Shanta: Jishnu, your blog post and mine on cash transfers generated a lot of comments. Some people argued that giving poor people cash will not “work” because they will spend it on consumption rather than on their children’s education, which is something we care about. What do you have to say to that?
Jishnu: I don’t think the question “does giving cash to poor people work?” is well-defined. It can only be answered in the negative if we (the donors who give the cash) impose our preferences and judge what poor people spend on relative to those preferences. But if we give poor people cash so they will be better off, then—by definition—they are better off, regardless of how they choose to spend the extra money.
Shanta: Before you get too philosophical, let’s discuss the concrete issue of education. Everyone agrees that education is one of the best routes out of poverty. But if people spend the cash on other things, they deny their children the chance to learn and escape poverty. Wouldn’t it have been better to cut down on the transfer and use the savings to either subsidize education or give scholarships? Isn’t this what conditional cash transfers try to do?
Jishnu: I think the problem of thinking about it in this way ignores the fact that people may spend the money in other ways that could help them get out of poverty, even for a little while. What do we make, for instance, of the person who used the money to get bail for the main breadwinner in the family? So, let’s get even more concrete. Here are four potential reasons why people don’t send their children to school:
- The benefits are not worth the costs. For instance, the school may be too far (especially for girls) or the children who do go to school don’t learn anything once they get there.
- The benefits are high, but the costs are borne by people who may not be able to share in the benefits. This is a failure of contracting. For instance, parents believe that if children get educated, they will move to the city and not care for them in their old age.
- People are credit constrained. The benefits are high, they will be shared, but at the moment I can’t borrow the money to pay for schooling. This may be particularly salient in the face of bad shocks:
- When there is an illness in the family, kids are pulled out of school to save money, and perhaps to engage in income-earning opportunities.
In the first case, if the costs are too high, the better option would be to provide transport, or build schools that are closer to home, as was done in Afghanistan with stunning results. That the benefits may be too low was demonstrated in Cambodia: Conditional cash brought children to school, but they did not learn anything. In the second case, I would subsidize education because there is an externality: the inability of households to contract for future transfers. In the third case, a smooth cash flow through direct cash transfers may work better than conditional cash transfers because the family is not penalized when illness strikes as they fail to meet the condition without understanding the nature of the underlying problem, I find it very hard to advocate for one type of transfer over the other.
Shanta: I think you’re saying something more than the old cliché “tailor the solution to the problem.” We’ve known since Samuelson’s 1954 paper that there are two rationales for public expenditure—market failure and redistribution. Too often, in practice, governments use the same instrument to address both. For instance, there is some evidence that there are externalities associated with primary education (the benefits to society are over and above those to the individual student). Even if there were no externality (all the benefits were private), we would still want poor people to have access to education, since it’s a way out of poverty. So we provide free education for all. But this combines the two rationales with one instrument. For the externality, we should subsidize education to the amount of the difference between the social and private benefit—for everybody. For redistribution, we should make sure poor people become less poor through targeted redistribution.
I wonder if there is a more general point here. When the rationale for government intervention is redistribution, the default instrument should be cash transfers (because these have the greatest potential to improve the poor’s welfare). Any deviation from this should be justified on grounds that it achieves the redistribution motive better. When the rationale is market failure, then the instrument should be a tax, subsidy or regulation.
Jishnu: With one wrinkle. When the motive is redistribution, moving away from direct transfers to the poor requires somebody (the government) who represents the poor to decide on how they should spend the money. But what if the government is insufficiently attentive to the needs of the poor? This has been a constant sticking point in aid debates, to the point that it is argued that aid sustains unresponsive governments far beyond their past-due dates.
In my own work I often wear two very different hats. One strand of my work highlights, for instance, the very poor quality of health care available to Indians—particularly those who live in poverty. I have argued that this, in part, reflects a failure of the government to ensure basic accountability over service providers—to the extent that 40 percent of Indian doctors are absent on any given day; and when nurses’ wages are tied to attendance (you don’t get paid if you don’t show up to work), they break the time-stamp machines. In this line of inquiry, I am not working with the government; I am working to demonstrate a failure in their ability to attain their own stated goals.
My second hat is the one I wear when working on specific World Bank or government programs. For instance, I have helped design cash transfers after an earthquake in Pakistan and worked in the initial days of India’s health insurance program. I have also researched these programs, but that is long after my implementation hat has worn out. When working on these programs, I am trying to ensure that they function well and that some systems are being put in place to ensure their sustainability. What I am not doing is asking whether these programs make sense in the first place (did those cash transfers keep a government in place longer than it should have, should insurance be provided through private companies, financed by the government) and what the broader implications are for the overall organization of state-citizen relationships.
At least for me, it is difficult to separate out these two tasks sufficiently to ensure both that I am fully committed to improving a program and, at the same time, ask whether that program makes sense from a broader societal perspective. If the answer was “no”, would I still keep at it the same way? What if the answer was “Yes, but only if everyone believes that it would work”? I give this example to highlight a conundrum we face every day at The World Bank. We see non-functioning institutions. We work with them every day. But it is hard to both work on improving programs and simultaneously question whether those programs make sense.
Shanta: Sure, but some of us don’t have the luxury of wearing two hats (or having enough heads to accommodate them). As chief economist, my view is that there is only one client—the poor in my countries. The government is the intermediary to reach the poor. If the government is a functioning intermediary, then I work with them (and I have pretty low standards for what “functioning” means). But if the government does not seem to represent the interests of the poor (and there is evidence, often derived from your work, to that effect), then I work directly with the public, informing them of the evidence, so that they can bring pressure to bear on government to give them what they are due. Sometimes, this makes governments (not to mention my managers) uncomfortable, but if we are serious about ending poverty, I’m convinced that this is how we should operate.