IMF finds poor less poor


It's rare that the International Monetary Fund does makes the front page of the New York Times. When it does, it's probably an exception that proves a rule. On May 20, the newspaper's Steven Greenhouse reported that in a "long overdue" recalculation, Fund economists had finally corrected big underestimates of the income of "developing" economies. The revision is particularly friendly to China, which moves up from the 10th-largest to the third-largest economy in the world, placing it ahead of Germany, and not all that far behind Japan. Other beneficiaries of mathematical upward mobility are India, Indonesia, Mexico and Brazil.

China's promotion was the main focus of the Times story; the economic rationale and political implications of the IMF's new method were left unexplored. It's too much to expect that a newspaper venture into complex territory; it's always safer to stick with a couple of simple hooks — New Technique! Swelling China!

In the old days, the IMF converted national currencies into US dollar values using market exchange rates. These measures were used both to compare countries' income levels and to compute global and regional economic growth rates. From now on, the Fund will use a technique called purchasing power parity (PPP). PPPs, in the words of Organisation for Economic Cooperation and Development (OECD), which computes PPP figures for the rich industrialised countries, "are the rates of currency conversion that equalise the purchasing power of different currencies. This means that a given sum of money, when converted into different currencies at the PPP rates, will buy the same basket of goods and services in all countries."

Market noise

In a fascinating confession, the normally market-worshiping IMF notes that "market exchange rates may not reflect their PPP equivalents for a variety of reasons. For example, market rates may be distorted by speculative bubbles, exchange market intervention, asymmetric speed of adjustment in goods and asset markets, or macroeconomic shocks." In other words, free markets don't do what we've been told they're supposed to do — adjust frictionlessly to underlying reality. Instead, the Fund admits that prices can and do deviate profoundly from supposed "real" values.

Such market irrationalities apply mainly to the rich countries, whose currencies are traded to the tune of $US1 trillion a day. Hot money often flows into a currency one month and out of it the next, with no fundamental change in a country's economic position. But for the poor countries, there's another reason that market exchange rates don't tell the whole story.

In theory, market exchange rates tend — emphasis on theory and tend — towards values that equalise the price of common, widely traded goods. For example, TV's of comparable quality should cost roughly the same in the US, Mexico, and France, since appliances are shipped around the world. If televisions were expensive in one country, every producer would try to sell TVs there, resulting in a rices towards the global average. Or if TVs were cheap, producers would shun the market, forcing up prices. Similarly, since an exchange rate represents a country's general price level for traded gods, a high exchange rate makes a country's exports expensive in world markets; its producers would have to cut prices and/or its currency drift lower to bring prices in line with prevailing levels. Again, this is a theory of tendencies, not a description of daily reality.

But services, from haircuts to education to lawyering, aren't traded. The price of services is determined largely by wages, and since wages are low in poor countries, untraded services would look cheap next to traded goods. So a TV may cost the peso equivalent of $US400 in Mexico, but a haircut may be gotten for 50 cents: the price of TV's is set on world markets, but that of haircuts is set locally. Since goods prices drive exchange rates, but services are an important part of normal consumption expenditures, market rates systematically understate the domestic buying power of Third World currencies.

Approximating value

Market exchange rates can produce very low income estimates. At market rates, for example, the average African gets by on $US340 a year, according to the World Bank — a number that seems impossibly tiny to a First Worlder. Even the $US2450 figure for Mexico, an upper-middle-income country in the bank's classification, looks incomprehensible. The reason such low incomes can sustain life — however barely in many instances — is that the cost of living is much lower in the South than in the North. PPPs aim to compensate for these differences.

To derive PPP-based exchange rates, analysts figure what a basic market basket of goods and services costs in the universe of countries covered, and then figure what exchange rates would equalise the basket's cost in all countries. The broadest PPP exercise is carried out by the United Nation's International Comparison Program (ICP), which covers roughly 400 products in 137 countries. ICP data has been massaged by economists at the University of Pennsylvania to produce a massive database called the Penn World Table (PWT). Other projects are carried out by the OECD and the European Community, to survey their member states (24 and 12, respectively); the OECD's shopping list covers over 2500 categories.

The process sounds simpler in principle than it is in practice. Housing is measured by the physical dwelling, with no adjustment for location, meaning that a 120-square metre apartment with one bedroom and one bath would have the same "real" value in Amsterdam or Kinshasa, Harlem or Berkeley. Government services like education and health are often free; how do you measure their value? The ICP values elementary education by the number of teachers employed, "appropriately standardised for training", and medical care by measures like number of hospital bed-days. To measure value, price must be measured against quality. Does a higher teacher/student ratio mean that education is better, or teacher productivity is lower? Does a longer hospital stay for the same disease signify better or worse care? Though the final PPP values are carried out to two decimal posite of thousands of judgment calls.

