As part of our interview series, we ask renowned experts in the field about the future of research in development economics, and for their advice for young researchers. For our third interview, we got the opportunity to talk to Francis Annan.
Francis Annan is an Assistant Professor at UC Berkeley’s department of Agricultural and Resource Economics. Francis Annan’s research centers on development economics and microeconomic issues, with a focus on digital financial markets, insurance, and firms in sub-Saharan Africa, India, and the United States.
This post makes reference to a talk by Francis Annan where he presented one of his latest papers: Misconduct and Reputation under Imperfect Information
ETRM: So I think the first question is, how would you define development economics at frontier right now? For graduate students, what should they think when someone says, “development economics”?
FA: I think it’s a very great question. As a backdrop, I think there are a lot of misperceptions about what development economics is and I believe that has changed over time. From where I sit, when I hear development economics I think: taking theory seriously, taking empirics seriously, and taking policy seriously. It is really about finding a better way to do research that matters for real people in developing countries, but with a strong understanding of the theory, the policy implications and even the commercial implications of the research that we do. This obviously requires a lot of active partnerships in the field and in research that really matters for real people. That’s what comes to mind.
ETRM: You basically used the name of our blog in your answer. We love it! Our next question would be regarding mobile money. After we take into account misconduct that you raise in your research, do you think mobile money can address issues related to credit constraints in developing economies?
FA: Very fair point, I think mobile money has the ability to change credit markets, if well designed, if not, it is going to be a problem. We have seen all these recent papers talking about digital credit, for example, in Malawi, where consumers are victimized precisely because the credit terms are so opaque. Though these frictions are also found in traditional markets, and not by any means limited to just digital. But I guess the implications of having opaque contracts, which are digital in this case, are actually more severe precisely because consumers are less sophisticated. And these are the markets where we think mobile money has the ability to improve welfare or reduce poverty as a whole. So if you look at a population where consumers are poorly informed, there is a lot of shrouding in contracts and prices or transactions, and the credit terms are not favorable in fundamental ways to the users, then certainly it is hard to imagine what the first best will be. So I think at the very least in the second-best world, mobile money has the ability to actually change the dynamics of credit provision and credit acquisition in these markets. We are learning a lot, and research is influencing the design of digital contracts. Consumers are also learning a lot, so over time, we are going to overcome some of these informational frictions, and consumers will learn to self-select into the right sort of contracts, and mobile money will change credit markets and influence liquidity issues for households and businesses. The easiest thing to mention here is that through digital transactions, contracts that were not feasible in the past precisely because of lack of ability to infer who is likely to default versus not, is overcome. I think we get more efficient with contracts because unlike, say, the traditional microfinance environment, here, the operational cost of deploying contracts on the supply side is very cheap. Compared to microfinance where the operational costs alone account for about, I don’t know 30% of the costs. So, the long and short answer is that mobile money does have the ability, and the potential to inform credit markets to increase credit access, if well designed.
ETRM (V. Sharma): I completely agree with that. And, you know, it is so contextually varied. For example, my research was on mobile phones in India where I ran an RCT across 108 villages. And one thing that I would like to study in the future is the differences in mobile money adoption across countries, because in India, for example, there is no cost for using mobile phone payments, which is very different from what you described in Ghana.
FA: I want to push back a little bit more on adoption, and here’s why: I think the more I do research in digital financial services, the more I get really excited about the amazing set of questions that I think people should be thinking about more broadly, including mobile phones and other technologies. I think there is a desire to adopt digital finance but ask yourself, what quality of services are we providing to consumers? I’ll tell you in the market that I presented today, about 30% of the time you show up at a retail point, and the guy tells you “I don’t have enough liquidity, you have to go and come back tomorrow”; simply imagine going to the bank and they tell you “we don’t have money today, come back tomorrow”. So I think as we reflect on adoption in scale, we should reflect on intensive margin issues and how to promote quality adoption. Understanding issues about market structure, overpricing, privacy, intensive margin demand effects, transparency, you name it! I think that’s where the literature should be headed. We’re going to move past just adoption, and we’re going to think of how to promote quality adoption and services. And in fact, we’re going to think of how to design new markets and enter into “virgin” markets in communities where for example, formal financial services have never existed to begin with.
ETRM: Just a quick follow-up. In this scenario you are describing, the telecommunication providers are entering as potential big players in the finance sector, right? So how do you think that is going affect not just the finance sector but public service delivery in general?
