December 14, 2009
We Should Amend, not Discard, National Income Accounts
We Should Amend, not Discard, National Income Accounts-Becker
The justification economists give for concepts of national income, such as Gross Domestic Product (GDP) and Net Domestic Product (NDP), is that the price of a good measures both the utility households gain from consuming that good, and under competition price also measures the cost to sellers of producing the good. This interpretation of the value of goods sold in terms of both utility and production costs helps justify not only concepts like GDP, but also helps link consumer price indexes to the cost of producing a given level of utility to the average consumer, and also ties producer price indexes to the cost of producing given amounts of output.
A tremendous achievement of the 1930s was the construction of early measures of GDP and NDP by pioneers in national income accounting, like the Nobel-winning economists Simon Kuznets and Richard Stone. They recognized from the outset that their national income accounts only very imperfectly measure what is produced each year, let alone consumer utility. Despite these limitations- I discuss in detail a couple of these shortly- these national income accounts have proved very useful in quantifying changes over time in the real value of what is produced in numerous nations and regions of the world. They have also provided useful measures of changes in the degree of inflation at different time periods.
For example, measured real GDP shows clearly that the worldwide Great Depression of the 1930s was far more severe in its effect on production and incomes than the world recession of the last couple of years. The growth in real GDP during the third quarter of 2009 strongly suggests that this recession ended during that quarter. Data on the growth of GDP per capita also clearly indicate that China and other rapidly developing countries during the past couple of decades significantly reduced the gap between their own standard of living and that of the wealthy Western countries, although cross country comparisons of real GDP typically make additional price adjustments to national income accounts for differences in purchasing power.
Similarly, indexes of changes in consumer and producer prices unambiguously show that the middle and end of the 1970s were periods of substantial inflation in the United States and Europe, that Japan had significant deflation during the 1990s, and that many countries experienced mild deflation in prices during the recent recession.
So despite their various limitations, measures of aggregate changes in real output and prices, and other concepts related to the national income accounts, have been valuable tools for policy-makers, economists, and others. Although the United Nations Human Development Index (HDI), which tries to rank countries by their level of "human development", has the virtue of incorporating improvements in life expectancy-I return to this subject later- in all other dimensions it is far inferior to national income accounts. In constructing the HDI for different countries, this index gives equal weight to measures of their life expectancy, education level, and GDP. Since education is an important source of improvements in life expectancy and GDP, the HDI involves serious double counting by including education as an independent variable on an equal footing with these other variables. Moreover, even aside from the double-counting problem, the 1/3 weight attached to life expectancy, education, and GDP are completely arbitrary, and not based on any analysis of either consumer utility or costs of production.
Although national income accounts are far more useful than the HDI and other suggested alternatives, these accounts have serious omissions and other limitations. Posner discusses many of these, so to avoid too much duplication I will concentrate on only two. GDP and other output measures do not include work in the household (neither does the HDI), such as time spent on child rearing, preparing meals, and other valuable household activities. These activities have traditionally occupied much more of the time of women than men, so that GDP and related measures of output exclude very important production mainly by women.
New national time use surveys for the United States and other countries allow this omission to be corrected, at least to some extent. Time use surveys give the amount of time spent on child rearing and other forms of work in different households, and also the time spent on leisure. Other data sets provide evidence on the hourly earnings of women and men in different households when they are working in the labor force. By multiplying the time spent at household work, and perhaps also the time spent on leisure, by measures of their actual or potential hourly earnings, one can derive an estimate of the cost of household time that is conceptually similar to measures of the cost of producing market work.
Such measures of total output or income adjusted for household time would be imperfect, but they would be superior to income measures like GDP that exclude the value of household work. For example, when increases in economic development and education raise the time married women spend in the labor force mainly by reducing the time they spend working at home, national income accounts exaggerate the increase in total output by failing to subtract reductions in output produced at home.
National income accounts do not incorporate improvements in life expectancy, which is a major limitation because of the magnificent advances in health and declines in mortality during the past 100 years throughout the world. The HDI has the virtue of incorporating life expectancy into its measure of human development. However, this index ignores modern economic research that provides a method to use consumer and producer behavior to combine changes in national income with changes in various types of mortality risk into a single index of what is called "full" income. This method calculates the "statistical value of life", which measures how much individuals are willing to pay for various reductions in mortality rates (see the fuller discussion in my post on Dec.15, 2007).
