The Boys from Belarus
Two young economists, best friends since high school, are exploring—and expanding—the frontiers of economics.
Published September 1, 2006 | September 2006 issue
It is the birthplace of painter Marc Chagall, and makes a tenuous claim to Fyodor Dostoyevsky, but otherwise the small city of Vitebsk (see map below), about 360 miles west of Moscow, is little known to the outside world. Among economists, however, that will soon change, for the Belarussian town is also home to two of the profession's rising stars, Mikhail Golosov and Aleh Tsyvinski.
Golosov and Tsyvinski attended the same high school in Vitebsk, the same college in Minsk and the same graduate school in Minnesota. While grad students, they both worked at the Minneapolis Fed and the International Monetary Fund. And they now are assistant professors at two of the world's top economics departments, at the Massachusetts Institute of Technology and Harvard University, respectively. [In 2008, Tsyvinski accepted a full professorship at Yale University. A year later, Golosov joined him at Yale, also as a full professor.]
Not incidentally, Mike and Oleg—as they're known—are also best friends. Meeting in high school was luck, but later parallels have been, in Mike's phrase, a series of "coordinated decisions." And when Mike got married earlier this year, there was no question who would be his best man.
So it isn't surprising (even in a discipline that extols the virtues of competition) that the boys from Vitebsk have found huge professional payoff from close collaboration. Over the course of their nascent careers, the two have worked jointly on almost all their research—they've each produced 14 published articles or working papers; all but two were written together.
Academics often co-author papers, of course, but few do so with such consistency and fecundity. These partnerships can be difficult, as scholars struggle over ideas, methods, workload and ego; seldom do they have the fluid style developed by Golosov and Tsyvinski, who motivate each other, capitalize on comparative strengths and communicate smoothly.
Still, what makes their collaboration truly noteworthy is the quality of their work. Over the phone, via e-mail and on the days Tsyvinski drives across Cambridge to spend the day at MIT ("I've got a car; Michael doesn't"), the two are generating a stream of research that is helping economists rethink optimal policies for taxation and social insurance.
"In many ways, they're the future of macroeconomics and macroeconomic theory," observes Daron Acemoglu, an award-winning MIT economist. "In terms of their curiosity, skills and ideas, what we're seeing of them now is just the tip of the iceberg." V. V. Chari, their dissertation adviser at the University of Minnesota, is similarly impressed. "They're going to leave a very big mark on the profession," he says. "I think they'll be pushing forward the frontier of public finance and macroeconomics much, much further."
But to understand the nature of their research, it helps first to recall the work of two other economists, at another Cambridge.
British philosopher/mathematician/economist Frank Ramsey was a wunderkind of an earlier era, a professor at King's College, Cambridge. He died just short of his 27th birthday, but not before making major contributions to each discipline, and in economics, Ramsey developed seminal insights into optimal taxation and saving.
"How much of its income should a nation save?" Ramsey asked—and answered—in a landmark 1928 article. In an earlier paper, he'd tackled a related problem: If a nation wishes to raise a certain amount of revenue through proportionate taxes on different commodities, "how should these rates be adjusted?"
Both questions are central to economic growth and efficiency, and Ramsey's solutions were elegant, even poetic. Regarding saving, Ramsey introduced the term "bliss" to denote the highest possible rate of enjoyment, and by mathematically modeling the inevitable temporal trade-off—"the more we save the sooner we shall reach bliss, but the less enjoyment we shall have now"—he proved that the "rate of saving multiplied by marginal utility of consumption should always equal bliss minus actual rate of utility enjoyed."
Regarding taxes, Ramsey showed that optimal taxation depends highly on the elasticity of demand for the goods being taxed. "[T]axes should be such as to diminish in the same proportion the production of each commodity taxed," he determined. The term "Ramsey pricing" is now used by economists to indicate optimal pricing according to demand elasticity.
