Page 2

The (Limited) Power of Persuasion

Cross-posted from Broadstreet, a new blog devoted to historical political economy.

Among the numerous consequences of COVID for everyday life, there is this: Many Americans will not venture beyond their own dining rooms for Thanksgiving this year, and so will miss out on the opportunity to speak with someone whose political views differ from their own. I confess to enjoying these conversations—I even think I have learned a thing or two—but not everybody is a political scientist. More typical, I suspect, is the angst implied by an op-ed in The Hill last year, which offered advice on “How to Shut Down Your Trump-Supporting Family Member at Thanksgiving Dinner.” Dread at what Tucker Carlson and Sean Hannity have wrought since you last sat down to turkey and stuffing—this is the American holiday condition.

Listen, the Fox News effect is real, as Stefano DellaVigna and Ethan Kaplan showed in a much-cited paper. But we give too much credit to Roger Ailes if we grant magical powers to propaganda and misinformation. Political scientists should be especially skeptical—after all, it was William Riker who famously dismissed rhetoric in favor of the more consequential “heresthetic.” Skillful politicians, Riker argued, focus on out-maneuvering their opponents rather than persuading them. More recently, models of Bayesian persuasion have shown that biased information can be effective at changing the beliefs of even fully rational individuals, but there are limits: insist that the sky is always blue (or that a million people showed up at one’s inauguration), and people will tune out the message. Given a choice, they may even change the channel, as Konstantin Sonin and I argued in “Government Control of the Media”—a phenomenon currently at work in Belarus, as citizens desperate for information on COVID turn away from state media.

Historical political economy provides a useful vantage point on these issues. As Ali argued the day after the election, there is nothing older than fake news. And among the twentieth century’s numerous purveyors of a false world view, perhaps none is understood to be more successful than Adolf Hitler. Volker Ullrich’s magisterial Hitler: Ascent, 1889–1939 is typical in this regard. No crude propagandist, Ullrich argues, Hitler knows how to read the room: demagogic rabble rouser at the Hofbrauhaus, polite anti-Semite at the Wagners’ salon. Time and again, he moves his audience to tears, to anger, to action. Not least, there is the trial after the failed Beer Hall Putsch of 1923, when Hitler “use[s] the tribunal as pulpit” and “turn[s] a fiasco into a propaganda triumph.”1 After Hitler was released from prison, most of Germany’s federal states banned him from speaking in public, apparently fearing his words as much as his stormtroopers.

But how effective were Hitler’s speeches really? Not so much, argue Peter Selb and Simon Munzert in “Examining a Most Likely Case for Strong Campaign Effects: Hitler’s Speeches and the Rise of the Nazi Party, 1927–1933.” Employing a sort of differences-in-differences estimator that accounts for the deliberate allocation of campaign resources, Selb and Munzert find that Hitler’s campaign appearances—his speeches—had effects on Nazi vote share in the neighborhood of one percentage point. One percentage point. As Selb and Munzert suggest, if you can’t find strong local effects for speeches by a charismatic politician largely shut out of traditional media, you are unlikely to find them anywhere.

That said, maybe the point is precisely that the Nazis did not have access to radio before they were in power—an idea explored by Maja Adena, Ruben Enikolopov, Maria Petrova, Veronica Santarosa, and Ekaterina Zhuravskaya in “Radio and the Rise of the Nazis in Prewar Germany.” Certainly Joseph Goebbels thought radio important. After Hitler seized control of the country, Goebbels wrote, “Now it will be easy to carry on the fight, for we can call on all the resources of the State. Radio and press are at our disposal.”2 Yet the actual effects appear relatively small, though perhaps pivotal given the political instability of the Weimar Republic. Had Nazi propaganda not been banned in 1930, Hitler’s party might have increased their vote share on the order of four percentage points. Had Hitler not gained control of radio after being named chancellor in January 1933, the Nazis might have received three percentage points less support in the March 1933 elections. Perhaps more consequential is how Hitler and Goebbels subsequently used radio to build a narrative around public policy: In “Highway to Hitler,” Nico Voigtländer and Hans-Joachim Voth show that the impact of Autobahn construction on voting in the 1934 plebiscite was greater where radio was more available.

Adena, Enikolopov, Petrova, Santarosa, and Zhuarvskaya identify effects of Nazi propaganda not only via changes in the content of radio programming, but also through variation in access to radio. They further examine the influence of radio after the consolidation of Nazi power—and here is another lesson, one not immediately implied by models of Bayesian persuasion. Nazi radio encouraged various anti-Semitic acts in regions with a history of anti-Semitism, but actually discouraged it in areas without that history. The propaganda backlash implied by these results is not unique to the Nazi setting. In work on contemporary Ukraine, for example, Leonid Peisakhin and Arturas Rozenas find that access to Russian television increased support for pro-Russia parties in Eastern Ukraine among individuals with pro-Russian attitudes ex ante, but if anything decreased it among those with the opposite views.

All of these studies examine the impact of comparatively short-term exposure to propaganda. A further question is what effect long-term exposure might have. In “Nazi Indoctrination and Anti-Semitic Beliefs in Germany,” Nico Voigtländer and Hans-Joachim Voth demonstrate that Germans born in the 1920s and 1930—those who were exposed to Nazi propaganda as students and as members of the Hitler Youth—harbored more anti-Semitic attitudes decades later than did those who were born before or after the Hitler generation. Precisely because such exposure was universal, this sort of cohort analysis has its limits, though Voigtländer and Voth further show that the effects of growing up under Hitler were strongest in parts of Germany with a history of voting for anti-Semitic parties. As with radio, childhood indoctrination seems to have been most effective where there was fertile soil for Nazi propaganda.

