I've been studying a large program in the US that's called the Earned Income Tax Credit, on which the government spends about $40 billion per year. It is intended to encourage low-income people to work more by essentially subsidising their work. For every extra dollar they earn, the government gives them money. So, say, for every $10 you earn, the government gives you $4. So effectively you are making $14/hour instead of $10/hour.

I think the idea of this program is great in principle, since it helps lower income individuals while encouraging them to get into the labour force. But I think there's a problem in the way it is set up. The way it works now is that you get a check when you go to file your taxes. On April 15, you and I may pay taxes. But lower income people in the US get a check from the government. We have done some pilot studies talking with people who get these checks, and the problem is that they do not really see how the amount they get from the government is related to how much they earn. So they understand that around February or March they get a big check for about $3,000 to $5,000 or something like that from the government. For them, it is a very large amount of money if they're making say $15,000 a year. The problem is that if they don't understand that working more affects the total amount of their income, then the whole point of the program is defeated. The idea of the program is to encourage you to work more by making you realise that your wage is higher.

To understand how serious this problem is, we're doing an experiment with H&R Block, the tax preparation company, with 40,000 low-income clients in Chicago, where we give them very simple information about how this program works. The tax preparers tell half the people (randomly selected) what I've just said to you: suppose you earn $10/hour, you should really think of your wage as $14/hour because the government gives you $4 for every $10 you earn. Next year when they come back to file their taxes, we're going to test if those 20,000 people to whom we gave this information increased their earnings relative to those who were not given that information (the control group). We've already done the experiment in 2007; in 2008 we will get the feedback. The question is whether 20,000 people in Chicago are currently changing their work patterns this year based on the experiment.

What are your thoughts on welfare policies?

In any country, welfare support is a very controversial area because I think there are two strong competing arguments. Both in India and the US you see there's tremendous growth in incomes at the top end. Some people are becoming incredibly well off while others are not, and this suggests that you want to try to help the lower income people share the fruits of growth. I just read that 80 percent of people in India live on less than Rs 20 a day. At the same time, we have a growth rate of about 10 percent per year in India. The growth rate benefits a very small fraction of the population. You then think that we need to try to help these lower income people in some ways through education subsidies, welfare payments, earned income credit programs, and so on. This is the argument for welfare that liberals would make.

The danger in this argument is that you must take market forces into account. How will you finance such a welfare program? If you finance it by taxing companies or wealthy people a lot, you essentially create a situation where I think it is like India in the 80s, there is not much growth because there is not that much incentive to start these new companies and make big investments because the government is taking a lot of the return. And on the other end, there's less incentive to work because the government is just supporting you. This is the argument against welfare that conservatives would make.

Both arguments are compelling and no doubt correct to some extent. So, you have to be very careful, that when you try to implement these big welfare policies, you do not reduce the level of economic growth such that everybody is not as well off. This is a very tough 'equity-efficiency' tradeoff. A lot of my research has focused on in what areas provide welfare support: where it is most valuable and how can we minimize these efficiency costs. In general, I think that short-term income support programs can be very effective (eg, unemployment insurance, health insurance). I also think that there is probably tremendous value in improving literacy and primary school education for lower income individuals, so that you can help them earn a good living rather than support them with welfare payments forever.

Image: Former US president Bill Clinton expanded the earned income tax credit programme in 2000.

Photograph: Paul J Richards/AFP/Getty Images