Ramsey's work had a major influence on subsequent economic research into saving and tax theory. But it was based on an assumption that, in some sense, crippled its power. "I propose to neglect altogether questions of distribution," he wrote in the taxation paper. "We also ignore altogether distributional considerations," he wrote in the saving paper. By avoiding issues of redistribution—a key purpose of taxation and a factor in national saving behavior—Ramsey theory told only part of the story.
Roughly 40 years later, another British economist, James Mirrlees, now at Trinity College, Cambridge, but then at Oxford, delved into "the theory of optimum income taxation." Unlike Ramsey, Mirrlees was centrally interested in questions of distribution. His seminal 1971 paper on the topic focused on "the labour-discouraging effects of the tax ... relative to the redistributive benefits."
Nowadays economists use the terms "efficiency-equity" or "incentive-insurance" to describe trade-off. How can a nation design a tax system that will provide adequate social insurance—protection against the shocks of unemployment, disability or old age—while maintaining incentives to work?
Mirrlees recognized that any such system must cope with asymmetric information: A citizen knows more about his or her ability to work than does the government and can deceive the government to take advantage of social insurance policies. Given that asymmetry, how do policymakers design a tax system that provides optimal incentives?
Mirrlees' solution involved designing a system that was "incentive compatible," meaning that its rules must give workers an incentive to reveal their true abilities to work. The "revelation principle," formulated mathematically, provided the answer to the dilemma.
Mirrlees' work, recognized in 1996 with the Nobel Prize in economics, thus addressed the issue that Ramsey assumed away. But Mirrlees, too, made a huge simplifying assumption. "Intertemporal problems are ignored," he wrote. "The economy discussed below is timeless. Thus the effects of taxation on saving are ignored."
In essence, then, Mirrlees and Ramsey provided diametric perspectives—both profoundly valuable, both incomplete. Macroeconomists have used Ramsey growth models to understand optimal taxation. And public finance economists have used Mirrlees' insights to better grasp incentive-insurance trade-offs. But each approach lacked the essential truths of the other.
A dynamic duo
This, then, is the problem that Golosov and Tsyvinski have been tackling over the past several years, merging Ramsey's dynamic approach to optimal taxation with Mirrlees' static model of taxation under asymmetric information.
"Most of the work in public finance is essentially static in nature," notes Tsyvinski. "It's about the incentives that are intratemporal today, rather than about dynamic incentives that are central to Ramsey. So we're trying to think about dynamic issues in Mirrlees, motivated by the macroeconomic literature of Ramsey approach to derive the implications of the social insurance for distribution, et cetera. In some sense, our approach unifies both taxation and social insurance."
Golosov gives an example. "A general Ramsey approach shows that pretty much under no circumstances do you want to distort savings, so you want to use labor taxes rather than capital taxes," he explains. "But once you introduce these other features from Mirrlees, the answer becomes much less clear. It actually shows that you may want to distort savings."
In a series of papers published or forthcoming in some of the profession's most prestigious journals—the Review of Economic Studies, the Quarterly Journal of Economics and the Journal of Political Economy—Golosov and Tsyvinski explore various angles of this theme.
In a 2003 article co-authored with Narayana Kocherlakota, then at Stanford University and the Minneapolis Fed, they re-examine two central conclusions of the Ramsey and Mirrlees literature (respectively, capital should not be taxed and consumption goods should be taxed equally) in a broader context: dynamic economies, unobservable worker skills that evolve randomly over time and a tax authority that can choose less rigid types of taxing schemes than economists usually consider.
"This general class of environments is technically challenging," write the authors, who then meet the challenge and find that "the uniform commodity taxation theorem is generally valid, but the zero capital income taxation theorem is generally not."
In a more recent article, Golosov and Tsyvinski examine the role of government in designing and providing social insurance. Economists had shown previously that when consumption is publicly observable, private markets can provide insurance against many uncertainties, and government insurance programs simply "crowd out" private insurance.
But Golosov and Tsyvinski determine that when individual consumption is not publicly observable, the results change. While private markets can still provide a significant amount of insurance, it is not entirely efficient. "There is still a role for welfare-improving ... taxes or subsidies imposed by the government," they write. And they provide a solution: a linear savings tax that corrects an externality problem associated with private insurers and asymmetric information.
