Economic indicators are more than “statistics.” They are the factual base for public policies and actions that affect the economic well being of all Americans. It is essential for the vitality of a democracy that these data be impeccably objective, that they be prepared with the highest professional standards, and that they have no hint of political interference. Only with such integrity will the people have confidence in the data.
The indicators are produced mainly by agencies in the executive branch of the U.S. government, such as the Bureau of the Census, Bureau of Labor Statistics, Bureau of Economic Analysis, national Agricultural Statistics Service, and the Internal Revenue Service. From the 1970s to the early 1990s, there have been occasional allegations that the indicators were politicized by “cooking” the preparation of the statistics to make the president who is in office at the time look better. On further examination, these allegations of tampering with the data were shown to be unfounded. Although this is comforting, it still leaves the possibility of future tainting, which must be guarded against. While some data may be more vulnerable to “cooking” than others—for example, the estimation of the gross domestic product is based on more statistical judgments that conceivably could be shaded than is the unemployment rate, which is based on household survey information—the possibility of tampering exists with all data.
One institutional device for insulating statistical agencies from political pressure is for the head of the agency to be appointed by the president and confirmed by the Senate, as is done for the director of the census, commissioner of labor statistics, commissioner of internal revenue, and commissioner of the Social Security (since 1994). This may give the agency heads the appearance, if not the reality, of heightened stature for resisting pressures from their political superiors, which in these cases are the secretary of commerce, secretary of labor, and secretary of the treasury. Senate confirmation probably has greater weight when it is specified for a period of time that does not coincide with the presidential term, as is the case of the four-year term for the commissioner of labor statistics, and a six-year term for the commissioner of Social Security Administration. At the same time, a career civil service employee probably has more independence than a political official. Thus, to the extent that an agency head who is confirmed by the Senate in reality becomes a political official, the “independence effect” of the confirmation is diminished. There obviously are no tidy answers.
A second institutional device that may lessen political interference with the data is the Office of Management and Budget’s directive, which requires statistical agencies of the federal government to limit access to their facilities where the data for major economic indicators are being prepared preceding the day when they are released to the public (referred as a “lockup”). During the lockup, which may last several days, only certain employees of the statistical agency have access to the data. The lockup was originally instituted to prevent leads of unpublished data that give recipients of leaks an unfair advantage in financial markets, but it could have a secondary benefit of fending off attempts to interfere with the data preparation. On the other hand, the argument can be made that is many people know a number ahead of time, a political appointee could not have that number changed without causing a scandal, if, however, only a few people know the number, the political appointee could conceivably use some leverage (for example, the threat of funding cuts) to get them to change the number. Again, there are no tidy answers.
On balance, the above mechanisms of Senate confirmation of statistical agency heads and of limiting access to the data preparation through the lockup procedure lessen the chances that economic indicators can be compromised. But no system is foolproof when there is a determination to violate it. Thus, these and future measures to protect the integrity of data will reduce, but not eliminate, attempts to contaminate the data for political gain. Therefore, it is essential that the press and analysts be vigilant in safeguarding the integrity of the numbers and sound the alarm when they believe that data may be suspect. This is added insurance for maintaining accurate information, which is a bulwark of free society.
Frumkin, Norman (1994). Guide to Economic Indicators, 2nd. Ed., M.E. Sharpe
publishers.
GUIDE TO ECONOMIC INDICATORS
The U.S. economic indicators developed primarily by U.S. government agencies but also
by private organizations. The indicators reflect the overall dimensions of the domestic
and international aspects of the American economy as well as particular segments of it.
This introduction briefly describes how the indicators are used to track the economy and provides background material on using economic indicators, including how to interpret changes in them and how to evaluate their accuracy and presentation
The economy continually operates in recurring phases of rising and falling activity, which are referred to as business cycles. Economic indicators are used to measure overall economic activity for classifying it as rising (expansion) or falling (recession), as well as to determine the cyclical turning points of these expansions and recessions. The actual and technical determination of expansions and recessions is made by a committee of economists under the auspices of the National Bureau of Economic Research (NBER), a private nonprofit economic research organization. The dating of business cycles by this nongovernmental organization is adopted by the U.S. government as the official periods of expansion and recession. It is accepted by a wide range of economists and politicians, regardless of their differing views regarding economic analysis and policy formulation.
