The conference will be hosted at spatial.chat.
(If you need the password, please email email@example.com.)
Poster presentations will be held in all Halls and Seminar Rooms.
March 26, 2021Applied and Experimental
- Managing Participation and Transactions on a Digital Platform
Paul Belleflamme - UCLouvain, Muxin Li* - UCLouvain
- Coordination in the Network Minimum Game
Johannes Hoelzemann* - University of Toronto, Hongyi Li - UNSW Business School
Paper Link | Additional Link
- Positional Concerns and Social Network Structure: An Experiment
Armenak Antinyan - Wenlan School of Business, Gergely Horvath* - Duke Kunshan University, Mofei Jia - Xi'an Jiaotong-Liverpool University
Paper Link | Additional Link
- Keeping Up With the Joneses: Economic Impacts of Overconfidence in Micro-Entrepreneurs
Julia Seither* - Universidad del Rosario
- Classroom Networks and School Victimization: How Personality Protects Children Socially
Manuel Eisner - Cambridge, Denis Ribeaud - Zurich, Torsten Santavirta* - Uppsala, Miguel Sarzosa - Purdue
We provide a unified framework to compare the merits of various strategies that platforms can adopt to facilitate the interaction among users and generate value from these interactions. Using a micro-foundation of transactions among users (typically, buyers and sellers), we show the platform’s strategy over market involvements varies when it has different levels of ability to manage and abstract surplus from users. On top of charging membership fees, we allow the platform to facilitate price coordination among sellers, to make use of data analytics, or to sell first-party products directly to buyers. Our results are the following: a platform setting membership fees (i) encourages price coordination when it can only charge sellers, while keeping sellers from coordinating prices when it can charge both sides; (ii) is motivated to share information about buyers with sellers and enable sellers differential pricing when it can only charge sellers, while preventing sellers from using buyers’ information when it can charge both sides; (iii) is always willing to operate in the hybrid business mode when it can only charge sellers, while requiring the first-party product quality to be not sufficiently low when it can charge both sides.
Motivated by the problem of organizational design, we study coordination in the network minimum game: a version of the minimum-effort game where players are connected by a directed network. We show experimentally that acyclic networks such as hierarchies are most conducive to successful coordination. Introducing a single link to complete a network cycle may drastically inhibit coordination. Further, acyclic networks enable resilient coordination: initial coordination failure is often overcome (exacerbated) after repeated play in acyclic (cyclic) networks.
We experimentally study the positional concerns of individuals embedded in social networks. In the game, individuals compete for positional advantage with their direct neighbors by purchasing a positional good. The Nash equilibrium consumption is determined by the Katz-Bonacich centrality of the individual's network position, while the efficient outcome requires a lower consumption level. For most network positions, the gameplay converged to the Nash equilibrium, except for local centers that consumed less of the positional good than the Nash prediction. Despite this, the consumption of the positional good increased in the individual's centrality. The evolution of the gameplay was explained by the myopic best-response dynamics model.
Micro-firm owners have incomplete information about attainable incomes, and their relative performance standing. This might affect their beliefs about the quality of their business network and the returns to effort. Incorrect beliefs in turn, lead to sub-optimal choices that stunt firm growth. In this paper, I study whether feedback on relative performance standing affects beliefs and, as a consequence, firm output. I exploit data from a unique field experiment in Mozambique that provides novel causal evidence of the effect of performance rankings on firm outputs. One year after the intervention, at baseline high-performers are not affected by the treatment as the ranking contains only limited new information for them. However, low-performing firm owners at baseline significantly increase their revenues compared to low-performers in a control group. Treated low-performers close the performance gap to high-performing peers by 43%. This effect is, to a large degree, explained by significant increases in work hours and pro-social behavior. Exploiting variation in the observability of peer characteristics, I show that the treatment is particularly effective when subjects observe that the most successful peer is a woman. Low-performers that, additionally to their own ranking, observe a female top seller, outperform at baseline high-performers by 14%.