Theory and data
A third piece examines disability insurance, noting that it's one of the largest social insurance programs in the United States, benefiting more than 6 million people and costing $61 billion in 2001, far more than was spent on unemployment insurance or food stamps that year. But a disability is difficult to verify. An individual could pretend to have severe back pain simply to collect insurance. How much of the $61 billion is thus misspent?
Mirrlees had also studied this case of asymmetric information—where the worker knows more than the government—but when he did so (with MIT's Peter Diamond) he concluded that a linear savings tax could ensure a solution. But Golosov and Tsyvinski prove that Diamond/Mirrlees' system actually "does not implement the optimum."
The Belarussians then propose their own solution: an asset test in which a person receives a disability payment only if his assets fall below a certain level. In a dynamic setting, they find, this simple rule works optimally. The asset test eliminates the incentive to cheat the system (thus creating incentive compatibility). And they then bring data to their model, finding that implementing an asset test would provide significant benefit to the economy.
Kocherlakota, now chair of the University of Minnesota's economics department, points out how innovative this work is, not only for its insight into disability insurance, but for its reconciliation of two distinct areas of economics: public finance and macroeconomics. Both fields have a keen interest in tax policy, he notes, but in the past 20 years or so, "macroeconomists have been much more, I would say, theoretical, in their approach, and public finance people have been typically much more empirical."
Golosov and Tsyvinski blend the approaches. They provide a theoretical framework but also an empirical perspective. "Their JPE paper is very important in this because it shows it's not just pure theorizing in a vacuum," says Kocherlakota. "They show that you can actually use these new methods and the data together to answer questions that people view as first-order in importance. How do you structure disability insurance programs when we're faced with enormous incentive problems? They provide a real sharp way to go after that."
Not so similar
The close friendship and parallel lives of Golosov and Tsyvinski might suggest similar people. Not so. Golosov, 29, is tall, thin and quiet, with a cautious demeanor. Tsyvinski, 28, is built like an oak, barrel-chested with a round face and infectious smile. He talks rapidly, jumping from topic to topic, while Golosov thinks at length before answering a question and carefully rephrases if the initial response seems unclear. Tsyvinski lives on the 31st floor of a gleaming skyscraper in the heart of Boston's financial district; Golosov, newly married, lives in Somerville, a relatively placid "suburb."
"We have very different personalities," acknowledges Golosov, during one of his frequent visits to the Minneapolis Fed. "I'm kind of a quieter type. Oleg is more energetic, more talkative, more aggressive, more organized."
Tsyvinski, interviewed in his office at Harvard, agrees. "I'm more perky, while Michael is the epitome of stability and calculation and rationality." Asked about the large bottle of Tums next to his computer, Tsyvinski explains, "I drink a lot of coffee, and it upsets my stomach." Less coffee, obviously, isn't an option.
They harmonize well. "We have different styles which, I think, complement very nicely each other," notes Golosov. "Michael is good at the things that I'm not," agrees Tsyvinski. "So, for example, Michael can sit down and work on one line of the proof for a week, like 10 hours a day. I cannot concentrate for that long. But you know, I think I'm mildly ADHD, and because my mind races through so many things, I often can come up with different angles to look at the problem."
As Kocherlakota, also a dissertation adviser to Tsyvinski, observes, "I think Oleg operates a little more by short flashes of intuition, and Mike is a little more a guy who's real careful, pushing to get every detail right. ... That's not to say Mike doesn't have great intuition. He does. Or that Oleg isn't a careful guy, which he is. But in terms of comparative advantage, I would say that's what they bring to the table."
Working as a team
The two discuss economics incessantly, both on and off the job, and frequently come up with an idea that they realize might be turned into a research paper. Tsyvinski generally tries to come up with a simple example that helps pare the problem down to its essence, and then develops the intuition a bit further. "That's actually my strength," says Tsyvinski, who usually takes initiative at this stage in talking it through with other economists.