The NBER committee does not rely on a specific formula. The committee bases its decision on judgments about the overall direction in which the preponderance of the indicators are moving. For example, a recession is often but not always indicated when the real gross domestic product (GDP in constant dollars, which abstracts from rising or falling prices) declines for two successive quarters. Sometimes several indicators move contrary to the GDP trend. One of the challenges of economic analysis is to assess the import of such variations.
Most analysis of economic indicators, then, is concerned with changes in the indicators over time—the cyclical fluctuations between expansion and growth and recession and decline. While the absolute volume or level of economic activity is of some interest, movements from one period to the next are more important for tracking the economy. The indicators often show change in terms of percent or as index numbers, making relative comparisons over long time periods easier. It is customary to think in terms of one year in assessing economic performance. Over the period of a year, of course, some changes simply reflect natural seasonal variations. Thus, to interpret indicators effectively, an analyst must understand index numbers, annualized the annual movements, and seasonally. In addition, because those indicators that are measured in dollars (such as the gross domestic product or wage earnings) are affected by increases and decreases in prices, the analyst must understand the measurement of indicators in constant dollars that abstract from these price changes. It also is important for the analyst to understand how alternative methods of calculating growth rates and charting economic data graphically lead to different perceptions of economics performance.
Three measures are related to annual trends: the seasonally adjusted annual rate, annual change, and December-to December or fourth-quarter-to fourth quarter change.
The seasonally adjusted annual rate (SAAR) reflects what the yearly movement of the indicator would be if the same rate of change (adjusted for seasonal variations) were to continue for the next eleven months (monthly indicator) or for the next three quarters (quarterly indicator). This figure represents the same rate of change for the current month or quarter compounded over the rest of the year. The SAAR provides a quick view of how a very short-term movement compares with a twelve-month period. It also facilitates comparisons of growth rate of change for comparative purposes only—they are not meant to forecast what is expected to occur.
Annual change figures compare the average level of the indicator in one year with the average level of the next year. These averages are computed from data for the twelve months or four quarters of the indicator and, thus, help to compensate for the effects of unusually high or low activity periods when analyzing short periods during the year.
December-to-December or fourth-quarter-to-fourth quarter change figures focus on economic change from the end of one calendar year to the end of the next calendar year. This kind of data is often used in economic reports at the beginning of the calendar year to predict the coming twelve months. It provides a more current assessment of the most recent and coming twelve-month periods that the above annual change figures. However, any single period during the year may have abnormally high or low rates of economic growth or inflation. Because these data are not averaged over an annual period, they can provide a distorted view of annual change.
There are many factors-such as changes in the weather, holidays, school vacations, yearly automobile model changes, annual days, school vacations, yearly automobile model changes, annual tax returns, and so forth—that cause normal seasonal up and down movements in economic activity during the year. If not taken into account, these fluctuations could distort real trends in the economy. For example, Christmas buying can make the economy look prosperous in December when in fact, Christmas shopping is below average. Or the normal summer shutdown of auto assembly plants for the new model year can make the economy look dormant in the summer when, in fact, fewer plants have shut down than is typical. To prevent seasonal variations from distorting the economic picture, most economic data are seasonally adjusted. Seasonal adjustment attempts to eliminate movements in economic indicators caused by such factors as increased sales in November and December due to Christmas shopping, decreased construction work in winter because of cold weather, and the large number of students looking for work in summer months. the adjustments are based on experience in previous years and capture typical movements that are expected from the average experience. Because they cannot indicate special circumstances in particular time periods, however, aberrations should be watched for in analyzing current trends.
For those indicators that are not seasonally adjusted, comparing trends of several consecutive months with the same months of the previous year is an indirect technique of seasonal adjustment that helps determine when the indicator is actually rising or falling. However, this method cannot identity cyclical turning points on a current basis because it focuses on year-to-year monthly change and does not provide a seasonally adjusted view of current movements in their own right.
