Syndicate content

Playing with and Understanding Purchasing Power Parities

Kaushik Basu's picture

A fascinating feature of purchasing power parity (PPP) is more people hold an opinion on it than know what it means. This was in ample display last week, when the Global Office of the International Comparison Program (ICP), hosted by the World Bank, announced the latest PPP data for the world, pertaining to 2011. 

Putting aside complexities, PPPs may be viewed as an estimate what one US dollar can buy in different countries. In case a dollar in Ghana can buy three times what it can buy in the United States, then a person who earns 1,000 dollars each month in Ghana is said to earn 3,000 in terms of ‘PPP-adjusted dollars’. 

The calculation of PPPs is a massive exercise, involving the collection of detailed price statistics from around the world, followed by complex computations to work out the relativities.  Not surprisingly, the release of this data was widely anticipated and gave rise to newspaper headlines and front page stories around the world.

PPPs come laced with fun facts. It was discovered that compared to the last time PPPs were computed, in 2005, the economic geography of the world has changed considerably. The US is still the largest economy in the world but China is pretty much on its heels. In 2005, India was the fifth largest economy in the world, behind US, China, Japan and Germany. By 2011, it was the third largest having crossed Japan and Germany. Indonesia, which was a little over one third the size of Italy in 2005 PPP terms, was completely on par with it by 2011. 

But what grabbed the headlines last week were estimates based on back-of-the-envelope calculations that extrapolated from the 2011 data, something popular news outlets were quick to do. Thus The Economist, the Financial Times and many others announced how this year the Chinese economy is expected to surpass the United States’ economy, in terms of PPP-adjusted GDP. Fun with PPPs can be taken a step further by asking when exactly this will happen. According to the World Bank’s most recent published forecasts, US and China will grow in 2014 by, respectively, 2.8% and 7.7%. Using these to compute the daily growth rates and adjusting the GDPs with the recent PPP numbers, it can be shown that China’s economy will overtake the American economy on November 2nd this year. I have not yet been able to determine the time of day. And for those with an interest in astrological diversions, I may point out that November 2nd happens to be the birthday of the Chinese emperor Huizong (1082) and the 11th US President, James Knox Polk (1795).

While the above calculation is done carefully and is mathematically valid, it is important to understand that computing PPPs entails several assumptions and so we need to exercise utmost care in interpreting the numbers and the forecasts. 

To understand the conceptual problems, note that different countries do not consume the same goods. So which goods should we look at in comparing costs of living? One simple strategy is to use the goods that are common to the two nations. Assume for simplicity that between country X and US (which is treated as the benchmark for PPP computation for all countries), the only common good is the hamburger and that the same is true of country Y and US. Now suppose between 2005 and 2011, the US hamburger price is unchanged but hamburger prices have risen in X and fallen in Y. This will suggest that X’s PPP-adjusted GDP should be lowered and Y’s PPP-adjusted GDP raised. Thus, other things remaining constant, X’s (PPP-adjusted) GDP would have risen compared to that of Y.  

Now suppose X and Y are neighboring countries that consume 100 identical products and for all goods, except hamburgers, prices fell in X and rose in Y. It is perfectly reasonable for these countries to claim that X’s effective GDP has declined compared to that of Y’s, since, for all but one good, prices have fallen in X and risen in Y. And therein lies the problem that experts computing PPPs have to contend with. If each country’s PPP is calculated vis-à-vis that of the United States that will give us some clean numbers, but these may not make sense when we compare some of these countries directly with one another. Trying to maintain all these parities creates an impossible challenge. Scientific rigor can tell us what is patently wrong but it does not necessarily give us exactly one right answer. This forces us to use judgment and intuition, and opens up the scope for politics and pressure.

But instead of getting riled up, we should simply recognize that PPPs have these inherent shortcomings and treat statements of country size and power with, let us say, a fistful of salt. There are first the obvious reminders. For one country to be larger than another in terms of PPP-adjusted GDP does not necessarily make it a more powerful country. If the bigger GDP is driven in large measure by population size, then the country has more mouths to feed and so the amount left over for flexing muscle, waging war or conducting expensive diplomacy is less. 

Further, when two countries become engaged in a third country or region, the fact that one country’s money buys more within that country is immaterial because the bulk of the expenditure has to be incurred in a third country, in which case it is the simple, uncorrected exchange rate that is most relevant.    
 
Finally, what do the new PPP numbers do to our poverty computations? Recall that chronic poverty is currently defined as living on less than 1.25 PPP-adjusted dollars per day. Naturally, changes in PPPs are likely to affect the incidence of poverty. But before we jump to making these calculations mechanically, we have to recognize that the poor often consume very different goods from what the rich consume even within the same nation. Thus, if a variety of luxury goods prices fall and the prices of coarse grains and basic commodities rise, the overall PPP-adjusted GDP may rise but each dollar may actually buy less for the poor, suggesting that the PPP-adjusted incomes for the rich and the poor may well move in different directions.

