In this interview, Francisco Ferreira, director of the International Inequalities Institute at the London School of Economics and a Stone Center Affiliated Scholar, discusses the need for a better data-driven understanding of shifts in the income and wealth distributions, and of the reasons behind these changes.
What do you see happening in the next decade in inequality studies, and where would you like to see it go?
Ferreira: One area that has gained a lot of prominence recently is the subject of combining different datasets to gain a more accurate and comprehensive understanding of the distributions of income and wealth. I’m referring in particular to the fact that, having worked on the basis primarily of household surveys for almost half a century, economists — to a large extent led by Tony Atkinson, Thomas Piketty, and Emmanuel Saez — have started looking once again at tax data and other administrative datasets. Unfortunately, however, in most countries in the world, those datasets alone are not enough to paint a full picture of the distributions of income, because in most countries in the world a minority of people pay income taxes or, indeed, report their incomes to authorities.
This is now quite well understood, including by Piketty and his colleagues. And several people at the Stone Center have also been thinking hard about this. But I feel that the manner in which these datasets — tax data; data from national accounts; data from registries in countries that have those, as in Scandinavia; data from household surveys; data from social security and labor ministries — are combined still requires a lot more work, a lot more ingenuity, a lot more creativity and thought. These are not easy tasks, because the different types of datasets are organized in very different ways, contain different kinds of covariates, by which I mean information on variables other than income, like education, occupation, or family size and composition.
There has been some work recently on approaches for combining these various types of data, but I think they are in their infancy and clearly there is huge scope for more work on that. And I would also add that it’s important that there is methodological plurality and variety. There’s sometimes a risk that, too quickly, one approach becomes dominant, and then everybody does things in exactly the same way without testing the implications of alternative assumptions. This industry of combining datasets, which is a nascent one now, can benefit from a variety of different approaches as we try to establish what the best practice ought to be.
Another area to return to with new eyes and new tools are questions of causality: identifying the determinants of changes in income distribution. Because of the revolution in the analysis of fiscal data in particular — associated with an emphasis on describing top income shares — there has been a return to basic description of time series. That’s hugely valuable, of course: before we can set out to understand changes, we must measure and describe them accurately. But it shouldn’t be the only thing we do. We should not, in my view, abandon the efforts that the economics profession had made before, in terms of understanding why these changes are taking place. To what extent do they reflect changes in tax policy, or changes in the labor market? How much does technological progress contribute, versus trade? What changes are due to new patterns of labor force participation, or to changes in schooling? And so on.
Establishing causality in the domain of changes in income distribution is challenging because it’s very hard to study it through the various experimental methods that have become the gold standard for identification in economics. It is, of course, possible to study experimentally how a particular policy, in isolation, might affect poverty or inequality. But it is much harder to take the actual changes we have observed in a specific society over a period of time, and then to apportion “responsibility” for those changes among various competing alternative explanations. The distribution of income (or wealth) is part of the description of the general equilibrium of an economy, and responds to social, cultural, and political phenomena as well; parsing out causality is inherently difficult.
In the past we have often relied on decomposition analysis, instrumental variables methods or, occasionally, structural estimation methods to address these questions. But because these approaches are in general less “clean” than, say, randomized control trials, they have become less fashionable. But if we want to understand why things are happening to the distribution of income, and not only what is happening, then I think this will require going back to some of these non-experimental techniques, and to try to improve on them, rather than to just indulge in speculation around the description.
There’s been a proliferation of inequality studies centers in the last decade. Do you see more collaboration happening between centers? Is that part of your vision for the III?
Ferreira: First, I think we should say that there is a good deal of collaboration already. This is an area where people generally talk to each other quite a lot. Branko [Milanovic, of the Stone Center] is visiting LSE as a Centennial Professor for part of his time, during the next three years. I am, of course, delighted to be an Affiliated Scholar with the Stone Center. I am currently involved in at least two different projects with [Stone Center Affiliated Scholar] Nora Lustig and her team at the CEQ, based at Tulane University. François Bourguignon, from the Paris School of Economics, is also involved. Both Janet [Gornick, director of the Stone Center] and François are very closely associated with the Luxembourg Income Study [now known as LIS] and, of course, the Stone Center hosts their U.S. office.
There is also collaboration with the team at PSE; I’ve been there to examine Ph.D. theses and Piketty comes over rather frequently and gives talks at the III. There are ongoing projects with Facundo Alvaredo and, recently, we had an interesting conversation with Anne-Sophie Robilliard about the World Inequality Lab’s work on Africa.
