Strategy Seminar 2023-2024

Consider an economy with equal amounts of N types of goods, to be allocated to agents with strict quasi-convex preferences over lotteries. We show that all feasible and Pareto efficient allocations give almost all agents a binary lottery.
Even if all preferences are the same, some identical agents  necessarily receive different lotteries. In a  continuum economy, a feasible and efficient allocation satisfying No-Envy exists. Such allocations yield all agents binary lotteries. Assuming the reduction of compound lotteries axiom, social welfare deteriorates by first randomizing over these binary lotteries. Full ex-ante equality can be achieved if agents satisfy the compound independence axiom.

In the interwar period, some American firms began  to invest in basic scientific research.
Using newly assembled firm-level data from the 1920s and 1930s, we find that companies invested in research because inventions increasingly relied on science, but American universities lagged behind both Europe and the scientific frontier.
Firms close to the frontier, relying on disciplines which were underdeveloped in American academia, were likely to invest in research, especially if they were large and operated in concentrated industries (could internalize the benefits). Corporate science seems to have paid off, resulting in novel patents and high market valuations for those engaged in research.

Managers’ use of remote monitoring software increased following the transition to working from home during the Covid-19 pandemic to compensate for reduced observability. Higher observability entails quantitative measures not directly related to productivity, potentially incentivizing workers to prioritize quantity over quality. For example, office workers may increase observable work hours by directing effort inefficiently. Observing the number of completed tasks incentivizes workers to perform many meaningless tasks rather than prioritize productive ones. We design an experiment where workers can allocate effort based on perceived task difficulty and manipulate the structure of the signal to the manager. We show theoretically that quantitative information in the signal distorts incentives. In equilibrium, workers prioritize productive tasks less, reducing overall productivity. Our results confirm that removing quantitative information from the signal increases productivity by shifting workers’ strategies. Enriching the signal with quantitative information, however, does not have the opposite effect.

We study social learning, conformity, and differentiation within and across groups (Right and Left political blocs; Indians and Pakistanis; Scots and English). Our experiments cross-randomize observed ingroup and outgroup norms in several domains, including public opinion, investments in stocks, social values, and factual assessments. Results show that, overall, individuals follow their ingroup much more than they follow their outgroup. Behavior varies systematically across individuals: those who strongly identify with their group not only follow ingroup norms more, but may choose the opposite of observed outgroup norms. Such counteracting seems to rest on intergroup animosity and on ingroup-biased perceptions of others' knowledge. Thus, when identities diverge, the same information can have contradictory effects.

Can campaign finance regulation curb the political influence of economic actors? In this article, we identify a new factor that may hinder its effectiveness—the social structure of organizations. We argue that such regulation creates cooperation dilemmas in firms’ leadership and propose that a specific feature of organizations—family ties—help solve such problems. We evaluate this argument by studying a Supreme Court ban on corporate contributions in Brazil. Using a difference-in-differences design and data on family ties in Brazilian public companies, we show that, following the ban, members of firms’ controlling families substitute individual for corporate contributions. Furthermore, we document the presence of peer effects in the contribution behavior of family members, suggesting that family ties transmit influence. These bifurcated effects illustrate how organizational structure can be a source of de facto power by limiting the effectiveness of programmatic reforms, and thus contain a cautionary tale for policymakers.

This paper investigates the impact of an emergency department (ED) reform on congestion levels in a hospital with multiple interchangeable internal medicine departments. The reform switched patient routing from a “first-available-bed”  approach to equally distributing patients across departments. Comparing outcomes before and after the reform with baseline trends from the previous year, we find that equal-load routing results in a 25% lower average ED wait times and a 20% lower admission length of stays without increases in readmission or mortality rates. Using a queuing theory model in which departments choose effort levels, we show that these findings are consistent with a strategic response of hospital departments to free-ride incentives induced by first-available-bed routing. Our results imply that a balanced routing improves resource allocation and overall hospital efficiency.

Large and persistent disparities in labor market conditions across regions are well-documented even within small geographic areas. While various place-based policies have been implemented to address these disparities, little attention has been given to policies incentivizing commuting from disadvantaged areas to adjacent, tighter labor markets. This paper studies a unique policy implemented as a Randomized Controlled Trial (RCT) in Israel, providing a short-term employment subsidy to residents of disadvantaged localities conditional on working outside their locality of residence. Using rich administrative and survey data, we find positive effects on employment and earnings, with the most substantial effects observed among individuals with the weakest initial labor market attachment. For this group, the program increased employment by 9% even after 24 months, well beyond subsidy exhaustion. We show that this persistence is related to the workplace location condition by comparing the policy to a standard, unconditional employment subsidy. Overall, our findings indicate that the policy alleviated barriers to participation in labor markets outside the locality of residence.

Tipping is a complex phenomenon driven by social pressure, and cutting across various stakeholders – firms, customers, and workers. Analyzing the long-run impact of policies related to tipping is therefore challenging. To facilitate such analysis, we develop a modeling framework, in which a tipping norm is formed endogenously in a market comprised of a price-setting firm offering service to potential customers. Customers choose whether to consume the service or not, and if yes, how much to tip the server afterwards. With tipping, customers show reciprocity to the server by sharing a fraction of their surplus, but also undergo social pressure to comply with the prevailing norm of tipping. This tipping norm is shown to evolve endogenously through a dynamic process of sequential market adjustments over time: the average tip determines in each time period the tipping norm for the next period, causing the firm to adapt its price and customers to adapt their tips accordingly. We show that, in the short run (in between time periods), the average tip-to-price ratio increases, converging in the long run to a stable market-equilibrium outcome. Characterization of this equilibrium outcome allows us to derive qualitative results about the long-run impact of different exogenous factors on tipping: We find that the equilibrium tip-to-price ratio increases when customers are more sensitive to social pressure, when their feelings of eciprocity grow stronger, when their range of service valuations spreads out, or when they consider the service more valuable. Building on this framework, we further investigate several economic implications of tipping pertaining social welfare, labor cost, and service quality, thus, uncovering incentives and trade-offs to which the tipping mechanism give rise, from the firm’s, the worker’s, and the customer’s perspectives.

