"You Can Build the Infrastructure from Zero": A Conversation about Digital Adoption in Emerging Economies


Tom Heintjes: Hello, and welcome back to another episode of the Economy Matters podcast. I'm Tom Heintjes, managing editor of the Atlanta Fed's Economy Matters magazine. Today, I have the privilege of sitting down with Federico Mandelman, a research economist and associate adviser here at the Atlanta Fed—and who has been a guest on the podcast before, although it's been a while. So welcome back, Federico.

Federico Mandelman: Hi, how are you, Tom?

Federico Mandelman, a Research Economist and Associate Adviser in the Research department of the Atlanta Fed, during the recording of a podcast episode
Photo by David Fine

Heintjes: I'm really pleased you're back on the podcast, because I've looked forward to this conversation for a while. Federico, you recently wrote a working paper titled "Digital Adoption, Automation, and Labor Markets in Developing and Emerging Economies." That's kind of a mouthful of a title, but I thought it was really an interesting paper and certainly worth having a conversation about. So thanks for sitting down with me. I wanted to start off by asking you: we hear a lot about the potential effects of digital technologies on advanced economies like the U.S. You know, "will autonomous vehicles reduce the need for human drivers?" and so forth—but we typically don't hear much about technology's impact on developing economies. What led you initially to look into this phenomenon?

Mandelman: Well, there is very little data on emerging economies—that is precisely the problem. There is very little effort in collecting information on automation and digital adoption in these economies. A priori, it seems that this is not relevant for these countries. The idea is that poor countries are not in the frontier of technology. They still rely on traditional methods and therefore are far from adopting breakthrough innovations. However, quite the opposite is true—developing countries can easily adapt to new technologies because they do not need to replace the infrastructure that's already been built with existing state-of-the-art technology. Since they can build from zero, they can easily bypass previous technological stages, and that simplifies the technology adoption. Just to give you an idea, I was in China ten years ago and no one used credit cards.

Heintjes: Right—the China of ten years ago is very different from the China of today, I would imagine.

Mandelman: Yes, exactly. So you had to pay for everything with cash, no credit cards. And I talked to a friend that went recently to China, and we were having a conversation and he said, "Yes, it's impossible to use a credit card in China." And I said, "Of course—they only use cash." And he said, "No one uses cash. Everyone uses an app in their telephone. Whenever they want to make a payment, they show the QR code and they make a payment with that app." So really, they use cell phone apps to make payments.

Heintjes: So a huge change in a decade.

Mandelman: Yes. So essentially this is like leapfrogging, right? You jump over technologies more easily when you don't need to change the infrastructure from the existing technology.

Heintjes: Right. And so you also see that in not just China, but other developing economies?

Mandelman: Yes, exactly that. To give you an example, going back to these QR codes: in gas stations in Argentina, only cash was used before. But I went there this summer, and again people were using apps and displaying these QR codes to make payments. There is no infrastructure to take credit cards in gas stations like here in the U.S. But if you have a cell phone, like most people do, they can use an application to make electronic payments. I think that will be more complicated probably to have that technology in the U.S., probably because people are already used to using credit cards. So that's the idea. So these countries are not that different. They don't have technology, but probably that's an advantage that they have in the new way of telecommunications technologies that we observe.

Heintjes: Right. Well, your paper—which I should note was coauthored by Alan Finkelstein Shapiro of Tufts University—refers to recent studies suggesting that more than 60 percent—that's six-zero percent—of jobs in developing and emerging economies are likely to be susceptible to automation. Sixty percent is quite a staggering number. If we heard that data point applied to the U.S. labor force, it would be front page news here.

Mandelman: Yes, it's true that in developed countries you have many of these so-called nonroutine cognitive occupations. These occupations require creativity, management skills, or complex problem-solving skills—like computer programmers in Silicon Valley, doctors, lawyers, professionals in research and development, managers, etc. In developing countries, you don't have many of these jobs. In developing countries, you have a much higher share of occupations that are in repetitive jobs that execute a routine task, which are susceptible to be replaced by computers or robots at a larger scale.

Heintjes: Right. Well, Federico, one important distinction that your paper makes is that self-employment has a much higher rate in developing and emerging economies than in developed economies. Just how different are rates of self-employment in developing and emerging economies compared to economies like the U.S. or other advanced economies?