At the same time, Golosov works on elaborating the mathematical structure of the problem, trying to understand which forces are important for the result. He'll often take a special case and move it to a general statement. And if the paper involves a computational aspect—an empirical evaluation of the theoretical model, as in the disability insurance paper—Tsyvinski likely takes on that task.
"Most of the paper's main results are often derived in a month or two months, Tsyvinski explains, "but the generalization and writing the paper and working out the kinks, that takes a year or a year-and-a-half."
Observes Acemoglu, who has worked with them both, "There's a real personal bond. ... It is a very synergistic relationship."
As it happens, the relationship began not in Belarus, but in Holland, where the two were assigned to the same host family during a student exchange program. The friendship continued back home, and while neither had dreams of becoming an economist, both decided that attending the Belarus State Economic University (BSEU) would be their next step.
"When I was finishing high school, in 1994, that was a very interesting time," recalls Golosov. "Belarus was moving from the communist system to the market economy. But literally nobody had any idea how a market economy was supposed to work."
"[BSEU] was supposed to be the best economics university in Belarus," he notes, but they learned very little about modern economic theory. Indeed, the year before they entered, "the core economics course was all based on Karl Marx's work." But at that stage, both were more interested in learning business than economics, and that was the school's specialty.
They studied finance, accounting, some business law and "very trivial mathematics," says Golosov. The new nation had just launched its credit card system, and in their freshman year both Golosov and Tsyvinski worked for the Belarus Interbank Settlement Center. In the last year of college, Tsyvinski worked in a "boutique investment bank," he says, as a currency trader.
Both sought opportunity outside of Belarus. Tsyvinski won a fellowship to study abroad and spent his junior year at the University of Missouri-Columbia. Still interested in business, he found that scholarships for MBA programs weren't available, but funding for doctoral programs in economics was. "Minnesota kindly took me," he says.
Golosov entered the economics program at the University of British Columbia in Vancouver, intending to get a master's degree. "I didn't necessarily really want to go into academia," he explains, "but I really wanted to have some Western education." It was there that Golosov realized how little economics he really knew. "It was shocking how big a gap existed."
There was a math gap as well. Once he decided to get an economics Ph.D., he realized he needed better math background, but he knew too little, he says, to judge the level of difficulty of different classes. "Essentially by mistake, I took one of the more advanced math courses, and by the time I realized it was way too advanced, it was kind of too late to back out. ... It was a very, very, very difficult experience. But by the end of that year, I felt I was pretty much on top of the math."
At least picking a school wasn't difficult. "I was accepted to a bunch of places, but Minnesota was a very good program, and Oleg was already there," Golosov explains. "I didn't know a single other person in the U.S., so going where your best friend lives seemed like a very good choice."
The two have high praise for the University's economics program and its partnership with the Federal Reserve. "This economics community, the Fed and Minnesota," Golosov begins, "it's impossible to overemphasize how important they were for me. I'm very, very grateful to the many people here with whom I worked."
Says Tsyvinski, "I'm looking at the graduate program here [at Harvard], and I'm trying to, in some sense, replicate what I had at Minnesota ... [where] they move you from somebody who's never seen many of the issues to the frontier of economic research. They give you enough tools so you can actually do it on your own." And at the Fed, he says, a grad student is "treated as a junior colleague, or as a faculty member. So you sit at the same table during lunch, you go to all the seminars."
It's there that barriers drop and ideas flow. "Every day you have lunch with some faculty or some economist here at the bank," says Golosov. "[J]ust naturally you'll start talking to him about your research and you learn a lot. All these barriers which typically exist in grad programs, they disappear here."
To date, the greater part of Golosov and Tsyvinski's work has focused on blending the insights of Ramsey and Mirrlees to better understand optimal tax and insurance policies. They call this the "New Dynamic Public Finance" in a paper written with MIT economist Iván Werning, to draw the contrast with Mirrlees' static perspective. But their research—both past and future—explores other directions as well.
