Tyranny of numbers

If you are someone who pays attention to economic news and have not been hiding in a cave for the past few months, you must have heard of the famous book by the French economist Thomas Piketty, Capital in the Twenty-First Century.  Since its translation was published in English earlier this year, it has sold more than half a million copies, which is astonishing for a book with many tables and charts (a publisher once told me that each chart cuts sales by 10%  — there goes that bit of wisdom).

Last month people were also talking about Piletty’s book in Tehran, and this month’s Mehrnameh, published this week, has a section discussing it, including a short interview by yours truly.   I must confess, as I did to the interviewer, that like most people who have bought the book, so far, I have only read the introduction (I have read, however, many of the book reviews — more pages of reviewes than of the book itself! Read an excellent early review by Branko Milanovic here).

There are two parts to Piketty’s argument and both have interesting implications about Iran.  The first is the claim that the rate of growth of capital (r) exceeds the growth rate of output (g), which causes the ratio of wealth (K) to income (Y) to increase over time and the share of labor to decline.  Piketty considers this a fundamental law of capitalism, one that undermines its legitimacy and requires political action to counter.  there is here in Piketty’s thinking hints of Marx (the falling rate of profit) as well as of Keynes (save capitalism from itself).  The second important point is about the share of the top incomes.

There is an interesting implication of this “law” or hypothesis for Iran’s growth process.  Because of the existence of oil revenues, there is a further reason why K might grow faster than Y, and why public policy is needed to restore the balance.   This argument is based on the claim that, in the absence of corrective action, oil income subsidizes capital and lowers its price relative to labor.   This is because foreign currency, and hence the price of imported capital goods, decline with rising oil incomes, pushing production in a more capital-intensive direction.  Oil and gas sold to industry, generally at a fraction of its world price, is another reason why capital becomes cheap relative to labor.

There are many reasons why a country should prevent the price of capital from falling relative to labor, such as having a large pool of unemployed youth and uneven creation of wealth à la Piketty.  I have argued here and there that in Iran energy prices should be raised to world levels and the earning used to subsidize wages of young workers.  I had thought of this argument as arising from efficiency considerations, but now I think there is also an equity angle to consider.

The implication of Piketty’s second point, about top incomes, is equally interesting and relevant for Iran.  Piketty made his reputation some years ago showing that surrvey data do not capture the incomes and wealth of the very rich, the top 1 or 0.1 percent, and that to capture those one had to use tax returns data.  He and his colleagues went on to found the top incomes data base that forms the empirical basis of his book.

It is no secret that household surveys in Iran and elsewhere do not catch the very rich in their net.  Since there are no tax returns data in Iran, or if there are such data they also miss the top incomes, we will never know what the true level of inequality of income in Iran is or has been.  What the Household Expenditure and Income Survey (HEIS) of the Statistical Center of Iran, the workhorse of income distribution analysis in Iran, will tell you is far from the truth.

Here is what HEIS (2012) would have you believe about the share of the top incomes in Iran: in 2012 the top 10% controlled 16.3% of total incomes and the top 1% earned only 3% of the total.  The same numbers for the US, estimated from tax returns data, are 50% and 20%, respectively.  The distribution of incomes in Iran  and the US, as observed in survey data, are not that different, so unless the unobserved parts are very different — more precisely, the right tail in the case of Iran’s distribution is much thinner — the HEIS estimates of inequality are wide of the mark.

The highest incomes recorded by HEIS in recent years (in billion rials per year) are as follows: 2009, 4.1; 2010,  2.2; 2011, 3.1; and 2012, 8.3.  Few people would regard these as accurate. The 8.3 billion rials in 2012 amounts to about 700 million rials ($21,000) per month of income.  This is what a landlord could earn from less that 10 luxury apartments in northern Tehran.  Speaking of rent, the highest rent income reported in 2012 is a paltry 80 million rials per month, or $2500.  No one can believe this to be the highest amount of rent earned in Iran.

Missing the top incomes can throw the calculations of inequality way off — well, not all calculations because trends in inequality are less affected by the missing top incomes.  A simple exercise on the Gini index (which is, incidentally, not known for its sensitivity to top incomes– GE(2) is) demonstrates my point.  Suppose there were 10 multimillionaires, people with 10 to 100 million dollars in assets, that were in the HEIS sample (out of more than 38,000 households) but interviewers could not get them to respond.  This many people in the survey would mean about 5000 families with wealth above $10 millions nationally.  Not so unreasonable.

What would the Gini index have been had they responded?

Let’s say for the sake of demonstration that these 10 families own assets, in billion rials, equal to 300, 600, …., 3000 each.   Then assume that they all earned the going deposit rate of about 20% on the rial value of their assets, yielding from 60 billion to 600 billion rials per year for each family.   I add these ten families (assuming a family size of 4 for each and sampling weight equal to Tehran’s observations) ) to the data and compare the Gini index for the new augmented distribution with that from the original HEIS .  The results are quite surprising: the addition of top incomes raises the Gini index from 0.33 to 0.58, which is the Gini value in the least equal of Latin American countries.

In the absence of reliable tax returns data, all hope is not lost; there are ways to get closer to the true distribution of income.  But there are many other reasons besides better understanding of inequality why Iran needs a reliable system to record individual incomes and wealth.  No modern society can live without one and no lasting improvement in the distribution of income is possible without an effective system of taxation.