- Fast-Growing Networks and Distrust in Science (Even without Fake News)
Zaruhi Hakobyan - U Luxembourg, Christos Koulovatianos* - U Luxembourg
- Incomplete Information, Social Norms, and Beliefs in Networks
Theodoros Rapanos* - Södertörn Univ., Marc Sommer - Univ. of Zurich, Yves Zenou - Monash Univ.
Paper Link | Additional Link
- Local Interactions in a Market with Heterogeneous Expectations
Mikhail Anufriev - University of Technology Sydney, Andrea Giovannetti* - University of Technology Sydney, Valentyn Panchenko - University of New South Wales
- Relaxing Common Belief for Social Networks
Noah Burrell* - Michigan, Grant Schoenebeck - Michigan
We build a model of network dynamics with decision-making under incomplete information in order to understand the determinants of the observed gradual downgrading of expert opinion on complicated issues and the decreasing trust in science. We suggest a search and matching mechanism behind network formation of friends, claiming that the internet has made search and matching less costly and more intensive. According to our simulations, just combining the internet's ease of forming networks with (a) individual biases, such as confirmation bias or assimilation bias, and (b) people's tendency to align their actions with those of peers, can lead to populist dynamics over time through a vicious circle. Even without fake news, biases lead to more network homophily and, over time, more homophily leads to actions that put more weight on biases and less weight on expert opinion. Networks make fundamental biases be enhanced by peer-induced amplification factors, a finding suggesting that education should perhaps focus on mitigating fundamental biases by promoting evidence-based attitudes towards complicated social and scientific issues.
We develop a simple Bayesian network game in which players, embedded in a network of social interactions, bear a cost from deviating from the social norm of their peers. All agents face uncertainty about the private benefits and the private and social costs of their actions. We prove the existence and uniqueness of a Bayesian Nash equilibrium and characterize players’ optimal actions. We then show that denser networks do not necessary increase agents’ actions and welfare. We also find that, in some cases, it is optimal for the planner to affect the payoffs of selected individuals rather than all agents in the network. We finally show that having more information is not always beneficial to agents and can, in fact, reduce their welfare. We illustrate all our results in the context of criminal networks in which offenders do not know with certitude the probability of being caught and do not want to be different from their peers in terms of criminal activities.
We study the relation between the structure of information networks, trading behaviour, and market dynamics in an asset pricing model with heterogeneous expectations. We use a canonical Brock-Hommes (1998) model in which traders choose prediction rules based on their past performance. By modeling information sharing as a diffusion process via network, we characterize price volatility and threshold for instability for different networks and diffusion protocols. We establish an ordered relation between network connectivity, agents' trading behaviour, and price volatility for general classes of networks, including regular, random, and scale-free networks.
We propose a relaxation of common belief called factional belief that is suitable for the analysis of strategic coordination on social networks. We show how this definition can be used to analyze revolt games on general graphs, including by giving an efficient algorithm that characterizes a structural result about the possible equilibria of such games. This extends prior work on common knowledge and common belief, which has been too restrictive for use in understanding strategic coordination and cooperation in social network settings.
- Statistical Treatment of Social Network Data
Eric Auerbach - Northwestern, Rui Du* - Northwestern
- A Recursive Logit Model with Choice Aversion and Its Application to Transportation Networks
Austin Knies* - Indiana University, Jorge Lorca - Central Bank of Chile, Emerson Melo - Indiana University
- Risk Preferences and Risk Pooling in Networks: Theory and Evidence from Community Detection in Ghana
Daniel S. Putman* - Innovations for Poverty Action
Paper Link | Additional Link
- Escaping Saddle Points in Constant Dimensional Spaces, an Agent-based Modeling Perspective
Grant Schoenebeck - University of Michigan, Fang-Yi Yu* - Harvard University
We consider a choice problem in which a decision maker must choose between communities represented by a social network for treatment. For example, a firm must choose between locations to build a plant, a government agency must choose between organizations to audit, or a non-profit may choose between villages for an economic development project. Our choice problem is complicated by the fact that the networks representing the communities are high-dimensional (relative to the available data) and defined on heterogeneous populations that are difficult to directly compare. We propose a covariate-stabilized principal components-based decision rule for this problem. To demonstrate the use of this methodology, we consider the setting of Banerjee et al. (2013) in which social network data is used to predict villages with high rates of participation in a social program. A key finding is that while the authors' proposed diffusion centrality measure is highly predictive of program participation, there is additional information in the network data that can be used to choose villages with high participation rates.
