Last time we looked at the measurement of GDP and inflation, and at the end of class I noted that there is some concern about the accuracy of our inflation measures (the CPI and the GDP deflator).
There are three principal sources of bias in our measures of price changes that have been identified: new goods bias, quality change bias, and substitution bias. In each case, these biases become increasingly important as the length of the period being compared increases (i.e., comparing this year vs. last year, these biases are small; but if we're comparing 1996 to 1976 or 1966, they may be very important).
New goods bias refers to the fact that new goods keep replacing old goods, and these new goods often are more expensive (as well as being better) than the old goods. (Note that this bias occurs steadily with the GDP deflator, while being relevant only periodically -- when the market basket changes -- with the CPI). By virtue of the fact that more expensive new goods replace cheaper old goods in the index/deflator, there is an upward bias introduced into the estimate of changes in the price level.
Quality change bias results from the fact that many goods undergo improvements in quality over time. These quality improvements often entail price increases (e.g., safety equipment or new technology in cars), but we do not want such price increases to be included in a measure of inflation (which should reflect price increases for goods and services of constant quality).
Estimates of the impact of new goods and quality change suggest that perhaps one to two percentage points worth of calculated inflation should instead be attributed to new goods and quality change -- i.e., measured inflation overstates the underlying "true" rate of inflation by 1-2 percentage points.
Substitution bias is a problem with the CPI. We saw in the micro part of the course that one response of consumers to changes in (relative) prices is to substitute away from commodities that have become more expensive and toward commodities that have become cheaper. However, since the CPI is based on a fixed market basket of goods and services, such substitution is not taken into consideration, except when the market basket is revised (as occurs about once every 10 years).
Again, then, the implication of these different sources of bias is that our measures of inflation overstate the true, underlying level of inflation. This overstatement is significant for several reasons. First, as we noted last time, cost-of-living adjustments (COLAs) are based on the CPI. Hence, if the CPI overstates the true underlying rate of inflation, COLAs for social security will be too large, overcompensating the elderly for such changes and straining the federal government's budget.
Indirect support for this argument may be seen by the fact that 30 years ago the incidence of poverty among the elderly was higher than that among the rest of the population, while today it is lower. Children have replaced the elderly as an especially poverty-prone group in the American economy.
Further, many other contracts and agreements provide for COLAs (e.g., union wage contracts, adjustments to tax brackets), so again if the CPI overstates inflation these adjustments will be too large (and hence, affect income distribution and government tax revenues). To put the matter a bit differently, if (for example) COLAs were limited to changes in the CPI minus 1.5 percentage points, tax revenues would increase and social security expenditures would decrease, thereby reducing the government budget deficit.
A second major reason for concern about the accuracy of our measures of inflation is that they are used to adjust nominal GDP to real GDP, and hence to estimate real economic growth. A one to two percent upward bias in estimated inflation entails a corresponding downward bias in estimated real economic growth. Thus, the implication here is that real growth is probably higher than implied by the official statistics.
We see articles in the newspaper frequently that growth in the U.S. economy has been too slow. The argument I've just made is that if we take seriously the notion that the U.S. inflation rate has been consistently overestimated, then real economic growth has been higher than the measured growth rate. This is the argument made in the article "Coming This Year: Marx for Dummies" from the Wall Street Journal (following page).
A third reason for wanting to measure inflation and real economic growth accurately pertains to international comparisons of real GDP. Such comparisons are useful for assessing differences across countries in the material standard of living of the population, and in changes of these standards over time. If our measures are overstating inflation and understating growth, then international comparisons of levels and trends in the standard of living will be flawed (unless everybody else has the same bias).
Further, as Parkin's discussion of China demonstrates, adjustments to real GDP for purposes of making international comparisons that simply take account of exchange rates may give a misleading picture of the population's real standard of living.
An alternate and generally preferable approach to making such international comparisons is one based on purchasing power parity. In brief, if we want to assess real GDP in country X and compare it to real GDP in the U.S. using a purchasing power par-ity approach, we would evaluate the quantities of goods and ser-vices produced in country X at the prices prevailing in the U.S. In the case of China, this approach yields a much higher estimate (6x) of real GDP per capita than one based simply on the exchange rate, principally because there are a number of goodsthat are inexpensive in China but relatively expensive in the U.S.
Real GDP and Economic Welfare
Our interest in real GDP as an indicator of the material standard of living reflects an underlying view that the material standard of living of the population is a good indicator of economic welfare of the population. However, as Parkin notes, there are several other elements of economic welfare broadly defined that are not included in our measure of real GDP. One of these items we've already discussed -- viz., quality improvements that are not taken into consideration and hence overstate our estimates of inflation.
Production that takes place outside of the market -- most notably, in the home and volunteer work (but, cf., subsistence agricultural production in many developing countries) -- is not included in calculations of GDP. On this count, then, GDP understates the true extent of productive activity in the economy (cf., housewife and work vs. market work).
Further, given the long-term trend toward increased labor force participation of married women and corresponding shifts in production of many activities from the home to the market (e.g., child care, food preparation), it is likely that on this count measured GDP will overstate the growth in economic welfare.
A third category of activities not included in GDP is represented by the underground economy. This includes both illegal (criminal) activities and unreported activities (not reported in order to avoid taxes or get around regulations; cf., home repair industry). The magnitude of the underground economy has been estimated at anywhere from 6-20 percent of calculated GDP.
Whether or not this biases estimates of GDP growth depends on whether this percentage increases, decreases, or remains stable over time. This in turn is likely to reflect trends in illegal activities (many would argue that incomes earned by drug dealers do not constitute an economic good) and also changes in the structure of taxes (cf., flat tax proposals, tax avoidance vs. tax evasion).
Two additional considerations that are most relevant to consideration of economic welfare are leisure time and environmental quality. GDP does not include any consideration of leisure, but clearly this is a "good" in the economic sense and an important part of economic welfare. Historically, leisure time has increased as the standard work week declined, early retirement became more prevalent, and vacation time increased; hence, on this count measured GDP growth probably understates improvements in economic welfare.
Taking the long view, the transition from an agrarian economy to an industrial economy has been associated with significant increases in environmental pollution. Since clean air and water may be considered as economic goods, this increased pollution is in part an unmeasured cost of economic growth. Further, expenditures that are made on pollution control are included in GDP, and indeed some economists have tried to develop measures of "Net Domestic Product" or "Net Economic Welfare" that subtract from calculated GDP expenditures for reducing "bads" like crime and pollution (cf., "Genuine Progress Indicator").
All things considered, then, it is clear that real GDP is a highly imperfect measure of economic welfare.
© 1996 David Shapiro
Return to the Econ 4 index page.
Return to David Shapiro's homepage
For questions, comments, and suggestions, write to: dshapiro@psu.edu.