Even the ICP, the most ambitious PPP project of all, on which the IMF's figures are based, is less universal than it appears at first glance. Of the 137 countries covered in the PWT, only 77 have ever been surveyed in detail; the remainder are covered by statistical approximations. Even some of the surveyed countries haven't been examined in some time. Chinese figures, to take an important example, are based on a spotty 1975 survey — that, despite the profound changes in the last 18 years.

The subjective factor

Different analysts can come up with very different results. Both the World Bank and the PWT use the same raw ICP data, yet their "real" income figures are often far apart: The bank's estimate of China's income is a third lower than the PWT's, of Zaire's, more than twice as high. The OECD's estimate of Norway's "real" income is 15% higher than the PWT's. Though these are extreme examples, they show that estimating PPPs is an inexact science. The PWT's compilers grade the quality of their data for every country; worldwide, the average is a B when countries are weighted by income or C when weighted by population. For Latin America, the averages are C/C; Asia, C/C; China, D/D; and Africa, D+/D+.

According to the Wall Street Journal, the IMF discarded the World Banks's and the PWT's estimates of Chinese PPP incomes because they would have placed the country second in the global rankings, ahead of Japan, and nearly half the size of the US. Such a promotion made the IMF "uncomfortable". Instead, the Fund settled on an obscure, unpublished US Census Bureau estimate.

One should be wary of measurements of the fundamentally immeasurable. While specific money incomes and prices do exist, such things as real incomes and general price levels do not. As Keynes argued in The General Theory, "To say that net output to-day is greater, but the price level lower, than ten years ago... is a proposition of similar character to the statement that Queen Victoria was a better Queen but not a happier woman than Queen Elizabeth — a proposition not without meaning and not without interest, but unsuitable material for the differential calculus. Our precision will be a mock precision if we tried to use such vague and non-quantitative concepts as the basis of quantitative analysis". Mathematical economics often builds elaborate structures on chimerical foundations. This doesn't mean that attempts to devise measures of real welfare are pointless — economic analysis would be impossible without them — but it does counsel humility in the face of complexity.

While "real" measures do not exist, money does, yet economists will often go to amazing lengths to deny that money is one of our fundamental principles of social organisation. For two centuries, much of mainstream economics has assumed that money — gold, paper currency, or the entire institutional structure of finance and credit — is just a neutral medium of exchange, with no great effect of its own. The entire PPP exercise is within this tradition.


Suspicions of hidden agendas are excited by the IMF's approving quotation of Robert Lucas' remark that income differences between North and South are "literally too great to be believed". Lucas, one of the founders of a capitalism-is-perfect school called the New Classical Economics, said that in a 1988 article that attempts to develop a model of economical development that is mechanical enough to be "put on a computer and 'run'". His model explicitly ignores finance, "treating all exchanges as though it involves good-for- goods."

"In general," he declares, "the importance of financial matters is very badly over-stressed." Happily, this assumes away the $US1.4 trillion the world's poor owe its rich as well as the perpetually deteriorating terms on which the Third World trades with the First (in which depreciating commodities are exchanged for expensive manufactured goods).

Or, as the IMF puts it in the May 1993 edition of its World Economic Outlook: "[T]here are clearly purposes for which PPP — would not be appropriate, for example, to use PPP-based weights to aggregate measures of international trade and capital movements, which are transacted at market exchange rates, or data for external debt and debt service."

It means something that world TV prices place them way beyond the means of most African households; Sony would never accept PPP estimates in lieu of cash. More importantly for development, power plants, machine tools, drugs and books are priced at conventional rates of exchange, not some "real" level. Debts, too, are denominated in hard currencies; the IMF won't cash PPP cheques. And all foreign investors care about are cash wages; PPPs don't close the vast gap between First and Third World labour costs.

While disguising the hard discipline of the cash nexus, PPP calculations also make the poor seem less poor, thereby reducing further what little pressure there is for global redistribution. On this score, we can turn to New York Times economics columnist Peter Passell, always a reliable guide to vulgar centrist opinion. Embracing the new method with enthusiasm, Passell wondered why it took the Fund so long. He answers, with the help of MIT economist Rudiger Dornbusch: "[P]olitics dominated science. Systematic underestimates of living standards gave poor countries a stronger moral case for aid from the rich ones: Exchange-rate based calculations show the industrialised countries hogging 78% of the world income, while purchasing-power calculations put the figure at just 54%."

The poor, through their control of the IMF and Northern media, have been putting one over on the rich. But now that we know that the poor aren't so poor after all, we can forget the vast differences between North and South in literacy, life expectancy, and maternal and infant mortality and feel better about ourselves.

[From Left Business Observer
. For subscriptions write to: 250 W 85 Street, New York, NY 10024-3217, USA.]