AF: I like to start with analogs because it’s always useful to reflect on things that are close to you. I grew up in Ghana. I used to live close to a neighbor who was a really rich guy, like rich in the sense that he did public and government contracts. In many developing countries where you see procurements, government outsource a lot of different projects. One of the biggest frictions in developing countries, in India in particular, is that if you ask 10 firms, “Are you willing to do government contracts?” about 50% will say “no”. Why? Because payments are delayed. In the Ghanaian context, when I went back home, and learned this man that I used to know who had all the contracts and employees had actually committed suicide. What happened is, the government contract money was not coming and lenders were going after him. He couldn’t pay them and he couldn’t pay his employees and other input suppliers. Anyways, that got me thinking, how do we leverage digitization to deliver public services. So I have an ongoing project called “digitizing bureaucracy” (joint with Apoorv Gupta at Dartmouth College). The key point here is, how can we leverage technology to facilitate the processing of public contracts for businesses by eliminating potential coordination failures and payment delays? And what are the implications of that on firm productivity? Not only that, but what are the implications on public sector productivity? If you eliminate all these frictions, how do you crowd in high quality firms into procurement? You know, we have developing countries where roads, including other basic infrastructure, don’t last precisely because the person who is building the road knows the money is not going to show up in time. So they invest in poor quality inputs, the wrong technology, and delivers poor quality of development projects. The bottom line is that technology or digitization or automation has an amazing potential to promote the delivery of public services in particular by eliminating bureaucratic delays. We can then study how this impacts firms with procurements that have a lot or little people to pay. To the extent that “digitizing bureaucracy” minimizes the contact rate between public officials and firms, this also has the ability to reduce bribery.
ETRM: What advice do you have for graduate students who want to do primary data collection? And given there are all of these policy interventions by governments or research institutes, how do we take into account the contamination from existing interventions? Is there a way to avoid this?
FA: I think this is a very important point, I think we have made a lot of progress in this area. My usual way to start a project is to look at, for example, the AEA registry if it’s an experiment. So for example, if you want to do a project in Ghana, go ahead and enter Ghana into the registry to get a sense of what projects are ongoing. It’s important to make sure you don’t repeat already existing research, but also you can prevent overlaying one intervention over an already existing one. That’s fundamental! That’s why I promote a lot of this transparency in initiatives in research precisely because it helps us overcome some of these frictions and avoid wasting public funds or donor money. Another way to do this is by including in your baseline survey questions about whether participants are part of any other programs. This can be important because you may know which areas have ongoing interventions, but you won’t know who the beneficiaries are. So, it makes sense to include questions that allow you to screen for other programs. Then you can make a judgment on whether this is an environment where you should intervene or not, and if so, how you want to do it. These approaches are imperfect, but I think they are non-trivial.
ETRM: Our last question is related to our previous one: what advice do you have for graduate students in development economics today, if they, for example, should try to pursue an RCT or not?
FA: So, going back to the way I see development economics: theory is important, empirics are important, and partnerships are important. I’m very hopeful about research in development economics, and therefore I support and promote students in development. They should see this as a very interesting environment, precisely because they are changing lives in some fundamental way. Now, I think, in terms of advice, we should make theory very important in the projects that we do. Also, as a graduate student, really ask yourself: what kind of interventions can I do that are low risk in terms of my ability to do an endline? What I have found is that information interventions are very easy to run and low cost. While you may not achieve large outcomes, you can observe interesting intermediate outcomes such as changing beliefs. You should also look at what I call “at-risk environments.” For example, consider those involved in illegal gold mining activities in Ghana. Now, if you do interventions about illegal gold mining, you are never going to expect the results to change in the next year or two. So the question is, how do you find the most at-risk communities? In this example, the next places that illegal mining is going to enter. Then focus on those markets and you’d expect to see some short-term impact. So I think finding a better context is very important as a PhD student, along with the type of interventions you do, whether they are short lived or long. You can always deliver if you think about it very carefully. And do not underestimate how long it takes to run an experiment. This is so so important! It can take a long time, not only just because the field work, but in terms of securing funds or research money. In fact, COVID actually taught those of us who had projects in the field a lot. I think you should do excellent baseline work. For example, the people during the pandemic who had phone numbers of all their respondents were still able to deliver their research. Those that were expost “not smart enough” to have this information really had to struggle. I don’t love to see graduate students in that soup where you can never make progress because something global happened. So again, to maximize the use of the baseline, make sure you establish the right contacts, something that promotes easy access to the respondents in the field. And so that’s my general advice. I think it’s doable to do experiments within the span of your PhD, just start early, plan well, maximize the baseline and get a lot of support from your advisors, especially when it comes to funding because this can be a non-trivial step.
