Mark Rosenzweig is Frank Altschul Professor of International Economics at Yale University, and was one of the original leaders in bringing theory and micro-level data to addressing development questions. We caught up with him after a recent symposium, which honored his achievements, and celebrated him turning 70 and continuing to produce important new work.
2. A lot of your work starts with a question about why we see certain behaviors in developing countries (e.g. why don’t we see more people migrating in India given the dispersion in wages?), puts together a theory to explain this (e.g. because of caste-based rural insurance networks), and then uses micro-data to test predictions of this theory and provide structural estimates, which are then used to generate policy counterfactuals (e.g. what might happen if formal insurance improves?). While this delivers areas for policy direction, it doesn’t tend to directly test how specific policies work in practice, with all the implementation issues that arise, constraints that may also matter that are outside of your model, and behavioral responses made by individuals (e.g. they might not trust that formal insurance will pay out). How do you think about the amount of effort researchers should be devoting to explaining why the world works the way it does, versus testing concrete policy efforts for changing this?
I agree that many studies, even if revelatory about the world, do not often provide a specific policy initiative precisely because they abstract from many details that may matter for policy implementation in order to highlight fundamental issues. But, I think that pitting the testing of policy interventions against understanding how the world works is a false dichotomy. Evaluating concrete policy efforts also can help us understand the world better. To figure out why an actual policy worked or did not work requires obtaining an understanding of what is going on. So a research puzzle could just as well be why implementing a specific policy that seemed like a good idea in theory did not work as why, say, democracy does not always lead to high growth rates. A good policy is one that has taken into account all the appropriate behavioral responses within the existing institutional constraints. Knowing why a specific policy worked or did not work is also much more valuable for implementing policies than just knowing what worked or not, especially if there is heterogeneity in institutions and constraints and capabilities across time and place.
3. One recent example where you were testing a policy being implemented was work with Mushfiq where you offered formal insurance to these informal risk-sharing networks using a randomized experiment. Did working on an RCT change any views you may have had about its usefulness as a tool, or surprise you in any way?
This study is an example of how implementing a specific policy can add to general knowledge, which was the aim. The only surprise to me was that what we saw in the field as a result of the experiment closely conformed to theory, with less fear that the result was spurious. I never doubted that RCT’s were useful as a tool, no more or less than my doubts about the usefulness of, say, logit or GMM. There are many banal and useless examples of studies using every specific method, and also great studies.
4. You are known for asking “where’s the economics” as a critique of a number of talks by economists. How do you think about the value added of economists relative to other social scientists and other researchers more generally, that is, what, in your mind, makes something an economics paper?
I view economics as a field in which we try to understand human behavior. So, any study that models human behavior and shows that that knowledge helps us understand what is going on is good economics. I also often say, “where’s the model?”, which is almost the same thing. And this is the same point as distinguishing between knowing what works and knowing why something works. Now, there is an inductive part of science that is important. So a study that uncovers a new fact (which could be a new correlation or an RCT showing that subsidized loans increase goat stocks) adds to empirical knowledge, but it is not specifically economics, even if the variables are money or income.
5. At your symposium, you gave a very nice speech in which you attributed part of your career to luck in being in the right places at the right time when changes were taking place in the way development was done, and part to the insights learned from moving around different institutions. What advice would you give to a young researcher in development today in terms of what to look for when deciding where to work, and when it is time to move?
Collaboration in research is fun and productive. Being around good researchers fosters better work. And having colleagues and collaborators with different skills and points of view is also productive. But, it is also true that it is independent thought that makes for new ideas. Being different is an essential ingredient for a good research career. The advice I give to young researchers is not to listen to advice from senior scholars, since every path to a productive career is different, and times change.
6. Amongst your many papers, is there a paper now that you feel is perhaps even more relevant now than when you wrote it, or that somehow slipped through the cracks in terms of more people knowing about it?
I’ve got a bunch of papers sitting on my desk (actually, literally) that were written over many years that I need to return to and get out because I think they are still relevant - on the consequences of local democratization, on marriage markets, on family immigration sponsorship, for example. But I keep getting interested in new things.