The concept of "full" income combines changes over time in GDP with estimates of the value placed on reductions over time in mortality risks (or increases in mortality, as in some African countries in recent years due to the Aids epidemic). Such full income measures combine changes in life expectancy and in ordinary income not in an arbitrary way, but by extending the willingness to pay concept used in national income accounting to valuations of changes in life expectancy.
A common finding is that the usual measures of per capita incomes grew only a little more rapidly during the decades 1960-2000 in poor and less developed countries than in richer countries, even after accounting for the rapid growth in the world's two most populated countries, China and India. That finding on changes in world inequality is greatly altered when comparisons are made not of GDP per capita but of full income per capita. Phillipson, Soares, and I construct such measures of full income for about 100 countries for these decades (see our "The Quantity and Quality of Life and the Evolution of World Inequality" The American Economic Review, March 2005). Since mortality declined more rapidly in poorer countries than richer ones, adding the value placed on declines in mortality to measure changes in full incomes have a much greater effect on adjustments for poorer countries than for richer countries. The result is that the growth rate in full incomes are generally much more rapid in poorer countries than in richer ones, which implies that inequality in full incomes per capita declined greatly across nations during the past several decades, even though inequality among countries in GDP per capita did not change much.
Despite all the limitations of national income accounts, they are still the best available way to measure changes over time in the amount of production in an economy, and even for comparisons of real output per capita among different countries. Nevertheless, these accounts can be made even better by including the value of time spent at home, and the value of improvements in life expectancy. A demonstration that such changes in national income accounting are feasible would encourage serious efforts at including some of the other improvements in national income accounts discussed by Posner and others.
Should We Jettison GDP As an Economic Measure? Posner
This past September, the "International Commission on the Measurement of Economic Performance and Social Progress," which had been appoined by French president Sarkozy and was chaired by the well-known economist Joseph Stiglitz, issued a report of almost 300 pages criticizing Gross Domestic Product as a measure of social welfare and proposing a variety of alterations and alternatives. Such criticisms are nothing new, but Stiglitz relates them to the deficiencies of GDP as a measure of our current economic situation, and this makes the report especially timely. For U.S. GDP (and that of many other countries) grew in the third quarter of this year, leading a number of economists and journalists to declare the end of the recession that began in December 2007 and that dramatically intensified in the financial collapse of September 2008.
There are three types of objection to the GDP as a measure of welfare. The first is that it is defective even from the narrowest economic perspective. GDP is the market value of all goods and services produced in a year. It thus explicitly excludes nonmarket values. But its treatment even of market values is defective because it excludes depreciation. Suppose the calamitous effects of Hurricane Katrina on New Orleans and other parts of the Gulf Coast had caused (it probably did cause) a surge in the output of various services such as emergency relief, building repairs, and construction. The market value of those services would be counted as part of GDP, without subtraction for the depreciation of the value of property that the flooding triggered by the hurricane caused.
Another objection to GDP as a measure of economic welfare, also a criticism based on economics though now viewed a little more broadly, is failure to adjust for monetizable, but not monetized, economic values. An obvious example is household production, which can be valued in money terms by estimating what the household producer would earn in the market; that is only a lower-bound estimate, but it is better than nothing. Leisure, which is also a value, can be monetized similarly.
Quality is another economic dimension that can be monetized, as is recognized in calculating the consumer price index; without an adjustment for quality change, the rate of inflation would be greatly overstimated. More to the point, it would be wrong to conclude that if the cost of making a product declines and competition forces the producers to cut price, there has been a loss of value. A great deal of the modern increase in the standard of living is due to improvements in product quality that do not result in cost increases commensurate with the improvements, and often result in lower costs. A dramatic example is modern dentistry. A possible further example is increased longevity, which can be monetized; the problem is relating increased longevity to the enormous expenditures on health care.
Turning to bads, economists can and do estimate the costs imposed by crime, pollution, and traffic congestion, but these costs are not subtracted, in the calculation of GDP, from the costs of police and prisons, costs of pollution control, and costs of dealing with congestion—all those costs (except a loss of production from congestion delays) are included in GDP. Put differently, the monetizable value of investments in police, pollution control, and reducing congestion does not enter into GDP.