It is beyond the scope of this lecture to detail the various statistical techniques of making seasonal adjustments or to analyze whether any particular indicator should be seasonally adjusted. However, the source publications of monthly or quarterly indicators would tell whether, in fact, each indicator is seasonally adjusted or not.
There are many kinds of questions that can be raised about the accuracy of economic indicators. Conceptual issues, such as whether the indicators measure why they purport—for example, if the unemployment rate truly represents the proportion of people out of work or if the consumer price index truly represents inflation to the consumer—are beyond the scope of this book. More practical considerations, such as how closely the underlying data represent the definitions of the indicators, are also incapable of being measured. Such errors clearly do exist because secondary data are used as source information for constructing economic indicators. For example, data based on income tax returns that were originally developed for assessing the economic effects of existing income tax laws and of proposed changes in tax laws, are used for estimating certain components of the gross domestic product and for obtaining information on small firms for the economic censuses. Even though some data sources may not exactly correspond to the definitional concepts of certain indicators, secondary sources are used in constructing economic indicators to hold down the costs of data collection and to limit the reporting burden on the public.
There are two fairly simple ways to evaluate the accuracy of economic indicators, however. The effect of data errors and the relative accuracy of an indicator can be estimated by taking into account the extent of revisions to the preliminary data and, in the case of indicators based on surveys, the sampling reliability of the surveys. Quantitative measures of the effect of these errors have been developed in some cases.
Error due to revisions reflects changes in the figures from when they are initially provided to the later, more accurate information. The size of revision error is based on the past experience of these changes. For example, two-thirds of the revisions between the advance and the latest estimates of real gross domestic product have been within a range of –1.3 to 2.2 percentage points. Thus, based on a 67 percent confidence, it is likely that an advance quarterly estimate of real GDP growth at an annual rate of 2.0 percent will be revised within a range of 0.7 to 4.2 percent. Raising the confidence level to 90 percent increases the likely revision to a range of –0.6 to 5.5.
Error due to sampling results from the likelihood that data obtained from a sample of a population differ from what they would be if the entire population were surveyed. Estimates of sampling error are developed from mathematical formulas of probability, and there is a predetermined direct relationship between error size and its chances of occurring. For example, the sampling error for housing starts based on a 67 percent confidence is plus or minus 3 percent. Thus at a confidence of 67 percent, it is likely that a monthly figure of housing starts at an annual rate of 1.5 million units ranges with 1.455 and 1.545 million if all starts were surveyed. Raising the confidence level of 95 percent doubles that sample error to a range of 1.41 to 1.59 million.
When such estimates are available, it is important to take them into account. However, whether estimates of error are available or not, it is clear in all cases that any single number provided by an indicator cannot always exactly represent reality. Because of the various sources of error inherent in economic data, in general, an indicator should be considered as representing a range rather than an actual figure. Analysis of the specific or related data, as well as estimates of the size of revisions or sampling errors available, can suggest whether actual measurements fall closer to upper or lower bounds of that range.
Economic indicators are developed from data gather in surveys of households, businesses, and governments, and from tax and regulatory reports submitted to the federal and state governments. The indicators are available weekly, monthly, quarterly, or annually, depending on the data series. Because policy makers in the administration, Congress, and Federal Reserve Board want the indicators as soon as possible following the month or quarter to which they refer, the figures are initially provided on a preliminary basis and are revised in subsequent months as more complete and accurate data are obtained. Revisions are sometimes substantial, and it is important that preliminary information be treated as tentative. The use of preliminary and revised information results from the tension between the need for both timely and accurate data. In analyzing current information, it is desirable to view the most recent data in the context of previous trends and to wait for the more accurate revised data to determine whether there has been a continuation or change from the trend.
In addition to revisions that are made on a current basis, more comprehensive revisions are made annually, every five years, or as “benchmark” revisions. For particular indicators, they result in a revision of all historical data as in the case of the gross domestic product; in other cases, application of the new definitions and data-estimating methodologies is limited to future figures of the indicators, as for the consumer price index. The decision about whether to revise historical data is based on consideration of several factors—the need to have a consistent series over time balanced
Against the lack or weakness of comparable data for earlier time periods, the theoretical question of whether to “rewrite history” by including factors that previously were not considered in economic analysis and policy making, and the additional costs for statistical programs to make the more extensive revisions. When the historical data are not revised, there is a break in the series where the previous data are not fully consistent with the new data. The inconsistencies in data should be recognized when analyzing long-term trends.