Mandelman: Most importantly, and first of all, I think I have to highlight that self-employment in developing countries is very different than in developed countries.

Heintjes: How is that?

Mandelman: In the U.S., when we think about self-employed individuals, we think about skilled people: creative, very energetic—that they launched their own companies. In developing countries, it is the opposite. In developing countries, where you have very limited or nonexistent safety nets, the urban unemployed will seek subsistence in these occupations. So the self-employed in developing countries have typically very little or no capital, and have to work on their own—that is, they don't hire other people. And usually when we think of someone that is self-employed in a developing country, we are thinking not of an entrepreneur but quite likely a street vendor. So self-employment is a very different phenomenon. Salaried employment, on the other hand, which seems to be the norm in rich countries, is very scarce in developing countries. So the lack of business creation in these economies implies that about 45 percent of the labor force in these countries work on their own—that's very different than in the U.S.

Heintjes: I see. Well, you know, I don't want to get us on a long tangent about world history, but there must be historical or sociological factors behind the greater share of self-employment in developing and emerging economies. For example, your paper mentions the greater barrier to credit access in developing and emerging economies that might impede firm creation. Is that sort of thing a factor at work here?

Mandelman: Yes. While many other factors are in play, the greater extent of high self-employment in these economies I would say is the other side of the coin of lack of business creation. It is usually very hard to be sure of the actual factors behind the lack of firm creation and salaried employment, as the institutional setting is key in those countries. We can mention, however, that business creation requires some basic conditions that are in short supply in these countries. So for instance: lack of basic infrastructure, from highways to internet connections; bank credit; human capital; an imperfect rule of law; or bothersome red tape—all these factors essentially contribute to a lack of firm creation in these countries.

Heintjes: Those are very important factors. Well, I thought it was interesting that in your paper, your models' outcomes show that as the creation of salaried firms increases, labor is—I guess you can say it was reallocated away from self-employment and into salaried employment, but that leaves unemployment basically unchanged, doesn't it?

Mandelman: Yes, essentially that is the main empirical finding of the paper. Controlling for many other factors, we find that countries that adopt more of the information and communications technology tend to create more businesses, and by doing so, they pull people out of self-employment into paid, salaried jobs—which is quite different from what typically happens in developing countries, where technology adoption and automation are usually associated with displacement of workers in routine occupations like manufacturing, or administrative workers.

Heintjes: I see. I don't know if your paper went into the impact on income so much, but is there an effect on the incomes of workers when this reallocation happens?

Mandelman: No, we don't go over that aspect in the paper.

Heintjes: Okay, I just wanted to make clear that you didn't really delve into that aspect of it. So Federico, is greater technology adoption alone a sufficiently powerful force to generate a move away from self-employment and into salaried employment, or are there other factors at work?

Mandelman: One of the things that I didn't mention before is that advancements in telecommunications help to overcome the lack of infrastructure by bringing people together and creating business opportunities. So for instance, rural and geographically dispersed towns are now better interconnected, thanks to ecommerce platforms. So to give you another example: in Argentina, you have small farms from rural towns that just have started to sell their products, like food produce, on websites like eBay. Not only that, they can also buy and receive by mail intermediate goods that they need for production, without the need to go to the capital city. And most interestingly, we see that in many developing countries, by tracking their operations and their cash flows in these e-commerce platforms, they can easily get access to credit through the fintechs' subsidiaries that are associated with these companies, as it is the norm in China.

So finally, another factor that I wanted to highlight is that mobile banking—that is, the possibility of using your phone to make financial transactions—explains alone at least 20 percent in the decline of the number of the unbanked in the world between the years 2011 and 2014. So digital adoption appears to be a fundamental factor promoting new business creation, by lowering the barriers to entry and creating salaried jobs. So going back to your question…I don't know if you can remind me because I got distracted with this.

Heintjes: Well, I wanted to ask you if greater technology adoption alone—

Mandelman: Ah, yes, exactly. So we saw that if we focus solely on existing incumbent firms that adopt the new technology, job creation is not powerful enough. So if established firms adopt technology, they become more productive for sure, so they can hire more workers. But on the other hand, we know that machines and computers will replace some of the existing workers as well.

Heintjes: Right; the automation that we've been talking about.

