Tyranny of numbers

Iran’s subsidy reform program of 2010 depended heavily on a monthly cash transfer for its social acceptance.   About 450,000 rials per person has been paid to all Iranians every month since then without any condition, so people were free to do whatever they wanted with the money.  We know little about what they actually did, but there is a sense in Iran that this was not the best way to redistribute the energy subsidies.

There is much talk of improving the targeting of the cash transfer program, but less motivated by a desire to improve its equity than by the desire to cut the program’s cost.   The most talked about proposal is to limit payments to the poorer households only, for example, to the bottom 7 deciles.  I have already warned in this blog that this is impractical, arguing that identification of people in the top three deciles is very difficult and likely not worth the cost. I believe that the same amount could be raised by some very practical measures, such as charging more for the gasoline sold to consumers in richer neighborhoods and with more expensive cars.

Others have mentioned the need to make the cash payments conditional.  I have in the past advocated giving half the money to the woman in the household so as to increase the likelihood that it is spent on socially beneficial things like child education, nutrition, and health.  This proposal acknowledges the important role women play in Iranian households, and that their expenditure decisions may favorable welfare and growth enhancing expenditures, such as in child education and health, as reported, for example, by Benhassine, Devoto, Duflo, and Dupas (2011).  At the same time, I have been more skeptical of making the cash payments conditional, mainly because of the added administrative and monitoring costs.   Iran’s program is the simplest in terms of implementation because so far it has not tried to identify the poor nor to monitor how they spend their transfer money.

I learned something surprising about the unimportance of conditionality last weekend from a presentation in this year’s NEUDC conference, held at Harvard Kennedy School.  MIT PhD student, Johannes Haushofer, presented a paper (joint with Jeremy Shapiro), which is not yet available but you can read a policy brief based on it here.  Their research has been receiving a lot of attention lately and was featured in the recent issue of the Economist.

They report the results of a randomized controlled trial (RCT) that transferred cash via a mobile phone and without any condition to poor rural households

 in Western Kenya between 2011 and 2012.

They find that as a result of the unconditional transfer, expenditures increased on a variety of goods, including food, education and health.   More importantly, expenditures on “bad” adult goods, such as tobacco and alcohol, did not increase.  They did not present evidence on labor supply, but after his talk Johannes told me that it probably had the opposite effect because the cash transfer increased investment in livestock and small business.  It and several other papers question the assumption of superiority of conditional programs like Mexico’s oportunidades.

The basic result in the literature on conditional (CCT) vs. unconditional transfers  (UCT) seems to be that CCT works best when parents are unwilling to invest in something the government cares about, such as in girls (or other “marginal”) students.   CCT’s for education are better suited for poorer countries, such as Burkina Faso where enrolments are low, but not for countries like Morocco, where enrolments are relatively high already (Duflo 2011).  In Iran, except among poor rural families, school enrollment are relatively high, so perhaps UCT works as well.  As far girl education is concerned, I have not seen any evidence of gender discrimination in schooling at the household level in Iran. What I have found is that girls’ education is more income elastic than boys’ education (more of a luxury), something that an increase in income can  improve without conditionality.

There was over the summer also some talk of turning the cash transfers into in-kind.  This would remove the monitoring problem because it limits the transfer to a specific kind of good or activity, but creates new problems.  Economists, including this one, prefer cash transfers because cash lets the person with most information about what works best to spend it.  A friend who runs an integrated poverty program in rural southern Egypt was telling me the other day that donations of water buffaloes (in-kind asset transfer) by the Egyptian government to poor households in Qena in upper Egypt had caused much trouble for the poor recipients because they could not feed themselves much less the hungry buffaloes!  Again, the poor know their needs and constraints better, though awareness of intra-household allocation issues within the poor households leaves room for debate about who in the household should get the money.