One more caveat is in order: Studies of Hitler’s speeches or Nazi radio implicitly focus on top-down propaganda. A potentially more insidious form of persuasion exploits preexisting social networks. As Georgy Egorov and Konstantin Sonin show in unpublished work on “Persuasion on Networks,” propaganda can be most effective if it targets peripheral members of a network, such as those who might congregate in singers’ associations or gymnastics clubs. In the slyly titled “Bowling for Fascism: Social Capital and the Rise of the Nazi Party,” Shanker Satyanath, Nico Voigtländer, and Hans-Joachim Voth show that Nazi Party membership grew more quickly in towns and cities with an abundance of such groups. Apolitical organizations, in this telling, facilitated the “unplanned propaganda of daily social life” that brought the Nazis to power.

In the preface to Hitler: Ascent, 1889–1939, Volker Ullrich asks, given the surfeit of scholarship on Hitler, if yet another biography is needed. Recent work in historical political economy has illustrated that there is still much to learn not only about the Nazi leader, but also about such general topics as propaganda. Whether future biographers will incorporate this research into their understanding of Adolf Hitler—that will depend in part on social scientists’ power of persuasion.


  1. Quotations from pp. 160 and 162, respectively.
  2. Quoted in Adena et al., p. 1886. Many Germans would have heard Hitler’s speeches on the Volksempfänger, or People’s Receiver, pictured above.

Life and Death During the Transition Depression

Cross-posted from Broadstreet, a new blog devoted to historical political economy.

The New York Times published an interesting pair of reports on Tuesday. The first related a recent study in Health Affairs that documents a staggering, and unexpected, decline in hospital admissions since the arrival of COVID in the United States earlier this year. With the important exception of those areas where COVID infections are spiking, hospital beds are sitting empty, as admissions have declined for ailments ranging from asthma to pneumonia to heart attacks. Increased exercise and improvements to air quality—I remember both during my Parisian lockdown earlier this year—may be driving much of the decrease.

Then again, maybe not. The second story reports on estimates of “excess mortality” by the Centers for Disease Control. According to the CDC, there have been 300,000 more deaths this year than would have normally been anticipated—this against the backdrop of 216,025 deaths officially attributed to the novel coronavirus. Some of the difference may be due to fear of contracting COVID in a hospital setting, leading even seriously ill patients to defer medical care.

Understanding the health impact of COVID is difficult because the pandemic is an economic as well as a health crisis. When U.S. states first started locking down, some policymakers and pundits predicted large negative consequences for public health. Peter Navarro led the charge from the White House, warning of the “very significant losses of life and blows to American families that may result from an extended economic shutdown.” Others disagreed, pointing to large improvements to population health during the Great Depression, which saw output declines of a similar magnitude.

We have been down this road before. The argument that economic collapse increases mortality echoes claims made during one of the most important economic events of the past half-century: the transition from centrally planned economies in the ex-communist countries of Eastern Europe and the former Soviet Union. In the worst-affected areas, the “transition depression” saw declines in output larger than in the United States during the Great Depression. Coincident with this economic collapse was a large spike in mortality among working-age men. It seemed as if shock therapy was killing the patient.

In a much-publicized article in the Lancet, David Stuckler, Lawrence King, and Martin McKee offered a particular version of this thesis: that mass privatization of formerly state owned enterprises led to increased unemployment, which in turn contributed to social stress (not directly observed) and thus mortality. “Mass Murder and the Market,” trumpeted the Economist. Post-Soviet reformers and their foreign enablers were guilty not just of economic mismanagement but of manslaughter.

Or were they? Here’s a little secret: medical journals are better equipped to handle randomized control trials than they are observational studies using country-level aggregates. What Stuckler, King, and McKee did made superficial sense—regressions on a country panel with two-way fixed effects—but as John Earle and I subsequently showed (in a letter to the Lancet, and then in this full paper), the Lancet study was a classic case of failing to check for differences in pre-existing trends. The estimated effect of mortality is substantially attenuated if one corrects for these differences. And it disappears entirely if we assume that it takes at least a year after privatization for men to lose their jobs, become unhealthy, and die.

But let’s take a step back: Is it even true that mass privatization resulted in increased unemployment? Not according to the best available estimates. In three of the four postcommunist countries studied by David Brown, John Earle, and Álmos Telegdy, privatization increased productivity, which holding output constant reduces employment—but it similarly increased output, implying no net effect on employment. (Liberalization, not privatization, was responsible for the output collapse, but that’s another story.) In Russia, in contrast, privatization initially decreased both productivity and output, but again with no net effect on employment.

Anyway, we can do better than cross-country analysis. In Russia, as John and I showed, some regions pursued mass privatization more aggressively than others—and the extent of privatization is uncorrelated with changes in mortality among working-age men. Nor do we see any pattern if we disaggregate mortality by cause of death. I could go on. If you’re interested, check out the complete exchange in the Lancet, which also includes a strong critique of the privatization thesis by Christopher Gerry, Tomasz Mickiewicz, and Zlatko Nikoloski.

So, if not mass privatization, what does account for the mortality spike among working-age men in some postcommunist countries during the early 1990s? Many of these men were drinking themselves to death, so it is reasonable to ask what was happening to alcohol markets during this period. Remember Gorbachev’s anti-alcohol campaign? That saved a lot of lives in the short run. But when effective prohibition ended,1 some of the men who would have otherwise died five years earlier died then. That catch-up presents itself as a mortality spike. And here’s the kicker: The spike is less pronounced in ex-Soviet republics with large Muslim populations—anti-alcohol campaigns are less consequential for public health where people are less likely to drink to begin with. Jay Bhattacharya, Christina Gathmann, and Grant Miller have a nice paper that demonstrates this. Meanwhile, price regulation in a context of general hyperinflation resulted in a dramatic fall in the relative price of alcohol. Daniel Treisman shows that variation across regions in the extent of this drop does much to explain variation in mortality.