We introduce a route choice model that incorporates the notion of choice aversion in transportation networks. Formally, we propose a recursive logit model which incorporates a penalty term that accounts for the dimension of the choice set at each node of the network. We make three contributions. First, we show that our model overcomes the correlation problem between routes, a common pitfall of traditional logit models. In particular, our approach can be seen as an alternative to the class of models known as Path Size Logit (PSL). Second, we show how our model can generate violations of regularity in the path choice probabilities. In particular, we show that removing edges in the network can decrease the probability of some existing paths. Finally, we show that under the presence of choice aversion, adding edges to the network can increase the total cost of the system. In other words, a type of Braess's paradox can emerge even in the case of uncongested networks. We show that these phenomena can be characterized in terms of a parameter that measures users' degree of choice aversion.
When risk preferences are heterogeneous, pooling covariate risk can lead to welfare improvements by shifting correlated shocks from more risk averse to less risk averse agents in exchange for a premium. However, the ability to pool covariate risk in this way depends crucially on whether agents prefer to share risk with others who have similar risk preferences. Do agents assortatively match on risk preferences? To investigate this question, I build a theoretical model of covariate risk pooling with heterogeneous risk aversion. I use rich data from four villages in southern Ghana to construct a bilateral risk sharing network and community detection algorithms to detect risk pooling communities, which bound the scope of risk pooling. Using econometric models of network formation, I estimate that individuals prefer to assortatively match in risk sharing networks. But, in detected communities the magnitude of assortative matching falls considerably. I compare the allocation of agents in communities to three benchmarks, including an optimal and worst-case scenario. In terms of assorative matching, I find that the observed networks deviate only slightly from optimal networks for this form of risk pooling.
We study a large family of stochastic processes that update a limited amount in each step. One family of such examples is agent-based modeling, where one agent at a time updates, so the state has small changes in each step. A key question is how this family of stochastic processes is approximated by their mean-field approximations. Prior work shows that the stochastic processes escape repelling fixed points and saddle points in polynomial time. We provide a tight analysis of the above question. For any non-attracting hyperbolic fixed point and a sufficiently small constant $\epsilon >0$, the process will be $\epsilon$-far away from the fixed point in $O(n\log n)$ time with high probability. We also show that it takes time $\Omega(n\log n)$ to escape such a fixed point with constant probability. This shows that our result is optimal up to a multiplicative constant. We leverage the above result and show that such stochastic processes can reach the neighborhood of some attracting fixed point in $O(n\log n)$ time with high probability. Finally, We show the power of our results by applying them to several settings: evolutionary game theory, opinion formation dynamics, and stochastic gradient descent in a finite-dimensional space.