So, I think a vibrant research network already exists, and that’s a wonderful thing. On the other hand, there is always more that one can do, and I would like to mention three thoughts. First, there ought to be much more collaboration with institutes and researchers in the Global South. At the III we do have some links, including to excellent researchers at the University of Cape Town and elsewhere in South Africa. We collaborate actively with researchers in Chile, have some ongoing work in India, and speak frequently to people in Brazil, for example. But it is too little. There is globally excellent work being done in places like CEDLAS, in La Plata (Argentina), and IPEA, INSPER, and other institutions in Brazil. There is lots of important work in China and India, of course. We need to build up our interconnectivity with scholars in the Global South much more!
Second, most of the collaboration that does take place is still within disciplinary boundaries. This is natural, as there are “language barriers” between different social sciences. But I think the payoff to interdisciplinary work is very high. Tony Atkinson used to say that there was much to be gained by “intellectual arbitrage” among different areas within economics, and I think he would have agreed that this extends across social sciences. At the III, we pride ourselves on multi-disciplinarity: we employ economists, sociologists, and anthropologists, and have faculty associates from at least 10 departments within the LSE. I know that the Stone Center also prioritizes interdisciplinarity, and has economists, sociologists, political scientists and others on board. I believe that is hugely important.
Finally, I’d like to say that collaboration should not mean uniformity or conformity. It is crucial, of course, that researchers maintain their intellectual independence, leading to methodological plurality. In a way, it’s important that we both collaborate and compete, in the sense that we try to approach questions independently from what others are doing, because that is part of the way in which science evolves.
Do you feel that the field of study itself, in terms of the kinds of students it’s attracting and the voices that are getting included, has changed over the course of your career?
Ferreira: I think there has been a substantial increase in the profile and perceived importance of inequality as a topic. When many of us who are now in our 40s, 50s, and 60s went to graduate school, economics was at the pinnacle of the neoclassical resurgence, and macroeconomics was largely built around representative agent models, which by definition had one kind of person: a representative agent! So they would study growth and fluctuations and cycles without ever thinking about inequality. There was no inequality in most of the important models. That started to change in the late ’80s.
There’s a 1997 paper by Tony Atkinson that’s not that well-known, called “Bringing Income Distribution in from the Cold.” In the ’90s, there were major breakthroughs in theory, and economic theory moved back to encompassing a multiplicity of agents. People started once again paying attention to the role of the wealth distribution in dynamic processes, including economic growth. We understood that nonlinearities in production functions could generate poverty traps, and that some kinds of inequality exacerbated inefficiency, so that the old equity-efficiency trade-off was not — is not — universal. Following that period, there was then a decade of massive expansion in empirical work. By moving the subject into the mainstream, it’s no longer just a thing for slightly odd Ph.D. students who are weirdly interested in these questions, like I was.
The other important change, which we have already discussed, is that there is more of a conversation across disciplines. Interdisciplinarity — of the kind we endeavor to promote in places like the Stone Center and the III — is still the exception. Academia works in silos. I like to compare it to ecclesiastical orders: you are inducted into a particular monastical order and that shapes how you behave and determines your credo. Even the way people participate in seminars is quite different across different disciplines. But that is beginning to change, as people within each discipline realize that it is important to listen to what the other groups are saying, even if it is a difficult thing to do because we are all so steeped in our own intellectual traditions, manners, and mannerisms. I think we are moving in the right direction here.
Do you have any advice for students who are starting out now?
Ferreira: As we discussed earlier, I think some of the important areas for future work are around the combined analysis of multiple different kinds of data on income and wealth, and then persevering with attempts to understand the causes of changes in inequality. Another area of inquiry that I think holds great promise is deepening our understanding of the mechanisms through which socioeconomic status is reproduced across generations, and therefore of what interventions could help break that reproduction. This is of course the broad theme of intergenerational mobility and inequality of opportunity. It is now reasonably well-established — in large part thanks to [Stone Center Senior Scholar] Miles Corak — that cross-sectional inequality and intergenerational persistence go together. Promoting greater mobility, opportunities that are less unequal, should be something of a holy grail for policymakers: it is where fairness, equality, and efficiency come together. And there is still much progress to be made in measurement and, in particular, in the causal understanding of the mechanisms that make or break mobility.
A final — and possibly unnecessary — piece of advice for Ph.D. students who are starting out now: speak to your classmates and colleagues. Ask them what they are working on and tell them what you are doing. Ask questions and work together. At the III we run a voluntary doctoral seminar series open to Ph.D. students from all departments whose work has anything to do with inequality. Attending them has been one of the highlights of my first year at the LSE. We are blessed with a group of very bright, deeply committed young researchers and it is wonderful to see how much they benefit from engaging with one another.
Francisco H. G. Ferreira is the Amartya Sen Professor of Inequality Studies and Director of the International Inequalities Institute at the London School of Economics. He is also affiliated with the Department of Social Policy and the Latin America and Caribbean Centre at LSE.