What’s the moral of this story? In postcommunist countries, there was no simple mapping from economic output to health outcomes. Nor would I expect one in the current crisis. With the economy disrupted in massive and differential ways, the impact on health is hard to pin down. Scholars will be working for years to understand life and death during the COVID pandemic.


1 Witness the very non-Soviet display of vodka in the image at the top.

Navigating the Frontier between History and Social Science

Cross-posted from Broadstreet, a new blog devoted to historical political economy.

The hottest debate in academia the past week has concerned the appropriateness of a new article on “Frontier Culture” by Samuel Bazzi, Martin Fiszbein, and Mesay Gebresilasse. In his post on Monday, Jared unpacked the paper’s argument and summarized the diverse evidence the authors bring to bear on one part of Frederick Jackson Turner’s (in)famous thesis. It’s what we do at Broadstreet: reasoned commentary on important work in historical political economy. Yet I expect the Twitter wars to continue.

Is mutual antagonism between history and the social sciences the inevitable consequence of the historical turn in economics and political science? I don’t think so, but if the relationship is to survive, each side needs to try—seriously try—to understand the other. As Tracy and I discussed in our inaugural post, we have been working hard to foster that understanding in our community of scholars working on Imperial and Soviet Russia. Let me describe a few things I think I have learned about historians from that experience and from the historical turn in my own research.

Historians worry more about the ideological origins of ideas than social scientists do

With the advantage of hindsight, it was probably inevitable that a paper with “Frontier Thesis” in the abstract would provoke a firestorm. As I read the commentary by historians on Twitter, the use of a phrase so closely associated with the nineteenth century suggests acceptance of the murder and expropriation of millions of native Americans and the cultural attitudes of those who displaced them. For a social scientist, in contrast, the Frontier Thesis is a rich source of ideas for empirical work.

For better or worse, social scientists are just more comfortable separating the text from the author and his times. Indeed, they may not even remember the author! Here is a test: Ask a graduate student in economics to explain to you the roots of utility theory in moral philosophy. Most will be unable to do so. To a historian, I suspect, that is simply inconceivable—that Ph.D. training does not school students in the origins of the most fundamental ideas in the discipline. Indeed, to many historians, the origin of ideas is the question, not just useful context.

Historians are kind of like “traditional” comparativists

For those not in political science, comparative politics is the study of the domestic politics of countries other than the United States. (Weird, I know.) Traditionally, it was the preserve of political scientists who were especially good at languages and not particularly afraid of contracting malaria. Comparativists really knew countries; they were the scholars who got invited onto the evening news to discuss a crisis in some remote corner of the world. Somewhat uncomfortably, comparativists attended the same department meetings as Americanists, who knew fixed-point theorems and probability theory. Comparativists were skeptical that you could understand politics in their countries with those things.

Historians are kind of like that. Having invested in learning the trees, they are uncomfortable with attempts to statistically describe the forest; historians know facts, not “stylized facts.” Try discussing your weather variable with a demographic historian. He will tell you—and rightly so—that there are more types of bad weather than are dreamt of in your methodology. Your response that the weather variable is nonetheless highly correlated with grain production is likely to fall on deaf ears.

I suspect that some of the students who a generation ago would have gone into comparative politics today gravitate toward history. That’s fine: it is useful to have an academy sorted by interest and comparative advantage. But it means that the conversations that I remember from my graduate coursework in political science are today more likely to happen when historians and social scientists find themselves in the same room.

Historians believe that social scientists are inattentive to data quality

The typical historian does not understand how much social scientists obsess over the error term. Consequently, when they see that a social scientist has used some data source that historians know to be a mess, they assume they cannot believe the results. Garbage in, garbage out. The idea that one can statistically compensate for measurement error is learned in econometrics, not historical method.

I have experienced this in my own historical work, much of which examines the causes and consequences of peasant unrest in late Imperial Russia. The primary data source for unrest is a multi-volume chronicle of the “peasant movement” compiled by Soviet historians working during the Khrushchev thaw. Those historians, in turn, relied on a variety of sources, including reports by Tsarist-era provincial governors and secret police. There is a lot to worry about here: the incentives of Russian officials to report what they knew, differential capacity to monitor unrest across space and over time, the presence in the chronicles of politically irrelevant events alongside major disturbances involving thousands of peasants. I’ve heard it all.

The thing is, the chronicles give us a lot of leverage to deal with these issues. We know the archives on which particular entries are based, so we can aggregate up from events reported only by officials whom we presume—based on what we know from historians—to be less biased in their reporting. Moreover, when peasant unrest is on the right-hand side of the regression equation, we can exploit the historical record to locate plausibly exogenous variation in disturbances. Historians identify problems with the data, but they also (sometimes unknowingly) give us solutions. We should express our gratitude.

Historians are deeply frustrated when social scientists ignore the historiography on a topic

Hey, political scientists are deeply frustrated when economists “discover” a question on which political scientists have worked for decades. Economists are deeply frustrated when humanists base their understanding of the economy more on Marx than on Econ 101. We could do this all day. The point is this: If you want people to take you seriously, you need to signal that you have read and understand their work. With all due respect to Bazzi, Fiszbein, and Gebresilasse, who wrote a really careful paper, a single paragraph acknowledging that “Turner’s work has attracted immense attention and vast criticism” doesn’t send that signal. It certainly doesn’t capture the discipline’s rejection not just of Turner’s particular arguments, but of his way doing history, his school.

I understand that there is essentially zero publication incentive to devote space to another discipline’s debates, and that even those who want to do so may find that their editors disagree.1 Fine, but if we don’t, we shouldn’t be surprised if the other discipline returns the favor. Moreover, if we really are interested in learning the literature, we shouldn’t assume that we can do it on the cheap. Interdisciplinary workshops help, as Tracy and I discussed in our last post, but the best way to understand another discipline is to take a class. I was fortunate at Wisconsin to have an outstanding historian of Imperial Russia as a colleague. David McDonald let me sit in on his graduate seminar, where we read and debated the historiography of Tsarist Russia. It was the start of an education, still ongoing.