- Desperateness in Contract Bargaining under Supply Chain Networks
Lei Hua* - UT Arlington, Alper Nakkas - UT Arlington, Kay-Yut Chen - UT Arlington, Xianghua Wu - UT Arlington
- Strategic Formation and Reliability of Supply Chain Networks
Victor Amelkin* - Amazon, Rakesh Vohra - Penn
Paper Link | Additional Link
- The Intraday Dynamics of the U.S. Tri-Party Repo Market
Matthew McCormick - OFR and U.S. Department of Treasury, Mark Paddrik* - OFR and U.S. Department of Treasury, Carlos Ramírez - Fed Board
- Shock Propagation in the Banking System with Real Economy Feedback
Andras Borsos* - Central Bank of Hungary/Central European University, Bence Mero - Central Bank of Hungary
Paper Link | Additional Link
- Monetary Plucking in a Network Economy
Amar Jyoti* - Indian Institute of Technology Madras, Arun K. Tangirala - Indian Institute of Technology Madras, Vipin P. Veetil - Indian Institute of Technology Madras
This paper theoretically and behaviorally studies contract bargaining in two-sided supply chain networks where retailers on the demand side purchase products from suppliers on the supply side. The retailers may have heterogeneous market valuations on the products ordered from the supply partner. In such a supply chain network, a retailer and a supplier must have a business relationship or "link" to bargain and trade with each other. However, a firm on one side of the supply chain network might not have a business relationship with every firm on the other side. Our experimental data suggest systematic deviations from the theoretical benchmark and reveal behavioral regularities on contracting behaviors. In particular, we show that players who link with more (or less) potential partners and/or who have more (or less) perceived values tend to earn more (or less) than expectation in games. We develop a new behavioral theory, referred to as desperateness theory, which explains and predicts the contract bargaining behaviors in two-sided supply chain networks. We demonstrate that firm(s) who link with less potential partner or who have less perceived values in the networks are more desperate of making contract agreement, and thus need to "sacrifice" part of the contract bargaining payoffs when bargaining with the corresponding "advantageous" firm(s). We also find evidence that the higher the total desperateness within the supply chain network, the lower the total supply chain profit.
Supply chains are the backbone of the global economy. Disruptions to them can be costly. Centrally managed supply chains invest in ensuring their resilience. Decentralized supply chains, however, must rely upon the self-interest of their individual components to maintain the resilience of the entire chain. We examine the incentives that independent self-interested agents have in forming a resilient supply chain network in the face of production disruptions and competition. In our model, competing suppliers are subject to yield uncertainty (they deliver less than ordered) and congestion (lead time uncertainty or, ``soft'' supply caps). Competing retailers must decide which suppliers to link to based on both price and reliability. In the presence of yield uncertainty only, the resulting supply chain networks are sparse. Retailers concentrate their links on a single supplier, counter to the idea that they should mitigate yield uncertainty by diversifying their supply base. This happens because retailers benefit from supply variance. It suggests that competition will amplify output uncertainty. When congestion is included as well, the resulting networks are denser and resemble the bipartite expander graphs that have been proposed in the supply chain literature, thereby, providing the first example of endogenous formation of resilient supply chain networks, without resilience being explicitly encoded in payoffs. Finally, we show that a supplier's investments in improved yield can make it worse off. This happens because high production output saturates the market, which, in turn lowers prices and profits for participants.
Repo markets offer a source of secure funding to financial firms and are an essential part of a well functioning financial system. However, as repo markets are over-the-counter, and information on demand/supply is generally opaque, mismatches in expected and realized collateral funding needs can create short-term stress. The expectation of funding-shortfalls can lead to borrower panic, seen through sudden funding rate spikes and fire-sale like liquidations. To assess how firms measure the potential for funding strains, we examine the U.S. overnight repo market's daily provision of funding to the financial system, using transaction level data from the tri-party report market. In doing so, we measure how features, such as collateral, time of day, and counterparty relationships, influence how funding is priced and distributed. Then we construct an equilibrium model of how information on funding is transmitted through the over-the-counter network of borrowers and lenders to explore how funding-shortfall panics arise.
In this paper we propose a model of shock propagation in the banking system with feedback channels towards the real economy. Our framework incorporates the interactions between the network of banks (exhibiting contagion mechanisms among them) and the network of firms (transmitting shocks to each other along the supply chain) which systems are linked together via loan-contracts. Our hypothesis was, that the feedback mechanisms in these coupled networks could amplify the losses in the economy beyond the shortfalls expected when we consider the subsystems in isolation. As a test for this, we embedded the model into a liquidity stress testing framework of the Central Bank of Hungary, and our results proved the importance of the real economy feedback channel, which almost doubled the system-wide losses. To illustrate the versatility of our modeling framework, we presented two further applications for different policy purposes: (i) We elaborated a way to use the model for SIFI identification, (ii) and we showed an example of assessing the impact of shocks originated in the real economy.