We examine friendships and study partnerships among university students over several years. At the aggregate level, connections increase over time, but homophily on gender and ethnicity is relatively constant across time, university residences, and different network layers. At the individual level, homophilous tendencies are persistent across time and network layers. Furthermore, we see assortativity in homophilous tendencies. There is weaker, albeit significant, homophily over malleable characteristics—risk preferences, altruism, study habits, and so on. We find little evidence of assimilation over those characteristics. We also document the nuanced impact of network connections on changes in Grade Point Average.

The talk explores the concept of partial identification, a critical methodological approach in economic research where incomplete data or model specification restrict what we can learn or identify. The talk is structured into two primary segments.

The first half of the talk will focus on the formalism underlying partial identification. I will explain the theoretical foundations that characterize this approach, emphasizing how it differs from point identification. I will further explain the sources of partial identification, which typically arise from missing data, measurement error, or incomplete theoretical models. I will also discuss the statistical tools used to handle such scenarios. In the second half, I will discuss empirical applications, specifically examining how partial identification is utilized within the fields of labor economics and industrial organization. Through leading examples, this segment will showcase real-world scenarios where partial identification has been used. Case studies will include analyses of surveys with missing or corrupted data as well as counterfactuals. In industrial markets, we will survey estimation of games, auctions, and incomplete preferences.
 

This paper studies the impact of market access on individualistic cultural traits, focusing on the expansion of the railroad network and population growth in the U.S. from 1850 to 1920. A county-level analysis leverages excess variation in counties' market access that is orthogonal to the expansion of local railroad networks. An individual-level analysis uses a difference-in-differences approach to estimate the impact of migrating to locations with higher market access relative to lower market access. Both empirical strategies reveal that market access fosters individualistic cultural traits. The findings indicate that this relationship is due to the direct influence of commerce on culture and psychology, rather than indirect effects through market-induced changes in population diversity, economic development, access to information, or legal institutions.
 

 

Crowdsourcing contests have recently
gained popularity due to the  availability
of online platforms through which contest
organizers  engage an external crowd of
 solution providers. Because crowd size is
 key to contest effectiveness, considerable
 attention has been devoted to studying
 how participation decisions are affected
 by contest design, often focusing on the
 monetary reward as a key determinant of
 endogenous entry. While empirical
 findings frequently confirm theory-based
 predictions of a positive effect of reward
 size on participation, the literature also
 provides evidence of no effect or even a
 negative effect, possibly arising from
 expectations of higher competition
 intensity with higher rewards.
 We address this important inconsistency
 in the crowdsourcing contest literature
 by formulating and testing two alternative
 hypotheses about the effect of reward on
 participation. Whereas the first
 hypothesis  draws on expected utility
theory to  hypothesize that the effect
is concave- shaped, the second
hypothesis draws on  lexicographic
 decision theory to  hypothesize that
 the effect is U-shaped.
We tested these hypotheses in five large-
scale online experiments, wherein a total
 of 4716 registered users had the option
 to  participate in multiple contests by
 exerting  real effort over a period of up
 to two hours  to win rewards that
 ranged from $10 to  $90.  These
experiments provide strong,  consistent,
 and robust evidence in support  of the
 lexicographic hypothesis, independent
 of the number of concurrent contests
 and  the contest success function.
 Our main finding is that participation
 decisions  are mainly affected by
 whether rewards  are extreme or not
 rather than by their  monetary value

R&D and innovation are thought to have large positive externalities via knowledge spillovers and human capital. Governments often address private under-provision with R&D grants, but identifying high social return firms is a challenge. This paper leverages features of the grant making process, using propensity score matching and regression discontinuity designs to evaluate the effectiveness of the Israel Innovation Authority’s (IIA’s) grants to startups. Three years out, grants increase firm survival, the probability of attaining private follow-on funding, and the likelihood of positive sales. Grants are effective where market frictions are large but crowd out private funds where frictions are small. These results suggest a strategy of targeting firms in geographic areas and sectors with less available private funding.

Platforms use purchase histories to profile customers, create consumer segments, and disclose them to sellers. Sellers target price offers to these segments, generating new data that enables further profiling. We characterize the platform's ability to learn consumers' valuations using only information design in constructing the segments disclosed to sellers. We then evaluate the implication for market outcomes. We find that there is a threshold so that the platform cannot accurately profile consumers with valuations above it but can do so for those with valuations below it. The threshold is the seller's optimal uniform price in the no-information case. As a consequence, the use of purchase data to profile customers increases total welfare without harming consumers.

We analyze the optimal disclosure (discrimination) policy for a regulator aiming to maximize investment in human capital in a setting where workers can invest to become productive. A worker's cost for increasing the probability of becoming productive depends on her (privately observed) type. The worker's wage equals an employer's equilibrium posterior belief regarding whether she is productive. This belief depends on the employer's direct (partial) information about the worker's realized productivity — conveyed via a partially informative test — and about the employer's information about the worker type. We characterize the regulator's optimal disclosure rule — a signal about the worker types revealed to the employer. Specifically, we show that if the mean of the worker types distribution is sufficiently high, then the regulator's optimal disclosure rule is uninformative. We interpret this result in the following way; the optimal regulation is not allowing the employer to condition wages on information about the worker's type - banning the use of statistical discrimination.