1 Still, with 50-page appendixes now the norm for empirical papers, perhaps we could spare a page or two in the appendix for a more substantial literature review that is signposted in the main text.

An Interdisciplinary Conversation

A conversation with Tracy Dennison, cross-posted from Broadstreet, a new blog devoted to historical political economy.

Scott: Tracy, this is the first post at Broadstreet for both of us. Welcome!

Tracy: And welcome to you! It feels a bit like cheating to make the first post a shared effort, but it’s very much in the interdisciplinary spirit of the blog itself.

Scott: A historian and a political economist talking to each other…sort of like one of Gail Collins and Bret Stephens’ periodic conversations.

Tracy: Just like that, but different! Better, even. Like political economy without the politics.

Scott: In fairness, we’ve had practice. Last year, in Madison, we held the first annual Summer Workshop in the Economic History and Historical Political Economy of Russia. Economists, historians, and political scientists all in the same room together, discussing each others’ papers. It was mostly wonderful, though I recall we did have a spirited conversation about “the relationship” at the end.

Tracy: It’s inevitable. I think it’s part of the peace process. I can only assume there were no hard feelings, as everyone wanted to come back and do it again this year!

Scott: And that’s what we would have done—in Paris, moreover—if the world hadn’t turned upside down this past spring. In the end, we moved the workshop online, like most every other part of our daily existence.

Tracy: It was a big disappointment to have to move to Zoom. Not only because it meant giving up Paris, but because the inaugural meeting in Madison had fostered so many productive in-person interactions. It wasn’t clear how we’d be able to recreate that atmosphere online.

Scott: That’s an interesting question—whether we were able to do so. We opened up the workshop to anybody in the field, which attracted a much larger audience than we were aware existed. But do you think we were talking to each other or only past each other?

Tracy: I think the constraints of Zoom were evident—it’s difficult to hash out those thorny questions of methodology or interpretation without being in the same room with one another and without being able to continue those conversations later in the bar. That said, I was delighted by how well it went in the end.

Scott: We had some great papers, and an online poster session to boot. (Here’s the program.) Mark Harrison, one of the deans of the field, led off the workshop with a keynote address on the KGB in Lithuania.

Tracy: Yes, it was the perfect opening; it really set up the interdisciplinary context nicely. The paper examines a question that has long interested historians of the Soviet Union: how effective were attempts to prevent deviant—as defined by Soviet authorities—behavior? And how can we measure these effects? It touched on a number of important issues, including the reliability of Soviet source materials, variation across space and time, and the changing priorities of both officials and Soviet citizens.

Scott: Once we moved to the paper sessions, it was interesting to see certain themes emerge. We had two papers on deportations under Stalin, and two on what we might call “rural political economy.” I don’t think that was an intentional choice on our part. It just reflects the state of the field.

Tracy: It was exciting to see so much new work on some of the bigger problems in Soviet history—industrialization, markets and finance, political repression, collectivization—from younger colleagues in the field. I was impressed by the ways in which they were using new sources and new methods to approach these questions. They certainly provided the group with much to discuss and debate.

Scott: You and I have been in the room where these discussions happen for awhile now. One recurring debate has centered on the use of digitized archival records by economists and political scientists. Roughly speaking, I would characterize the conversation as, Historian: You have no idea how noisy these data are and how biased is their selection into the archives. Social scientist: I understand at least some of that, and my fancy research design is intended to compensate.

Tracy: That sounds about right. I think it often boils down to one really fundamental issue: historians’ skepticism about numbers and social scientists’ doubts about textual evidence (please don’t call it anecdotal!). Somehow our “relationship discussions” always seem to revolve around some version of this problem. This is something I hope to explore in greater detail in future blogposts.

Scott: We had some variant of this in our session devoted to a new paper by Andrei Markevich, Natalya Naumenko, and Nancy Qian on the Stalin-era “Great Famine.” It’s really fascinating work that addresses an old question—whether Ukrainians, who died in enormous numbers during the famine, were targeted deliberately or merely caught up in a monumentally catastrophic policy. Many well-known historians have weighed in on this—Stephen Kotkin, Timothy Snyder, and Anne Applebaum come to mind—but Andrei, Natalya, and Nancy are the first to try to answer this question econometrically. (The working paper is online for those interested in their answer.)

Tracy: Yes, it generated quite a lively discussion—heated at times (friendly heated!), but exactly the sort of thing we are striving for with this workshop. It is worth noting that many of those who participated in the discussion were continuing a debate on the same set of questions that had been initiated in person last year in Madison. I interpreted this as strong evidence that the workshop is facilitating sustained interactions across battle—er, disciplinary—lines.

Scott: Part of that discussion was whether we should be looking for comments by Stalin or others as evidence of intent, or whether we should let the numbers do the talking. We ended up in this sort of absurd situation where we could not agree on the meaning of the word “anecdote.”

Tracy: True. But it highlighted some important ways in which we do talk past one another. Social scientists see the use of text to support an argument as arbitrary and unreliable, since, in their view, one can just look through a source and cherry-pick some quotes. Historians feel the same way about quantitative evidence. Each side assumes the other understands the value of its methodology and the reasons for skepticism, but this is almost never the case. We need more group therapy, I guess.

Scott: Or more joint training. My History colleague Stephen Pincus and I have discussed teaching a graduate course on “reading across disciplinary divides,” or something like that. We don’t need to turn historians into economists or political scientists, and vice versa, but we do need to be able to read each others’ work critically. You’re one of the few scholars I know who really invested in learning the language of more than one discipline. How did you do it?

Tracy: That is a great idea for a graduate course, and I completely agree about the value of being able to critically read each others’ work. I did my doctoral research in the UK, where there are economic history departments in which a large number of historians and social scientists are forced to cohabit. It was a conscious decision to train as a historian, but along the way I seem to have internalized a social science way of looking at the world.