Since standard PPPs are calculated for each country as a whole with attention focused on the average person’s consumption and not specially on the poor, we need to make checks before we can use the 2011 PPPs to compute poverty. There is some prima facie evidence that between 2005 and now, prices of food and some other basic consumption goods have risen relative to other goods. Therefore, from the fact that a nation’s PPP-adjusted GDP has risen we cannot conclude that its chronic poverty has fallen in step. In brief, while the PPPs can be taken readily to amend GDPs, the same is not true for poverty incidences. This is the reason why the World Bank’s official poverty statistics arrive with a lag after new purchasing power parities become available. That is what is happening now; there is an active research program to determine how the PPPs affect the poverty statistics.  

With these caveats in mind, if you want to switch on your computer (that is, if you ever switched it off), get the new PPP-numbers and use them to make comparisons across countries, extrapolate into the future, speculate about who will overtake whom and make forecasts of when the world as a whole will hit the 100 trillion dollar GDP mark, you are most welcome.

Comments

Submitted by Zahid on

Simply brilliant. A delicious piece of writing:)

Submitted by YK on

Interesting article. but i think the conclusion of the following should be reversed, ie X’s (PPP-adjusted) GDP falls in comparison to that of Y: "This will suggest that X’s PPP-adjusted GDP should be lowered and Y’s PPP-adjusted GDP raised. Thus, other things remaining constant, X’s (PPP-adjusted) GDP would have risen compared to that of Y."

Surely the $1.25 ppp adjusted dollar per day is for extreme poverty not chronic poverty. Its accuracy is also so in doubt in high-cost locations such as in most cities where low-income groups pay high prices for rent, water, access to toilets, health care, keeping children at school and transport

Submitted by Fred Vogel on

Some clarity is provided about several points made by Kaushik; his points are shown in italic and underlined with my response following immediately.

"To understand the conceptual problems, note that different countries do not consume the same goods. "

The reality is that there is considerable overlap in goods consumed by different countries; but in different quantities. A major improvement over ICP 2005 was to classify products by their importance in each country and to apply weights in the estimation of product level PPPs between countries.

"If each country’s PPP is calculated vis-à-vis that of the United States that will give us some clean numbers, but these may not make sense when we compare some of these countries directly with one another. "

The index method used ensures that the PPPs between any two countries are the same whether they were computed directly or indirectly between all other countries.

Trying to maintain all these parities creates an impossible challenge."

While it is difficult to maintain parities between all combinations of countries, it is not impossible. The most advanced index number theory was followed and robust computation methods were developed to calculate the PPPs.

"Therefore, from the fact that a nation’s PPP-adjusted GDP has risen we cannot conclude that its chronic poverty has fallen in step. In brief, while the PPPs can be taken readily to amend GDPs, the same is not true for poverty incidences."

That is why the Bank uses a subset of the GDP (individual household consumption) for poverty analysis. PPPs are calculated for about 100 categories (basic headings) for household consumption items; these include about 30 basic headings for food. A data file containing basic heading PPPs and expenditure weights will be released to data users for their research purposes. It should not be said PPPs will not or should not be used for poverty analysis. Instead it should be said researchers can use the detailed data and apply their own assumptions to provide their measures of poverty.

Fred Vogel
Deputy Chair, ICP Technical Advisory Group

Submitted by Joaquín Montes on

PPP are a snapshot on the situation at a particular point in time. Looking at their series and their volatility, I have the feeling PPP are not comparable along time; or at least, we are lacking some additional factor not yet included in our analysis. Indeed, GDPs are easily comparable in tradables. Non-tradables are the thing PPP is dealing with. In order to make a series, we need a deflactor. If trend prices of tradables are different from trend of non-tradables, the series on PPP can't be built.

I remember the first projections gave the year 2050 for China to surpass USA; there was even a page on The Economist for that bet (it should be still accesable). The most recent projection, made perhaps only one or two years ago, was 2017; and suddenly it happened this year. We are lacking some factor!

Submitted by Andrea Planter-Otter, MD, MPH on

Hamburgers are a small part of consumption. GDP is made up of not just C but also I+G . The price deflators for G may be elusive. China does have relatively large I+G. Can you compare price trends? For instance US spends only modest amounts toprotect public health. The rate of return on preventing contagion is extremely high.

This shows how relevant such spending is: http://infotel.ca/newsitem/mers-101/cp28352553

But China spends proportionately much, much less on the public good of disease prevention and control. How does this enter the PPP calculations, prof. Basu? Should the GDP be adjusted for the expected value of unprevented disease epidemics - surely a major contingent liability.

Submitted by Ranjan Ray on

The PPP is a concept that, as calculated and made available by the World Bank, is of very limited significance. The reason lies in the lucid explanation of Kaushik Basu of what the PPP means. Prices vary so widely within large countries such as India and China, and between expenditure percentiles, that it makes little sense to assign a single value to the PPP for the whole country. The ICP should spend more resources on calculating intra country PPP s and intra regional PPP s than it does to calculating between oountry PPP s. Moreover, we should move away from expressing every country's PPP with respect to the US dollar. For example, one cannot work out the PPP s between say India and Vietnam without going through the US $. This is analogous to flying from Delhi to Hanoi via Washington. The Global Office of the WB should allow greater control to the regional offices that are better placed to calculate the intra regional and intra country PPP s. Given that the ICP is spending so much resources on producing a set of numbers that means so little, it is time for the ICP to try to undertake a more meaningful exercise.

Add new comment