Social Science Applications Forum

The seminar series, Social Science Applications Forum, will be held on selected Tuesdays from 4:30 – 5:30pm in CMSA Building, 20 Garden Street, Room G02.

The list of speakers is below and will be updated as details are confirmed.

Date Name Title/Abstract
02-07-17 Nikhil Naik, Harvard/MIT


Title: Visual Urban Sensing: Understanding Cities with Computer Vision

Abstract: Street View services have documented the visual appearance of cities from more than a hundred countries across the world in the past decade. I design computer vision tools that harness Street View imagery to conduct computer-driven automated surveys of the built environment at street-level resolution and global scale. In this talk, I will describe two algorithms that computationally evaluate urban appearance from imagery. The first algorithm, Streetscore,  quantifies the perceived safety of a street block, by harnessing data from a crowdsourced game. The second algorithm quantifies the growth or decay of cities from time-series Street View imagery obtained over several years. Finally, I will demonstrate the use of these algorithms for studying important questions in urban economics, sociology, and urban planning.

02-14-17 Mauricio Fernández Duque, Harvard


Title: Pluralistic Ignorance and Preference Complementarities

Abstract: I develop a theory of group interaction with preference complementarities, in which individuals know that others privately judge whether their type matches that of the majority in the group. The model sheds light on situations of pluralistic ignorance — a social situation where `a majority of group members privately reject a norm, but incorrectly assume that most others accept it, and therefore go along with it.’ (Katz and Allport, 1931). As opposed to past approaches, our model explains three stylized facts of pluralistic ignorance: that most individuals are acting reluctantly, that they think most others are not acting reluctantly, and that an inefficient equilibrium can be sustained in which most are reluctantly cooperating. We show some preliminary results suggesting that as the certainty over the population distribution of preferences grows, the probability of pluralistic ignorance increases when group size is small and decreases when group size is large.

02-21-17 Ravi Jagadeesan, Harvard

Title: Complementary inputs and the existence of stable outcomes in large trading networks

Abstract: This paper studies a model of large trading networks with bilateral contracts.  The model allows income effects, unlike in parts of the matching literature, and imperfectly tradeable goods, unlike in the general equilibrium literature.  In our setting, under standard continuity and convexity conditions, a stable outcome is guaranteed to exist in any acyclic network, as long as all firms regard sales as substitutes and the market is large.  Thus, complementarities between inputs do not preclude the existence of stable outcomes in large markets.  Even when there are complementarities between sales, this paper shows that tree stable outcomes are guaranteed to exist in large markets, under continuity and convexity conditions.

The model presented in this paper generalizes and unifies versions of general equilibrium models with divisible and indivisible goods, matching models with continuously divisible contracts, models of large (two-sided) matching with complementarities, and club formation models.  Additional results provide intuition for the role of uni-directional substitutability conditions and acyclicity in the main existence results, and explain what kinds of equilibria are guaranteed to exist even when these conditions are relaxed.  Unlike in two-sided large-market settings, the sufficient conditions described in this paper pin down maximal domains for the existence of equilibria.

03-07-17 Krishna Pendakur, Simon Fraser University



Title: Infant Mortality and the Repeal of Federal Prohibition

Abstract: Exploiting a newly constructed dataset on county-level variation in prohibition status, this paper asks two questions: what were the effects of the repeal of federal prohibition on infant mortality? And were there any significant externalities from the individual policy choices of counties and states on their neighbors? We find that dry counties with at least one wet neighbor saw baseline infant mortality increase by roughly 3%. We argue that such cross-border policy externalities are plausibly exogenous sources of variation in assessing the effects of prohibition’s repeal and should be a key consideration in the contemporary policy debate on the prohibition of illicit substances.


 Spring Break — NO SEMINAR
03-21-17 Danielle Li, Harvard


Title: Financing Novel Drugs

Abstract: The process of drug discovery is expensive and highly uncertain. Because of this, risk-averse firms facing financial constraints may be less likely to invest in novel drugs than may be socially optimal. We examine this hypothesis by studying how resource constraints impact firms’ decisions to invest in novel drug compounds. This paper has two contributions. First, we develop a new molecule-based measure of the chemical novelty of new drug candidates. Second, we use variation in the expansion of Medicare prescription drug coverage in the United States (which differentially benefited firms with more drugs targeted toward the elderly, as well as firms with more remaining market exclusivity on those drugs) to isolate exogenous variation in cash flow to firms. We find that firms which benefit more from the expansion of drug coverage develop more drug candidates as a result and, moreover, this increase is driven by an increase in the development of more chemically novel drug compounds.

04-11-17 Jann Speiss, Harvard


Title: Integrating Machine Learning into Applied Econometrics

Abstract: Machine learning focusses on high-quality prediction rather than on unbiased parameter estimation, limiting its direct use in typical program evaluation applications in economics. Still, many tasks that we think of estimation problems have implicit prediction components. In this talk, I (1) lay out some basic features and limitations of machine learning; (2) propose a framework for integrating it into estimation tasks; and (3) work out two examples from experimental analysis to highlight my approach: accounting for control variables, and testing effects on multiple outcomes.

