3    14.3     1.6     2.9        6.4*      5.1(-1.3)     2.4(-1.8)
1993   10.0    10.3     1.6     3.0        5.0*      6.2( 1.2)     4.0( 1.8)

1994   10.0     1.8      .4     5.4*       6.2       4.9(- .5)?    2.7(- .7)
1995    5.1    -2.0     4.1     5.6        4.3*      5.6( 1.3)     3.6( 1.5)
1996    5.9    -4.1     4.8     5.3*       6.0       6.5( 1.2)     4.4(  .8)
1997    7.7    - .7     5.7*    5.5        6.2       5.6(- .1)     4.3(- .2)
1998    8.2     2.2     8.8     4.7        6.0*      5.6(- .5)     4.1(- .3)
1999   12.6     2.5     6.1*    5.3        5.9       5.9(- .2)     3.8(  .0)
2000    4.4    -3.3     6.1     6.4        4.6*      2.6(-2.0)?     .3(-2.3)
2001    9.5     8.3*   10.4     1.8        2.0                ?

Footnote for Table 9.2

*The asterisks identify the median rate in columns (1) to (5).

?The question marks identify error terms when the S&P composite index declined in the preceding year.

The parentheses associated with the following year to year growth rates for nominal GDP in column (6) are the error terms obtained by subtracting the median rate for the variables in columns (1) to (5) from the actual following year growth rate for nominal GDP.

The parentheses associated with the following year to year growth rates for chain weighted real GDP in column (7) are the error terms associated with the median rate for the variables in columns (1) to (5) minus the fourth quarter to fourth quarter growth rate for the implicit price deflator.


Table 9.3

Evaluating the 4th Quarter to 4th Quarter Growth Rate Forecasts of the President's Council of Economic Advisers.

                                                                     

Year    Forecasted     Preliminary      Forecasted         Preliminary   
        Growth Rate    Growth Rate      Growth Rate        Growth Rate
        for Real GNP   for Real GNP     for the Price      for the Price
        or GDP         or GDP           Deflator           Deflator
           (1)             (2)n              (3)                (4)n
1978    4.8(- .5)B      4.3(-1.4)        6.0( 2.3)B          8.3( 2.4)
1979    2.2(-1.4)B       .8(-3.5)        7.4( 1.6)           9.0(  .7)
1980   -1.0(  .7)B     - .3(-1.1)        9.0( 1.0)          10.0( 1.0)
1981    1.8(-1.1)        .7( 1.0)       10.2(-1.6)           8.6(-1.4)
1982    3.0(-4.2)      -1.2(-1.9)        7.2(-2.6)B          4.6(-4.0)
1983    3.1( 3.0)B      6.1( 7.3)        5.6(-1.5)           4.1(- .5)
1984    4.5( 1.1)       5.6(- .5)        5.0(-1.5)           3.5(- .6)
1985    4.0(-1.5)B      2.5(-3.1)        4.3(-1.1)           3.2(- .3)
1986    4.0(-1.8)       2.2(- .3)        3.8(-1.6)           2.2(-1.0)
1987    3.2(  .6)B      3.8( 1.6)        3.6(- .3)B          3.3( 1.1)
1988    2.4(  .2)B      2.6(-1.2)        3.9(  .0)B          3.9(  .6)
1989    3.5(-1.1)       2.4(- .2)        3.7(  .1)           3.8(- .1)
1990    2.6(-2.3)        .3(-2.1)        4.2(- .2)           4.0(  .2)
1991     .9(- .7)        .2(- .1)        4.3(-1.3)           3.0(-1.0)
1992    1.9( 1.0)B      2.9( 2.7)        3.2(- .8)           2.4(- .6)
1993    2.9(- .1)       2.8(- .1)        2.6(- .4)           2.2(- .2)
1994    3.0( 1.0)B      4.0( 1.0)        2.7(- .4)           2.3(  .1)
1995    2.4(- .9)B      1.5(-2.5)        2.9(- .4)           2.5(  .2)
1996    2.2( 1.2)B      3.4( 1.9)        2.8(- .7)           2.1(- .4)
1997    2.0( 1.9)       3.9(  .5)        2.5(- .7)           1.8(- .3)
1998    2.0( 2.1)       4.1(  .2)        2.0(-1.1)            .9(- .9)
1999    2.0( 2.2)       4.2(  .1)        1.9(- .3)B          1.6(  .7)
2000    2.9( 1.2)       4.1(- .1)        1.9(  .5)B          2.4(  .8)
2001    3.2(-2.8)B       .4(-3.7)        2.0(- .2)B          1.8(- .6)
2002    2.7                              1.9
                   Mean Absolute Forecasting Errors 1978-98

             1.35            1.63             1.01                 .84

Footnotes for Table 9.3

The forecasted growth rates for real GNP or GDP and its implicit price deflator are from the Economic Report of the President. The figures in parentheses are the error terms obtained by subtracting the forecasted growth rate from the actual preliminary growth rate in columns (2) and (4) from the next Economic Report. In those cases where a forecast range was presented, a mid point is used to evaluate the CEA's forecast.

(2)n and (4)n. The actual preliminary growth rates in the following year's Economic Report of the President. The figures in parentheses are the first differences in these preliminary growth rates. The evaluations for 1995 and beyond are based on chain-type indexes.

B identifies years when the Council of Economic Adviser's forecasting error turned out to be smaller absolutely than the "no change in the growth rate forecasts" implied by the first differences in the preliminary growth rates.


Table 9.4

Changes in the Fed Funds Rate Two Months after NBER Peaks in Economic Activity and the Duration of Economic Recessions.


NBER Peak   % Point Change   Duration of
  Dates     Fed Funds Rate   Recession
                             in Months

Jan. 1980       3.37            6

Aug. 1957        .26            8

July 1990        .05            8

Nov. 1948        .02*          11

Dec. 1969        .01           11

July 1953      - .22*          10

Nov. 1973      - .38           16

Apr. 1960      - .60           10

Mar. 2001--    -1.10           10?

July 1981      -3.17           16

*The Changes in the Fed funds rate for the 1948 and 1953 peaks are approximated by the change in the discount rate on new issues of 91 day Treasury bills.




Return to Some Featured Highlights

Return to A Tabular Index to the Essays and Tables