Milton Friedman argued that detrended business fluctuations are better modeled as the falling of output below a maximum potential cieling than movements above and below a mean. We find Friedman’s hypothesis to be true for aggregate output but not for disaggregate outputs. In other words, time series of US aggregate output has a ‘plucking-ceiling’ but sectoral outputs have no such ceiling. We develop a network economy model to explain the difference in the behavior of aggregate and sectoral time series data. Within our model, firms are related to each other via buyer-seller relations on a production network. Monetary shocks have a heterogeneous initial impact on firms depending on their liquidity positions. Monetary shocks percolate through the production network disturbing relative prices. Positive and negative monetary shocks can generate a decrease in aggregate output from its maximum potential level because the shocks create miscoordination. The output of some sectors however can increase in response to a monetary contraction, or decrease in response to a monetary expansion due to its position within the production network. Our model calibrated to the US economy with a novel data set generates data with a plucking-ceiling at the aggregate level but not at the granular level. Within our model, the plucking effect emerges despite fully flexible prices and wages. The empirical predictions and policy implications of our model are wholly different from models which depend on asymmetric wage stickiness to generate the plucking effect.
March 27, 2021Applied Micro
- Talent Goes to Global Cities: The World Network of Scientists’ Mobility
Luca Verginer* - ETH Zurich, Massimo Riccaboni - IMT School for Advanced Studies
Paper Link | Additional Link
- Spatial Network Coordination in Rural Markets
Moritz Poll* - Brown University
Paper Link | Additional Link
- Subcontractor Networks and Affiliated Private Values: Evidence from Oklahoma Bridge Contracts
Robert Press* - University of Oklahoma
- Boardroom Networks and Corporate Investment
Suyong Song* - Iowa, Jiawei (Brooke) Wang - Iowa
Global cities boast higher rates of innovation as measured through patent and scientific production. However, the source of the location advantage of innovation hubs is still debated in the literature, with arguments ranging from localized knowledge spillovers to network effects. Thanks to an extensive data set of individual scientist career paths, we shed new light on the role of scientist location choices in determining the superior innovative performance of global cities. We analyze the career paths of around two million researchers over a decade across more than two thousand cities around the globe. First, we show that scientists active in global cities are more productive in terms of citation weighted publications. We then show that this superior performance is in part driven by highly prolific scientists moving and remaining preferentially in global cities, i.e., central cities in the international scientist mobility network. The overall picture that emerges is that global cities are better positioned to attract and retain prolific scientists than more peripheral cities.
Market days are the pulse of rural economic and social life in many parts of the world. They are a way of spatially and temporally aggregating thin market demand and supply to ameliorate food security and price volatility in. Market days are also a complex coordination problem of assortative coordination (everybody trying to be in the same place at the same time) and non-assortative coordination (every village trying to have a different market day to its neighbors in order not to compete for participants). Fundamentally, coordination of markets determines who participates where and when in market exchange and are an important determinant of market integration. If neighboring cities compete over buyers and sellers on the same day of the week, the resulting dispersed market exchanges will suffer price volatility and unsteady product variety. This can hamper growth in the best of times, but in the worst of times it will deepen food insecurity and economic crisis. Understanding how the coordination works and where it fails can guide the way to making profound changes in poor people's lives. The coordination of these markets has an important social impact as it determines who meets on a regular basis, how social cliques and clusters form, within which groups/circles information diffuses or epidemics spread. Coordinated markets are a blueprint for the human networks that are likely to form through market participation and that facilitate knowledge diffusion and social learning. They also form the environment in which consumers and producers engage in costly search for one another and are a potentially important part of intra-national trade barriers.