04-12-17 Alex Teytelboym, Oxford

Alex_Teytelboym (1)


Title: Refugee Resettlement

Abstract: Over 100,000 refugees are permanently resettled from refugee camps to hosting countries every year. Nevertheless, refugee resettlement processes in most countries are ad hoc, accounting for neither the priorities of hosting communities nor the preferences of refugees themselves. Building on models from two- sided matching theory, we introduce a new framework for matching with multidimensional constraints that models refugee families’ needs for multiple units of different services, as well as the service capacities of local areas. We propose several refugee resettlement mechanisms that can be used by hosting countries under various institutional and informational constraints. Our mechanisms can improve match efficiency, incentivize refugees to report where they would like to settle, and respect priorities of local areas thereby encouraging them to accept more refugees overall. Beyond the refugee resettlement context, our model
has applications ranging from the allocation of daycare slots to the incorporation of complex diversity constraints in public school assignment.

This seminar will run from 4:00-5:00pm 

04-18-17 Battal Doğan, University of Lausanne

Battal Dogan

Title: A precise representation of acceptant and substitutable choice rules

Abstract: Acceptant and substitutable choice rules have been useful in market design particularly because they guarantee stability and allow for achieving various design objectives such as diversity. What is known from the literature is that each such choice rule has a “collected maximal representation”: there exists a list of priority orderings such that at each choice set, collecting the maximizers of the priority orderings retrieves the choice. For each acceptant and substitutable choice rule, we provide the minimum size collected maximal representation. We show that “responsive choice rules” render a collected maximal representation of the largest size among all acceptant and substitutable choice rules.

05-02-17 Mohammad Akbarpour, Stanford


Title: Dynamic matching on stochastic networks

Abstract:  Motivated by the problem of paired kidney exchange, we introduce a dynamic version of the problem of maximum matching in a random network, where agents arrive and depart stochastically, and the composition of the trade network depends endogenously on the matching algorithm. We show that if the planner can identify agents who are about to depart, then waiting to thicken the market is highly valuable, and if the planner cannot identify such agents, then matching agents greedily is close to optimal. The planner’s decision problem in our model involves a combinatorially complex state space. However, we show that simple local algorithms that choose the right time to match agents, but do not exploit the global network structure, can perform close to complex optimal algorithms.


Date Name Title
11-01-16 Jakub Redlicki, Oxford


Title: What Drives Regimes to Manipulate Information: Criticism, Collective Action, and Coordination

Abstract: Authoritarian states often manipulate information in order to prevent regime change. The model uses the global games framework in which the regime is overthrown if enough citizens attack, however, the citizens are imperfectly informed about its strength and the regime can increase noise in their private information. Inspired by empirical findings in political science (King, Pan and Roberts, 2013; 2014; 2016), this paper illuminates (i) why regimes may aim to prevent collective action rather than criticism of the state per se, and (ii) why they might distract their citizens rather than try to persuade them. Unlike boosting the apparent strength by adding a bias, manipulating the noise is effective even if (i) the citizens can observe the regime’s manipulative action and (ii) the regime is no better informed about its strength than they are. I show that under these conditions the regime has no incentives to increase noise for the purpose of improving the citizen’s perception of the state per se. At the same time, minimising the size of all collective attacks may be an effective way of preventing regime change, especially when the citizens’ private information is intrinsically imprecise. Finally, I demonstrate that if citizens can coordinate better only at the cost of impeded information aggregation, the regime’s incentives to increase noise may become stronger.

11-08-16 Sifan Zhou, Harvard


Title: Non-Compete Agreements and the Career of PhDs.

Abstract: Non-compete agreements are legal contracts that employers use to restrain ex-employees from joining another firm or starting a new business in competition against them. While non-compete agreements protect employers’ investments in R&D and human capital, such protection may come at the costs of employees. I study the effects of non-compete agreements on doctorate recipients by analyzing the time-series and cross-sectional variations in the enforceability of these contracts in the United States. Using a difference-in-difference approach, I find that, for PhDs who work in the for-profit sector, tougher enforcement decreases their within-state job mobility. The reduced within-state job mobility is in part compensated by an increase in job changing across state borders, in line with the situation that non-compete agreements are more difficult to enforce across jurisdictions. Tougher enforcement also slows down the salary growths of these PhDs, consistent with a model in which general human capital is turned in to firm-specific human capital and workers lose bargaining power as they cannot threaten to leave for an outside offer. What is more, freshly minted PhDs are less likely to stay working in the state from which they received their degrees if the state has tougher enforcement. All these effects are consistently strong for engineering majors and less so for life sciences majors.

11-15-16 Ben Roth, MIT

Title: Keeping The Little Guy Down: A Debt Trap For Informal Lending


Microcredit and other forms of informal finance have so far failed to catalyze business growth among small scale entrepreneurs in the developing world, despite their high return to capital. This prompts a re-examination of the special features of informal credit markets that cause them to operate inefficiently. We present a theory of informal lending that highlights two of these features. First, borrowers and lenders bargain not only over division of surplus but also over contractual flexibility (the ease with which the borrower can invest to grow her business). Second, when the borrower’s business becomes sufficiently large she exits the informal lending relationship and enters the formal sector – an undesirable event for her informal lender. We show that in Stationary Markov Perfect Equilibrium these two features lead to a poverty trap and study its properties. The theory facilitates reinterpretation of a number of empirical facts about microcredit: business growth resulting from microfinance is low on average but high for businesses that are already relatively large, and microlenders have experienced low demand for credit. The theory features nuanced comparative statics which provide a testable prediction and for which we establish novel empirical support. Using the Townsend Thai data and plausibly exogenous variation to the level of competition Thai money lenders face, we show that as predicted by our theory, money lenders in high competition environments impose fewer contractual restrictions on their borrowers. We discuss robustness and policy implications.

Comments are closed.