But Scott, you are also a scholar of this kind. What’s your story?

Scott: I remember struggling before graduate school with the path I would take: political science or economics. Ultimately, I enrolled in a political science program, but I never really resolved the dilemma. I ended up doing as much coursework in economics as in political science, and my degree is in both. And then I discovered history as a field of study when I was a newly tenured professor. But I’m still working to learn the language of history as a discipline. The workshop and all of our conference panels help.

Tracy: I think having intellectual relationships with people from other disciplines is particularly important. As I recall, one of our aims with this workshop was to pave a way to interdisciplinary engagement for younger colleagues in our respective fields, to ensure they would have the ability to make connections across disciplines. Their participation in the project has been invaluable, and gives me hope for enduring collaborative efforts of this sort.

Scott: Since we have the megaphone for a day, we should give a shout-out to the many junior scholars who participated: Gerhard Toews, Natalya Naumenko, Dmitrii Kofanov, Giovanni Cadioli, Otto Kienitz, Viktor Malein, Brendan McElroy, Timur Natkhov, Natalia Vasilenok, Matthew Reichert, Eugenia Nazrullaeva, and Imil Nurutdinov. And we had presentations by a quartet of outstanding senior scholars: Ekaterina Zhuravskaya, Pierre-Louis Vézina, Mark Harrison, and Andrei Markevich. The state of the field is strong!

Tracy: It is! Fingers crossed we’ll be able to revisit these debates in person at next year’s workshop. In the meantime, there will be plenty of interdisciplinary collaboration on display here at Broadstreet. Stay tuned!

Leaving Paris

On a warm summer day in late August 2019, we moved into a third-floor apartment in Paris’s residential 15th arrondissement. Standing on our balcony and looking to the left, we could see the district’s mairie—its town hall—and behind it the Eiffel Tower. Directly across the small park in front of us was a row of apartment buildings and a contemporary office structure. Just up the street was the Vaugirard metro station: our gateway to the rest of Paris.

With a bit more foresight, I might have anticipated that we would lose access to that gateway as one of France’s periodic transit strikes disrupted transportation throughout the nation’s capital, and beyond. No understanding of French political economy, however, could have prepared me for the city’s complete lockdown in March as a pandemic hit Paris with particular ferocity. Even after the country’s reopening in May and June, I experienced Paris as Baron Hausmann had envisioned—as a pedestrian.1

In The Walkable City, Mary Soderstrom posits a conversation between Georges-Eugène Hausmann, who under Emperor Napoleon III transformed Paris from medieval city to modern metropolis, and Jane Jacobs, who fought against a similar effort to clear New York’s “slums” a century later. Baron Hausmann was the Robert Moses of his day: an ambitious planner unencumbered by nostalgia for the city’s past. Jacobs would have hated him. And yet, while American urban planning has turned sharply against Moses’s vision, Paris survives as a model for the walkable city.

The difference, I suspect, has something to do with technology. Not only did Hausmann work in a pre-automobile age, but steel-frame construction had yet to emerge as America’s seminal contribution to modern architecture. The quintessential Parisian neighborhood is thus capped at seven stories. With numerous such buildings, Paris is dense, but it is rarely crowded, and it is always light. Many neighborhoods have the feel of a village.

Parisians did not take to the Eiffel Tower at first, but it proved such a draw for tourists that they learned to like it. (The same cannot be said for the unfortunate Tour Montparnasse.) Today, the Eiffel Tower is the iconic symbol of a city that annually draws tens of millions of sightseers. No living Parisian could remember Paris without either tourists or, for those old enough, foreign troops—until this spring. Sometime in March, the city emptied out. During the height of the confinement, I would go for long runs down deserted streets, jogging in the roadway to avoid the few pedestrians about. Even after restrictions were lifted and well-off Parisians returned to the city, the tourists were absent. We could sit on the Champ de Mars in early July with only a handful of picnicking locals in proximity.

People ask if I feel cheated that my sabbatical coincided with the pandemic. Au contraire! I got to see Paris as few have. We were scared at first, but the fear passed even before the danger did, leaving solidarity in its wake. Every evening at 8:00, we joined our neighbors on the balcony to applaud the city’s healthcare workers—and ourselves. For weeks, this daily ritual was our primary social interaction outside of the home. Elderly couple waving a white scarf, woman on distant balcony with long black hair, man in glasses (Nicolas, we eventually learned) one building over…we shared the confinement with you.

By July, even as infections spiked anew at home, we felt secure enough to travel around France by car. After a whirlwind tour of WWI battlefields and the Normandy beaches, we settled into a comfortable rhythm, staying 2–3 nights in each region. We joined our friends Sergei and Katia in Brittany, and we met up with our friends Mike and Brandi in Fontainebleau and Valery and Sonya in Provence. We ate fresh oysters on the Atlantic shore and fresh trout in the Alps, we visited chateaux in the Loire Valley and hillside villages in Provence, and we drank the local wines wherever we went. It was wonderful.

Memory works in interesting ways. If I fast forward through part an audiobook that I have already heard, I can remember where I was at the precise moment that a particular phrase was spoken. But when I again walk the street where I listened to that book, my mind is elsewhere—no visual cue prompts remembrance of the spoken word.

What will I remember when I can finally visit Paris again? I do not know, but I can guess: the embankment of the Seine during the morning hours when running was allowed, the square in front of the mairie where I played tag with my son as we took a break from online schooling, the view from our balcony and the sound of the city applauding. Paris, j’espère te voir bientôt. Tu me manques déjà.

Place Adolphe Chérioux. Artwork by Mike Duncan.