One of Golosov's first papers, co-authored with Nobel Laureate Robert Lucas, examined—and found cracks in—the microeconomic foundation of the Phillips curve hypothesis, the idea that policymakers can exploit a trade-off between inflation and unemployment. With his other U of M dissertation adviser, Larry Jones, and fellow grad student Michele Tertilt, now at Stanford University, Golosov has also built models to explain how concepts of Pareto efficiency apply to theories of population growth.
And Tsyvinski recently co-authored a piece, forthcoming in the American Economic Review, with Christian Hellwig of UCLA and the late Arijit Mukherji of the University of Minnesota on the causes of currency crises, rejecting the theory that such crises result from imperfect information and lack of coordination among global players.
Along with V. V. Chari, Golosov and Tsyvinski have examined the impact of capital gains taxes on entrepreneurial choice, finding that to avoid taxes some entrepreneurs manage the businesses they create rather than starting new ones. The three are beginning work on a separate piece having to do with patents and copyright laws. "In spirit and execution," says Chari, "it is very different from their other work."
Still another line of research examines the political economy of tax policies. Conventional models assume governments that are all-knowing and benevolent. By introducing asymmetric information, Mirrlees eliminated the first unrealistic assumption. But what of the second? "Recently," says Tsyvinski, "we have gotten quite interested, working with Daron Acemoglu at MIT, on how to design policy when there's no fictitious benevolent social planner, when the model has real politicians."
Markets are good at some things; governments at others, he explains. Governments may have an advantage in providing social insurance or redistribution, but they are prone to "expropriating stuff [and] misusing information." Competitive markets may have less insurance or fewer benefits of a centralized mechanism, but they alleviate problems with misusing information or misappropriation of resources. "This is something I'm very excited about, bringing political economy together with optimal taxation," says Tsyvinski.
Those who walked into the June 23rd afternoon lecture at the Econometric Society's summer conference in Minneapolis might have been perplexed, for a minute, about the man who stood before them. Solemn and thin to the point of frail, he didn't resemble the burly, buoyant Oleg Tsyvinski who was scheduled to present "Tax Policies of the Leviathan."
"If you think about standard theory of public finance," began the speaker, "the first assumption you make is that the government is this fictitious social planner—totally selfless, totally benevolent, caring only about the welfare of the citizens. We want to move away from that assumption ..."
It was Mike Golosov, of course, who gave the talk that day, but in some sense it didn't matter whether Mike or Oleg appeared, for the message was consistent. The two work seamlessly, symbiotically.
And for the time being, they plan to do so here in the United States. "The U.S. is the center of modern economics," observes Golosov, "and I do not see myself moving out in the near future. I will be doing a lot of mixture of macroeconomics and public economics, I think. But it's really difficult to say, especially on these longer horizons. You just never know where it will bring you."
He recalls the day in 1995 when his father brought him a Harvard T-shirt after a business trip to Boston, saying, "Well, son, maybe one day you'll go to Harvard." At the time, says Golosov, "I thought it was more likely I'll go to the moon than Harvard." Years later, to his father's chagrin, he turned down an offer to join Harvard's faculty and accepted MIT's. "I still have the T-shirt, and he reminds me of that story very often," he says, a broad smile overtaking his face.
Tsyvinski once planned to be in the foreign service, then international business. Now he speaks of a possible political career in Belarus. "I'm definitely interested in politics ... and you know, I think it fits my personality," he says. "But not for a while. For now I'm going to try to push academia, push research forward."
Golosov has the same ambition. "Just seeing Oleg trying to push the frontier makes me want to push the frontier more," he says. "It makes me want to go back to my desk and start working too."
Precisely what they'll be working on is difficult to predict, say those who know them well. "It's awfully hard to typecast them," notes Chari, their teacher and collaborator. "They're exceptionally broad. But whatever problem it is that seizes their attention, they'll bring a real combination of technical skill, a nose for the central issues and strong computational abilities to that problem. They're a phenomenal team."