Mandelman: Exactly. The idea is that as existing firms become more productive, they hire more people; but on the other hand, the same machines and computers that they now use in production can replace some of their workers. So in net, we find that net job creation by established firms, by the already established firms, is positive but near zero as a result. So the key, once again, is business creation—business creation facilitated by technological adoption that lowers the barriers to firm entry by bringing businesses together through e-commerce, promoting credit with mobile banking, or eliminating red tape by governments—seems to be an important factor.

Heintjes: Right. There's been a lot of research into the effects of technology and automation on advanced economies, but not so much on developing and emerging economies—and your research paper is one of the few I've seen to examine this relationship, which is why I found it interesting and wanted to talk to you today. Why hasn't this been studied more in general, do you think? Is it a data availability problem, or what's at work here?

Mandelman: Yes, exactly. As I said before, the main problem is the lack of data. Only very recently multilateral and developmental organizations, like the World Bank, started to make some effort in collecting data on the extent of digital adoption and automation technologies in these economies. And we, as economists, have a natural tendency to ignore the things that we cannot be sure of, so that would be probably my explanation.

Heintjes: Well, as we've just noted, this is a relatively underexplored area. So how did you and your coauthor set about constructing robust models and all the things you need to do this kind of analysis? What were the biggest challenges to your modeling?

Mandelman: So that's our main problem. Full disclosure here: we have a hypothesis, and we've built a model to test this hypothesis. But without reliable and sufficiently descriptive data, the robustness of our quantitative analysis can be fairly put into question. Well, I will say—some aspects of it. I need to sell the paper here.

Heintjes: Yes.

Mandelman: But some aspects of our work can be put into question because the data we have is not very rich. It's not very detailed, it's not sufficiently descriptive or reliable. For instance, we do not have firm-level data on technology adoption, we just have aggregate indicators—but no firm-level indicators. We don't have a time series to track the evolution of technology adoption over time, so these are important constraints that we encounter in the analysis. But you know, it sometimes happens that theory helps by making evident that more data is needed. So with that idea, our research probably will contribute as a grain of salt in that direction, showing that this might be an interesting topic but the quality of the data is not there. So you need to build more data.

Heintjes: Well, I do respect your full disclosure—that's very good. Was there anything you found that surprised you in your research, anything that maybe challenged your own assumptions about the relationship between technology and the labor force, or the larger macroeconomy?

Mandelman: Yes—critically, the role of business creation. I did research on automation before, but solely looking at the U.S. So when I started looking at this issue, I completely oversaw this important aspect of firm creation or firm entry, which is of course critical in emerging and developing economies. So I am thankful to my coauthor, Alan Finkelstein Shapiro, who enlightened me in this regard.

Heintjes: Well, I have one more question for you. What would you say are the implications of your research's findings? What would you like policymakers, or even anyone reading your paper, to take away from your work?

Mandelman: Well, let me highlight two points. The first one is that automation and digital adoption may be important in developing countries—essentially that. They don't need to be in the technological frontier to adopt the new technologies. Quite the opposite: you can build the infrastructure from zero without the need to revamp the existing infrastructure. In other words, it is easier to adopt modern systems when the technologies have not been adopted before. Okay? So you start from zero—fresh start. And the second point that we highlight in our research is that in emerging and developing economies, automation and digital adoption can be potentially a strong factor in lowering the cost of entry for new firms. By facilitating business creation, they may generate the much-needed salaried jobs that these economies require. So that would be the two points that I would like to highlight in the end.

Heintjes: Got it. Well, let's just call this, then, the first of a series of conversations we'll have about this as you continue to research this area and have new findings. So you'll be back on to talk about this in the future, I hope.

Mandelman: Yes, sure—it's a pleasure.

Heintjes: Well, thank you. Unfortunately, we are out of time, Federico, but I want to thank you for your time and for sharing your insights about your research with us. It's always great to talk to you, and I hope you will be back on the podcast soon—about this or other research you're working on. And that brings us to the end of another episode of the Economy Matters podcast. I'm Tom Heintjes, managing editor of the Atlanta Fed's Economy Matters magazine, and I want to note that we'll have a link to Federico's paper on our site at frbatlanta.org. Thanks for spending some time with us today, and I hope you'll come back next month for another episode.