- Looking back at how development economics has evolved as a subject since you first started as a researcher, what is the change that most pleases you, and the change (if any) which you feel is a step in the wrong direction?
2. A lot of your work starts with a question about why we see certain behaviors in developing countries (e.g. why don’t we see more people migrating in India given the dispersion in wages?), puts together a theory to explain this (e.g. because of caste-based rural insurance networks), and then uses micro-data to test predictions of this theory and provide structural estimates, which are then used to generate policy counterfactuals (e.g. what might happen if formal insurance improves?). While this delivers areas for policy direction, it doesn’t tend to directly test how specific policies work in practice, with all the implementation issues that arise, constraints that may also matter that are outside of your model, and behavioral responses made by individuals (e.g. they might not trust that formal insurance will pay out). How do you think about the amount of effort researchers should be devoting to explaining why the world works the way it does, versus testing concrete policy efforts for changing this?
I agree that many studies, even if revelatory about the world, do not often provide a specific policy initiative precisely because they abstract from many details that may matter for policy implementation in order to highlight fundamental issues. But, I think that pitting the testing of policy interventions against understanding how the world works is a false dichotomy. Evaluating concrete policy efforts also can help us understand the world better. To figure out why an actual policy worked or did not work requires obtaining an understanding of what is going on. So a research puzzle could just as well be why implementing a specific policy that seemed like a good idea in theory did not work as why, say, democracy does not always lead to high growth rates. A good policy is one that has taken into account all the appropriate behavioral responses within the existing institutional constraints. Knowing why a specific policy worked or did not work is also much more valuable for implementing policies than just knowing what worked or not, especially if there is heterogeneity in institutions and constraints and capabilities across time and place.
3. One recent example where you were testing a policy being implemented was work with Mushfiq where you offered formal insurance to these informal risk-sharing networks using a randomized experiment. Did working on an RCT change any views you may have had about its usefulness as a tool, or surprise you in any way?
This study is an example of how implementing a specific policy can add to general knowledge, which was the aim. The only surprise to me was that what we saw in the field as a result of the experiment closely conformed to theory, with less fear that the result was spurious. I never doubted that RCT’s were useful as a tool, no more or less than my doubts about the usefulness of, say, logit or GMM. There are many banal and useless examples of studies using every specific method, and also great studies.
4. You are known for asking “where’s the economics” as a critique of a number of talks by economists. How do you think about the value added of economists relative to other social scientists and other researchers more generally, that is, what, in your mind, makes something an economics paper?
I view economics as a field in which we try to understand human behavior. So, any study that models human behavior and shows that that knowledge helps us understand what is going on is good economics. I also often say, “where’s the model?”, which is almost the same thing. And this is the same point as distinguishing between knowing what works and knowing why something works. Now, there is an inductive part of science that is important. So a study that uncovers a new fact (which could be a new correlation or an RCT showing that subsidized loans increase goat stocks) adds to empirical knowledge, but it is not specifically economics, even if the variables are money or income.
5. At your symposium, you gave a very nice speech in which you attributed part of your career to luck in being in the right places at the right time when changes were taking place in the way development was done, and part to the insights learned from moving around different institutions. What advice would you give to a young researcher in development today in terms of what to look for when deciding where to work, and when it is time to move?
Collaboration in research is fun and productive. Being around good researchers fosters better work. And having colleagues and collaborators with different skills and points of view is also productive. But, it is also true that it is independent thought that makes for new ideas. Being different is an essential ingredient for a good research career. The advice I give to young researchers is not to listen to advice from senior scholars, since every path to a productive career is different, and times change.
6. Amongst your many papers, is there a paper now that you feel is perhaps even more relevant now than when you wrote it, or that somehow slipped through the cracks in terms of more people knowing about it?
I’ve got a bunch of papers sitting on my desk (actually, literally) that were written over many years that I need to return to and get out because I think they are still relevant - on the consequences of local democratization, on marriage markets, on family immigration sponsorship, for example. But I keep getting interested in new things.
Join the Conversation