Even with all the suggested corrections made, GDP would be an imperfect measure of economic output, because government provides many services that are difficult to value: expenditures on the military and on foreign and domestic intelligence and counterintelligence are conspicuous examples. It would be odd to say that the "market value" of a bomber that costs $100 million to build and is sold only to the U.S. Air Force has a market value equal to the purchase price. Its market value in a meaningful sense would depend on its contribution to reducing the expected costs of foreign threats to U.S. security.
Which brings me to the third and broadest problem with GDP as a measure of welfare--that even if improved along the lines I have just suggested it would not really measure happiness or well being. Market value is a function mainly of cost. The value that people derive from goods and services is better measured by what they would pay for them if competition did not reduce their price to or near the cost of production; but that value ("consumer surplus") is difficult to estimate. Or consider—coming closer to current events that have sharpened traditional concerns with GDP's adequacy as a measure of welfare—the anxiety that people who are involuntarily unnemployed experience.
The second in command at the international commission was the economist Amartya Sen, a pioneer (along with the philosopher Martha Nussbaum) in attempting to develop measures of human "capabilities" and ranking countries according to their ability to equip their citizens with such capabilities (long life, adequate nutrition, education, etc.). The United Nation's Human Development Index attempts such a ranking, and some might think it a candidate for replacing GDP.
So should GDP be changed or abandoned or supplemented or supplanted? I think not, for three reasons. The first is that even the adjustments that every economist would favor in principle, such as subtracting depreciation from market value, involve contestable judgments (there is a measure called Net Domestic Product that subtracts normal depreciation from GDP). If private economists want to make such adjustments and offer up their own estimates of the economy's output, fine; but GDP is an official government statistic, and to avoid the suspicion and perhaps reality of political interference it is essential that government statistics be calculated in a thoroughly objective, uncontroversial manner. An extreme example is the U.S. census, which would be a lot cheaper to conduct if it used sampling, rather than trying to count every household and individual in the United States. But the loss of credibility would be damaging.
The objection to adjusting GDP would grow with every adjustment. And adjustments that were not monetizable would require controversial decisions on weighting. That is a standard problem with multicriteria rankings.
Second, crude as it is, unadjusted GDP is at least roughly correlated with adjusted measures of welfare. In the international commission's report, adjustments for leisure and other nonmonetized but monetizable values boost France's GDP from being 66 percent of America's to 87 percent. In another words, France is poorer than America, though not by as much as the unadjusted GDP figures suggested. Anyone who knows something about France knows that it's a wealthy country, though not so wealthy as the United States, and that the French don't work as hard as Americans.
Third, except at extremes (Norway versus Zimbabwe, say), the significance of GDP lies not in its use as a method of ranking nations, but its use as a method of measuring the business cycle in an individual nation. A chart of U.S. GDP oscillates around a trend line of about 3 percent per annum; there is a big dip in the Great Depression of the 1930s and a smaller though still significant dip since 2007. The oscillations in GDP since 2007 provide a rough but serviceable starting point in appraising the performance of the economy—a sharp drop of GDP in the last quarter of 2008 and the first quarter of 2009, a smaller drop in the second quarter, an increase (of 2.8 percent) in the third quarter, which still leaves GDP well below its trend line.
But it is necessary to emphasize that it is just a starting point. I disagree with economists who say the "recession" ended in the third quarter. The depression (as I think we should call it if only because of its enormous potential political consequences) has caused massive unemployment with all the associated anxieties and hardships, has greatly reduced household wealth, has caused private investment to turn negative, has cost the government trillions of dollars in lost tax revenues and recovery expenditures (TARP, the fiscal stimulus, the mortgage-relief programs, the auto bailouts, etc.), has undermined belief in free markets and altered the line between government and business in favor government, and is threatening a future inflation while deepening our dependence on foreign lenders. To view a change in GDP from negative to positive as signifying the end of a depression (by which criterion the Great Depression ended in 1933 and again in 1938) is to misunderstand the utility of GDP as a measure of economic activity.