  1. Underground rail came late to Paris: the first Metro line opened only in 1900.

Nondemocracy (Second Edition)

When I wrote the first edition of my textbook on Formal Models of Domestic Politics, I made a conscious decision not to include models of autocracy. The literature was too new, the big picture insufficiently clear. There was enough to cover on more established topics. Nondemocracy could wait.

That first edition went to press in 2012. In the eight years since, there has been an explosion of interest in authoritarian politics. I see it at academic conferences, where panels on autocracy draw overflow audiences. Creative empirical work has cast new light on the governance of nondemocratic regimes. Simultaneously, and in dialogue with empirical research, political scientists and economists have explored the institutions of autocracy using the language of game theory. The themes of the theoretical literature are now sufficiently clear to deserve a chapter in the second edition of my textbook.

Of all the topics in the text, this is the one in which I have worked most directly. I have drawn substantially on a review article I published a few years ago with Konstantin Sonin and Milan Svolik, though I haven’t stopped there. In summarizing the literature, I have tried to be a faithful ambassador, including important models that may or may not reflect my personal views on the nature of autocracy. Nonetheless, where the literature has branched—in how to model information manipulation, for example—I have made some choices that inevitably reflect my personal modeling experience, even as I provide signposts to other parts of the literature.

With that as background, let me offer a brief survey. The chapter begins with Daron Acemoglu, Georgy Egorov, and Konstantin Sonin’s model of coalition formation in nondemocracies. Theirs is an explicitly institution-free setting: politics is governed only by the relative power of competing factions. The model thus serves as a benchmark against which more institutionally rich models can be evaluated. Anybody who has seen the brilliant film The Death of Stalin will immediately recognize the environment.

The chapter then moves to incorporate institutions into the analysis. The selectorate model of Bruce Bueno de Mesquita, James Morrow, Randolph Siverson, and Alastair Smith appears here, transplanted from the chapter on coalitions in which it previously appeared. The basic idea is to show how coalition choice and economic policy depend on the institutional environment, as measured by the relative size of the “selectorate” and the “winning coalition.” A complementary approach is to ask where institutions come from to begin with. Here, I present the simplified version of Roger Myerson’s model of institutions as a commitment device that Konstantin, Milan, and I developed for our review piece.

Myerson’s model brings information into the story; this is the focus for the remainder of the chapter. To set the stage, the chapter includes a general discussion of Bayesian persuasion, the explicit or implicit framework for many models of information manipulation in autocracies. What distinguishes this framework from the cheap-talk model of Crawford and Sobel is the assumption that the sender can commit to a probability distribution over signals for every state of the world—say, to send the signal “economy is weak” with probability 0.6 and the signal “economy is strong” with probability 0.4 if the economy is in fact weak, and to send the signal “economy is strong” with certainty if the economy is in fact strong. Is this assumption reasonable? Here is what I write:

The commitment assumption that characterizes models of Bayesian persuasion is not without controversy. It is reasonable to ask in which settings it is likely to hold. In their seminal contribution, Kamenica and Gentzkow (2011) provide the example of a prosecutor attempting to persuade a judge of a defendant’s guilt. The prosecutor might call a witness, not knowing exactly what she will say. Or she might order a DNA test, hoping for a positive result but constrained by law to share exculpatory evidence. In either case, the prosecutor de facto commits to a distribution of signals for each possible state of the world. Similarly, an autocrat might choose to disqualify candidates from an election (Ma, 2020), not knowing with certainty the distribution of voters’ preferences and thus the mapping from candidates to election outcomes. Alternatively, we can think of commitment as reflecting delegation to an actor of more or less sympathy with the sender’s point of view (think of how reporting on incumbents’ economic performance is affected by the identity of cable news personalities), where the actor is costly to replace (e.g., because charismatic hosts are in short supply). The latter environment may reflect the reality that any dictator relies on others to disseminate her message.

As this discussion illustrates, the commitment assumption is useful in two important environments in nondemocracies: government control of the media and electoral manipulation. The chapter includes some discussion of each, focusing on my model of media control with Konstantin Sonin and on work by Arturas Rozenas and by Alberto Simpser and me on electoral manipulation. This leads naturally to the final section of the chapter—a simplified version of Sergei Guriev and Daniel Treisman’s model of “informational autocracy,” which incorporates information manipulation into the models of political agency covered elsewhere.

As with other chapters, there is a lot of theory in the exercises as well as the main text. Alongside various extensions to the models discussed above, I include a model of “ex post” (i.e., after the election result is known to the incumbent) electoral fraud borrowed from a recent paper by Zhaotian Luo and Arturas Rozenas. (The full paper jointly examines ex ante and ex post manipulation.) Building on work by Gary Cox and by Andrew Little, I also consider authoritarian elections as a mechanism for gathering information about potential challenges to the incumbent regime.


Someone recently thanked me for my “service” in writing Formal Models of Domestic Politics. Oddly, I have never thought of it in those terms. Writing a textbook is one of the most enjoyable things I have done in my academic career. It satisfies various compulsions: to figure things out, to tinker with models, to write as clearly and efficiently as possible. As Gregory Mankiw has recently argued, it’s not for everyone—but it is for me. All the better that others have found the text useful. If things go as planned, the next edition will be out next summer.

Reform and Rebellion in Weak States

Sometime during my first year as a junior faculty member, I was wandering the stacks in Wisconsin’s Memorial Library. I can’t remember what I was looking for, but I remember what I found: a multi-volume chronicle of the peasant movement in Imperial Russia. That serendipitous discovery aroused an interest in economic history and historical political economy that has become one of the pillars of my academic personality. With Evgeny Finkel, Tricia Olsen, Paul Castañeda Dower, Steve Nafziger, and Dmitrii Kofanov, I have examined peasant unrest during the Great Reforms of Alexander II and the Bolshevik Revolution. And now, this week, Evgeny and I have published a short book on Reform and Rebellion in Weak States that returns to the themes of our very first paper with Tricia.1

We begin with the question that motivated that paper: When does reform provoke unrest among its intended beneficiaries? It is an old question, one that arguably dates to Alexis de Tocqueville’s observation (in The Ancien Regime and the Revolution) that “The regime that a revolution destroys is almost always better than the one that immediately preceded it, and experience teaches that the most dangerous time for a bad government is usually when it begins to reform.” Our examination of peasant unrest before and after Russia’s emancipation of the serfs in 1861 suggests a possible answer.

