Penn World Table

BIBLIOGRAPHY

The Penn World Table (PWT) evolved from the International Comparison Program (ICP), which was begun in 1968 at the University of Pennsylvania under the auspices of the United Nations Statistical Office. The ICP benchmark comparisons were built up from 150 expenditure headings related to gross domestic product (GDP), for which detailed price comparisons were made for about three to six items per basic headings. These benchmark comparisons permitted estimates of purchasing power parities (PPPs) for the heading levels and aggregates such as food, government, and all of GDP. The initial benchmark comparisons were for 10 countries in 1970 and 34 countries in 1975. These numbers increased to 65 in 1985 and 150 in 2005. Between 1985 and 2005, the European Union countries moved to annual benchmark comparisons, while the OECD (Organization for Economic Cooperation and Development) and associate countries went to three-year benchmarks.

The early ICP comparisons developed a framework and methods but covered relatively few countries. Because there was a strong interest both in wider coverage and estimates over longer periods, the group at Penn, led by Irving Kravis, began to experiment with reduced information methods to extend the estimates to nonbenchmark countries. Robert Summers and Sultan Ahmad developed an early short-cut method to move to more countries, and in 1978 Kravis, Summers, and Alan Heston elaborated the method and published estimates for 100 countries (referring to 1970). The basic method applied the relationship between per capita income and prices for benchmark countries to estimate PPPs of nonbenchmark countries. Summers and Heston took this effort still further, developing estimates for benchmark and nonbenchmark countries that extended forward and backward in time. By the early 1980s the original relationship among the PPP, exchange rate, and per capita income became much less reliable as countries moved to free up exchange rates. In response, Summers and Heston developed an estimating equation based on alternative price sources (e.g., post-adjustment allowances) and per capita income that improved estimates of nonbenchmark countries. PWT version 6.2 covers 178 countries for some or all the years from 1950 to 2004, with 2000 used as the base year. All of the expenditure estimates are expressed in international dollars, which have the purchasing power over GDP of the U.S. dollar.

The level of detail in PWT includes consumption, investment, government, and the net foreign balance. Estimates are provided in current and constant prices, with both chain, Laspeyeres, and terms-of-trade-adjusted GDP series. Per capita, per worker, and per equivalent adult comparisons are provided, along with comparative price levels for GDP and the major components. In PWT 5.6, estimates of the physical capital stock were provided, and this will be resumed in 2007 using an improved methodology.

PWT is widely used as both a research and a teaching resource in development and econometric courses. However, the OECD provides PPP estimates that are more reliable for their member countries than PWT. The main advantage of PWT is that it covers a wider range of countries for a longer time period. The World Bank also provides PPP estimates each year in their World Development Report, but it is only at the GDP level and is not a time series. In PWT, quality grades are provided depending on the number of benchmark comparisons in which a country has participated, and on the consistency of their estimates over previous versions of PWT.

The wide use of PWT reflects the fact that the use of PPPs for the conversion of national currency expenditures to a common currency has proved a more reliable explanatory variable in many applications than the alternative converter, the exchange rate. Over time it has been used in a number of cases—such as the appreciation of the Japanese yen and the more recent appreciation of the Euro against the U.S. dollar—when there have been no underlying changes in PPPs or real income levels. The implication is that conversions at exchange rates would have greatly overstated changes in the GDP levels in Japan and the Euro countries relative to the United States. Thus, while PPP conversions were first thought to be an improvement because they better captured differences in real product across countries, it turns out that they are also more desirable for changes over time.

Another advantage of PPP conversions is that they take account of differences in relative prices across countries. The importance of this frequently shows up in comparisons of shares of expenditures in national currencies. There is a systematic tendency for the national currency share of investment of GDP to be overstated for low-income countries and understated for more affluent countries, compared to comparisons at international prices where relative prices of investment to GDP have been taken into account. One consequence is that it appears that low-income countries get less additional output from their investment at exchange rates compared to PPPs.

Similarly, energy consumption in physical units like BTUs, as a ratio to GDP from PWT, provides a much more acceptable measure of energy efficiency than if exchange-rate-converted GDPs are used. This has important implications for the projection of energy consumption and fossil fuel emissions for fast-growing low income countries like China and India. PWT has also been widely used in growth studies testing the hypothesis that per capita incomes of countries will converge over time.

Researchers have used PPP-converted GDP from PWT measures of both production and welfare, but it is an imperfect measure of both. As an indicator of well-being, it is but one important element, and researchers look to other variables as well, such as climate, crime rates, and air and water quality. The United Nations uses literacy and health measures in addition to per capita GDP for its Human Development Index of the Human Development Report of the United Nations Development Programme. Productivity comparisons based on PWT can only be carried out at the GDP level, whereas much greater interest attaches to productivity in sectors of production like manufacturing, retail and wholesale trade, or agriculture. The Groningen Growth and Development Centre (GGDC) has carried out such PPP-based productivity comparisons by sector for a significant number of countries, and in PWT 7.0 (for 2008 or 2009) there will be an attempt to integrate their estimates on a consistent basis with the PWT aggregates.

PWT 7.0 will also undergo a major revision as it integrates the 2005 ICP benchmark comparison. It will provide more expenditure detail, such as actual household consumption, including government-provided health and education expenditures. Alternative methods of aggregation will also be explored, and there will be a more detailed treatment of the net foreign balance, so as to distinguish between production and income of a country. This version of PWT will involve collaboration between other centers, including GGDC and the International Data Center at the University of California at Davis.

SEE ALSO Exchange Rates; Purchasing Power Parity

BIBLIOGRAPHY

Center for International Data, Institute of International Affairs. http://cid.econ.ucdavis.edu/.

Groningen Growth and Development Center (GGDC). http://www.ggdc.net/.

Kravis, I., R. Summers, and A. Heston. 1978. Real GDP Per Capita for More Than One Hundred Countries. Economic Journal 88 (350): 215–242.

Kravis, I., R. Summers, and A. Heston. 1982. World Product and Income: International Comparisons of Real GDP. Baltimore, MD: Johns Hopkins University Press.

Summers, R. and A. Heston. 1991. Penn World Table (Mark 5): An Expanded Set of International Comparisons, 1950–88. Quarterly Journal of Economics 106: 327–368.

World Bank. International Comparison Program (ICP). http://web.worldbank.org/wbsite/external/datastatistics/icptext/.

Alan Heston