The benchmark in more accurate because it is based on data obtained from a larger sample of survey respondents including, in some cases, the universe of the whole population, such as the five-year economic censuses and the ten-year census of population. It is also more accurate because there is more time to check the validity of the reported survey data. Thus, the benchmark provides more complete and precise information for checking the accuracy of the indicator at particular points in time and for revising historical data and estimating current data.
There are over one hundred U.S. Economic Indicators measuring the performance of the American economy. Here is a topical grouping of the indicators:
ECONOMIC GROWTH
Gross Domestic Product
CONSUMER SPENDING
Retail Sales
Personal Income and Saving
Consumer Attitude Indexes
INVESTMENT
Plant and Equipment Expenditures
Corporate Profits
Business Optimism Indexes
LABOR
Unemployment
Employment
Help-Wanted Advertising Index
Average Weekly Hours
Average Weekly Earnings
Productivity
Unit Labor Costs
Employment Cost Index
Collective Bargaining Settlements
PRICES
Inflation
GDP Price Measures
Consumer Price Indexes
Producer Price Indexes
Import and Export Price Indexes
CRB Futures Price Index
PRODUCTION
Industrial Production Index
Capacity Utilization
Manufacturers’ Orders
Inventory-Sales Ratios
HOUSING
Housing Starts
Home Sales
FINANCE
Money Supply
Flow of Funds
Bank Loans: Commercial and Industrial
Consumer Installment Credit
Interest Rates
Stock Market Price Indexes and Dividend Yields
GOVERNMENT
Government Budgets and Debt
INTERNATIONAL
Balance of Trade: U.S. Merchandise Exports and Imports
Balance of Payments: U.S. International Economic Transactions
International Investment Position of the United States
Value of the Dollar
CYCLICAL INDICATORS
Growth Cycles
Purchasing Managers’ Index
Leading, Coincident, and Lagging Indexes
LEADING INDICATORS:
Average weekly hours, manufacturing
Average weekly initial unemployment claims
Manufacturers’ new orders, consumer goods and materials
Vendor performance
Contracts and Orders, plant and equipment
Building permits, new private housing units
Change in manufacturers’ unfilled orders
Change in sensitive materials prices
Index of stock prices
Money supply
Consumer sentiment
COINCIDENT INDICATORS
Employees on nonagricultural payrolls
Personal income less transfer payments
Index of industrial production
Manufacturing and trade sales
LAGGING INDICATORS:
Average duration of unemployment
Ratio of manufacturing and trade inventories to sales
Change in index of unit labor cost, manufacturing
Average prime rate charged by banks
Commercial and industrial loans
Ratio of consumer installment and credit to personal income
Change in consumer price index for services
ECONOMIC WELL-BEING
Distribution of Income
Poverty
Business Firm Formation and Growth
Business Failures
Farm parity Ratio
Most of the indicators in this book are produced and published by federal government agencies. In the dissemination of economic statistics, there is a distinction between primary and secondary data. The term “primary data” refers to economic indicators published in journals or reports by the organization that produces the figures, while “secondary data” refers to indicators published by organizations other than the producer of the figures. The following monthly journals, which include both primary and secondary data, are the main vehicles for publishing the indicators: the Survey of Current Business of the Bureau of Economics Analysis in the U.S. Department of Commerce; the Monthly Labor Review and Employment and Earnings of the Bureau of Labor Statistics in the U.S. Department of labor; and the Federal Reserve Bulletin of the Federal Reserve Board. A handy secondary publication is the monthly Economic Indicators, which is operated by the U.S. Council of Economic Advisers for the Joint Economic Committee of Congress. The publications are cited as primary or secondary sources of the indicators under the “Where and When Available” category. They are sold by the U.S. Government Printing Office. The appendix tables of the annual Economic Report of the President are the most convenient source for historical data; single copies are available free from the Executive office of the President. An advanced reference book for indicators is Leading Economic Indicators: Forecasting Records by K. Lahiri and G.H. Moore (1991).