To understand our argument, consider a puzzle: How is it possible that the tsar and his ministers—overseers of a weak state that had just suffered a humiliating loss in the Crimean War—were able to effect a radical transformation of state and society that not only granted legal freedom to twenty million serfs but initiated a lengthy process of land reform that extended into the twentieth century? The short story is that, with innovation born of necessity, they relied on the very nobility whose land was to be transferred to the newly freed peasantry. Empowered by the reform statutes with control over the process, numerous landowners jumped at the opportunity to keep the estate’s best land for themselves and to ensure that former serfs received as little valuable land as possible. This opened a gap between what peasants thought they had been promised in the tsar’s Emancipation Manifesto (read out in village churches across European Russia) and what they actually received. That grievance translated into the largest wave of peasant unrest in the nineteenth century.

Our intuition, then, is that the delegation of reform’s implementation to those with a stake in the status quo plants the seeds of rebellion. But this raises further questions: To what extent do those with the power to block implementation internalize the resulting unrest? And is there always a trade-off between stability and reform, or is it possible to pursue the latter without risking the former? These are the sort of questions best answered with a model. In the book, we provide a simple one that builds on recent work in economics on contracts as reference points. In our context, the promise of reform represents an implicit contract against which its subsequent implementation is measured. From the tsar’s perspective, this promise is a double-edged sword. On the one hand, by allowing for disappointment, reform raises the possibility of unrest. On the other, the possibility of unrest encourages those with control over implementation—local agents, in our telling—to abide by the spirit of reform. A very general result is that any reform that incentivizes local implementation necessarily also increases the risk of rebellion.

This theoretical framework provides a lens through which to reexamine the Russian case. Relative to our earlier work, the analysis in the book pays more attention to the dynamics of unrest over the two-year period that follows publication of the Emancipation Manifesto in early 1861. A key finding is that it took time and the arrival of “peace arbitrators” (Leo Tolstoy was one) for peasants’ understanding of the intent of reform to settle. Once it did, and the actual process of negotiating terms began in earnest, landowner opportunism and peasant unrest were joined at the hip.

But wait: there’s more. In addition to a detailed analysis of the Russian case, for which we have rich data and a setting that allows for causal identification (emancipation directly affected only one of two large classes of peasants distributed unevenly across European Russia), we provide an extended discussion of four additional cases, each of which supplies some form of empirical leverage. The Tanzimat Reforms of the Ottoman Empire are a “most likely” case for our argument, given similarities in the nature of reform and the broader political context with the Russian case that motivates our theory. In contrast, the Lex Sempronia of Tiberius Gracchus—perhaps the world’s first major land reform—exhibits important internal variation in reform design, with top-down implementation among Roman citizens but not Rome’s Italian allies. (A big shoutout to my podcaster friend Mike Duncan of History of Rome and Revolutions fame for suggesting that we look at Ancient Rome.) Our extended riff on Tocqueville demands that we reexamine the French Revolution, where a detailed historical record allows us to trace the process by which the National Assembly’s declaration of the end of feudalism ultimately led to increased unrest. Finally, we briefly analyze several instances of land reform in twentieth-century Latin America, exploiting important variation not only across but within states.

It’s a fun read, and a short one. And it’s available for free download through June 17. Check it out.


  1. Technically speaking, what Evgeny and I have written is an “Element”: what Cambridge University Press calls their new short-book format. Ours is in David Stasavage’s series on Cambridge Elements in Political Economy.

Political Agency (Second Edition)

One of the pleasures of revising my textbook on Formal Models of Domestic Politics has been discovering work that speaks to our current politics. I began a series of posts on forthcoming changes to the text with a discussion of Wiola Dziuda and Antoine Loeper’s model of dynamic veto bargaining, which helps to explain Republican resistance to economic aid that might outlast the COVID crisis. Last month, I wrote about the Krasa/Polborn model of issue ownership, which suggests that a “Downsian” change in electoral platform may not resolve the GOP’s dilemma in the face of shifting demographics. Today, I turn attention to the politics of natural disasters, of which the COVID pandemic is a prime example.

Existing work—empirical and theoretical—suggests that a pandemic could have at least three possible effects. First, one might expect voters to react “irrationally,” punishing incumbents for events beyond their control, just as voters are alleged to have responded to shark attacks or losses in college football games. Even if one is willing to assume, however, that nothing about the course of the pandemic is within the control of governing elites, recent scholarship casts doubt on whether claims in the empirical literature generalize or hold in the cases originally examined. Second, it is in fact not obvious that voter “competence” is necessary for political accountability. Indeed, the strategic incentives of politicians who want to communicate that they are deserving of reelection may rely on the inability of voters to observe incumbents’ actions or to respond rationally.

Third, and perhaps most relevant to our times, the presence of a disaster such as a pandemic may provide an exceptional opportunity for voters to evaluate an incumbent politician’s fitness for office. Scott Ashworth, Ethan Bueno de Mesquita, and Amanda Friedenberg argue this point in a model of pure selection that is a primary addition to my chapter on political agency. In the Ashworth-Bueno de Mesquita-Friedenberg model, the possible impact of a disaster on an incumbent’s fortunes depends on whether she was previously “ahead” or “behind” in a race for reelection. If voters ex ante believe that an incumbent is more competent than her challenger, then a disaster—which may reveal the incumbent to be either competent or incompetent—can only reduce her probability of reelection. In contrast, those (possibly exceptional) incumbents who are disadvantaged relative to their competitors may benefit from events beyond their control: in essence, nature has gambled on their resurrection, providing them an opportunity to reveal their competence, should that be present.

