Tyranny of numbers

There appears to be some confusion surrounding the meaning and use of the PPP (Purchasing Power Parity) among some readers of this weblog and of Rastak, where Dr. Davoud Souri published a rejoinder to my comments on his poverty calculations. (I am sorry I cannot participate in Rastak discussions because I am not good in typing or composing in Persian!)

Comparison of poverty across countries is an important subject, and impossible to do without some sort of comparison of purchasing power of different currencies across countries.  Unfortunately, market exchange rates are not generally a good guide for such comparisons, hence the need for PPP comparisons.

A good starting point for a useful discussion of poverty and policies to combat it is to establish common ideas about what is regarded as high and low rates of poverty.  To establish such benchmarks, it is natural to refer to what other countries with similar level of productivity and income are able to achieve in poverty reduction.  It is one thing to complain about poverty if countries with similar incomes to Iran are doing much better and quite another if they are not, in which case the only point of reference is one’s own high and lofty ideals.

What does the PPP mean and how is it used in  poverty analysis?

First, it is only (but widely) used to compare standards of living across countries, not to establish poverty lines (none of the P’s in the PPP refers to poverty!).   The entire empirical growth literature is based on these numbers, so considerable amount of scientific effort goes into making sure the numbers mean what they are supposed to mean. The Word Bank (WB for short) publishes the numbers annually, but for years the real work has been done by researchers at the Penn World Tables at the University of Pennsylvania. Second, the PPP is not to be confused with the equilibrium exchange rate (one reader of Rastak proposes to buy all dollars at the (2007) rate of 3355 rials per dollar; I would too!).   The PPP is about how high prices are in the US relative to Iran, nothing more and nothing less. If anyone has a quarrel with the published numbers the burden of the proof is on them to show that the WB numbers are wrong. Convince the readers that what you can buy with 3355 rials in Iran is less than what you can buy with $1 in the US.  Failing that, ask someone who travels outside Iran or has a child studying abroad.  They know about the PPP.  (Someone I know had his daughter come home from Malaysia for a dental operation, air travel and all.)   Dismissing the World Bank calculations because their people do not live and shop in Iran is not a good substitute for doing the calcuations better.

Third, neither the WB nor I have suggested that the appropriate poverty line for Iran is 3355 rials or $1 per day.  The usual international comparison is based on $3 dollars per person per day (this is the so-called $2 per day rate based on the value of dollar in 1993). No one has suggested that this is the right number to use for Iran. Each country must pick its own number based on what it takes to have a minimum standard of living and in relation to average productivity in the country.  Poorer countries, such as China and India, often use the $1 threshold.   The number that I have picked in my paper on poverty and inequality is higher than $3).

But the problem with picking a poverty line specific for Iran is that it is hard to compare it internationally. For example, according to the World Bank, in 2005 the $2 per day (close to $3 in 2005) poverty rate for Turkey was 9.0%, while Iran’s was 8.0%. If you prefer a higher poverty line — such as the $10 line used in Dr. Souri’s study—chances are that you will find Turkey’s poverty rate is still a bit higher than Iran’s.

A fourth point is about the use of the poverty calculations. They have domestic use if there are government policies that trigger when a person’s income falls below the official poverty line.  Unlike Turkey, Iran does not have a national poverty line, possibly because there are no such policies in Iran. Poverty targeting in Iran is done differently than in many other countries, where a ministry is usually responsible for poverty alleviation. In Iran, the principal agency dealing with the poor is Komiteh Emdad, which is semi-public and relies on its own criteria to identify who is poor, mainly derived from information supplied by the community.

The argument about where the poverty line should be makes sense in politics and policy making but not in science.  I may think that life under $20 per day is very hard, but that is my personal opinion, and policy makers may decide to deliver services only to those below $3.  The point worth discussing is how to measure change in poverty from one year to the next.