Whether you think the pandemic increases or decreases Donald Trump’s chances of reelection thus depends on whether you think he entered 2020 the favorite or the underdog. If you believe, as I do, that he was behind, then COVID likely increases the probability of both a landslide loss and a narrow victory. Before you dismiss the latter possibility, ask yourself the following question: How will the American electorate respond if Trump moves beyond Lysol-gate and somehow manages to badger the FDA and the pharmaceutical industry into an effective vaccine ahead of schedule? It’s maybe not the most likely outcome, but it’s the sort of outcome that is more likely in the presence of a pandemic.

One more question: Why does it take so long to develop a vaccine anyway? I am no expert, but Bill Gates asserts that, had we invested heavily in vaccine production over the past decade, we could have been across the finish line in a year. Empirically, we know that voters fail to reward politicians for disaster preparedness, even though the returns to such investments are enormous. In the exercises (one could learn an enormous amount of theory just working through the exercises), I present a model by Sean Gailmard and John Patty that makes sense of such behavior. In the Gailmard-Patty model, voters prefer prevention spending but worry that politicians are investing in preparedness for the wrong reasons—say, to steer government contracts to friends or supporters. In equilibrium, even honest politicians may therefore fail to invest in prevention, as doing so may convey the impression that they are corrupt.


So, that’s about it. I have one more blog post planned on changes to my textbook, and it’s a big one: a discussion of a new chapter on models of nondemocracy. Stay tuned.

Delegation (Second Edition)

I consider the chapter on delegation in Formal Models of Domestic Politics to be unusually coherent. This is not patting myself on the back. Rather, the literature that this chapter summarizes, with its origins in the seminal work of Holmström, is of a piece. One paper follows another; all I had to do was follow the bread crumbs.

What, then, to add? As I have discussed previously, the second edition of my textbook features new models, new exercises, and clarifications. The new models push on both substantive and methodological frontiers, providing a mix of novel argument and novel technique. With this chapter, I was able to extend the theoretical narrative in an interesting direction while introducing a class of models that belonged somewhere in the text.

In particular: In the first edition, the chapter on delegation closed with a discussion of Gilligan and Krehbiel’s extension of the Crawford-Sobel cheap-talk model to delegation and information transmission in legislatures. This seemed a natural segue to a recent literature on cheap talk with multiple senders, where Galeotti, Ghiglino, and Squintani’s model of strategic information transmission networks has provided a framework for analysis of numerous political environments. Happily, a paper by Torun Dewan and Francesco Squintani provided the hook: their model of “Leadership with Trustworthy Associates” examines delegation to leaders. My chapter, which began with a model of delegation to bureaucratic agencies and grew to encompass delegation to legislative committees, thus now incorporates delegation by party elites to a leader tasked with making decisions on their behalf. The trail of bread crumbs is intact.

I do have one regret with this chapter. A pedagogical principle that guides the text is to present no more analysis than can be written on the blackboard and understood by Ph.D. students who have taken a semester of game theory and a semester of calculus. When that proves impossible with the original version of a model, I typically strip it down to something simpler; I try to avoid statements of the sort “X and Y show that Z is true.” I wasn’t quite able to do that with the Dewan-Squintani model without abandoning entirely the theoretical framework I wanted to demonstrate. There are a couple of hand waves. I hope the accompanying discussion compensates, but readers will be the judge.

Coalitions (Second Edition)

Halfway through my posts on the second edition of Formal Models of Domestic Politics, I am reaching the finish line for the manuscript itself. If all goes as planned, Cambridge will have the draft by next weekend. If you are reading this post, you might be reading the manuscript itself in a few weeks. Thank you in advance for your comments!

One of the joys of writing, and now rewriting, this book has been discovering papers that otherwise I might not have known. This is perhaps especially the case for the literature on legislative bargaining and coalition formation, which is not an area in which I have worked. I remember writing the first edition and discovering this amazing paper by Hülya Eraslan that showed that, notwithstanding the multiplicity of equilibria in the Baron-Ferejohn bargaining model, any stationary equilibrium (given particular recognition probabilities and discount factors) generates the same vector of expected payoffs. This is a remarkable result with great potential for applied work, where one might want to “plug in” a legislative bargaining game to proxy for the stakes in some contest for political power.

My reaction upon stumbling across a recent paper by Vincent Anesi and Daniel Seidmann was more bittersweet. The Baron-Ferejohn model that Eraslan examines assumes that bargaining is once-and-for-all: there is just one pie to be divided. In many contexts, however, a budget is negotiated annually or biannually, with the outcome in one period the default in the next. (I discussed a similar environment in my post on veto players last month.) Anesi and Seidmann show that, in this setting, most anything can happen in equilibrium. Contra Riker, winning coalitions may be more than minimal; in some equilibria, everyone gets a piece of the pie. Even more surprisingly, in period after period, some of the pie may be left on the table. (I have a hard time envisioning this with rhubarb pie, but appeals to welfare maximization will not get you far in this environment.) This is an “anything goes” result more typical of equilibria with history dependence, but Anesi and Seidmann work in a world of stationary strategies. It is a brilliant paper that fits naturally in the chapter on coalitions. It is a disappointing result, if you were hoping for predictive power similar to that of the Baron-Ferejohn model.

As with other chapters, I use the exercises to present related work. There is so much that could be included here; I have a long list of ideas for future exams. For the moment, I have stopped at two new exercises—one based on Maggie Penn‘s model of farsighted voting, in which the status quo is endogenous but broad societal forces rather than legislators drive the agenda, and one on Craig Volden and Alan Wiseman‘s model of bargaining over public and private goods. If I include any more, I will be past my deadline and over my page limit.