This paper offers a theoretical framework associating subcontractor networks in procurement auctions to affiliated private costs of potential bidders. Based on the methodology by Li and Zhang (2010), I construct a model that allows for cost affiliation depending on firm-pair observables. The extension is used to test for entry affiliation caused by overlapping subcontractor networks in a sample of Oklahoma Bridge building contracts from 2004 to 2011. The empirical analysis finds a statistically significant presence of affiliation, driven by subcontracting networks, affecting firms' decision to buy for project plans. JEL Codes: D44, D85, L14, H57 Keywords: Auctions, Networks, Affiliation, Public Procurement
This paper investigates whether firms follow network firms and whether firms strategically herd. Using novel data on board members, we utilize board-interlocks as firm networks where two firms share at least one common board member and estimate network effects on firms' investment decisions. Our identification strategy to resolve the endogeneity issue is to adopt peers' peers' characteristics as legitimate instrumental variables. Empirical findings confirm significant network effects on firms' investment and show that firms follow more when the information quality of network firms is relatively high, which supports strategic herding behavior.
- Racial Social Norms among Brazilian Students: Academic Performance, Social Status and Racial Identification
Alysson Lorenzon Portella* - Insper, Charles Kirschbaum - Insper, Naercio Menezes Filho - Insper
Paper Link | Additional Link
- Representing Lotteries As Networks
Daniel Quiggin* - Georgia State University
- Spatial Dynamic Differential Game Models: a Leader and Followers with Multiple Activities
Hanbat Jeong* - The Ohio State University, Lung-fei Lee - The Ohio State University
- Spillovers, Homophily, and Selection into Treatment: The Network Propensity Score
Alejandro Sanchez Becerra* - University of Pennsylvania
Paper Link | Additional Link
This paper investigates the relation between social status and grades in Brazil and how it differs between racial groups, as well as the effect of academic performance on racial identity. White students display a positive correlation between grades and same-race social status, while this correlation is not significant for nonwhite students. However, nonwhite students observe a positive correlation between other-race social status and grades, while this relationship is concave for whites, increasing for low-achievers and decreasing for high-achievers. Overall, nonwhite students with high grades have better social status than those with worse academic performance. Regarding racial identification, we find that among all students good grades polarizes racial identification, although estimates are imprecise. Among high school students, "good grades whiten", what contradicts the hypothesis that they should be more susceptible to incentives from affirmative actions.
Bloch, Jackson and Tebaldi (2021) point out correspondences between network centrality measures and a variety of problems, including evaluation of consumption streams. In this paper, I look at the closely related question of preferences over lotteries. More specifically, I consider whether lotteries can be represented as network graphs. I show that any lottery with rational probabilities can be represented as a network with a centrality-based allocation rule.
This paper introduces a new spatial dynamic panel data simultaneous equations (SDPDSE) model explaining decision-making processes of two types forward-looking agents: a leader and multiple followers. In an empirical study, they represent central and local governments. In each period, the leader decides levels of grants for followers, who simultaneously choose their multiple expenditures. Hence, our model's purpose is to account for the leader's resource allocations for followers and the intertemporal competitions of followers. With a micro-foundation, we establish a dynamic network game for agents with multiple activities. Derived optimal actions lead to a spatial econometric model. For recovering parameters in agents' payoffs, we implement a quasi-maximum likelihood (QML) method. We examine policy interactions among U.S states' expenditures on public welfare, housing and community development, and relations among those expenditures and grants from the central government. We detect positive spillovers in states' public welfare expenditures; the two expenditures behave as substitutes; and a significant positive effect of a federal grant on a state's public welfare expenditure.
Propensity score matching is often used to estimate treatment effects when there is selection on observables; however, it fails to identify causal effects when one person's treatment affects another's outcome. This phenomenon is known as spillovers. I propose a novel network propensity score matching approach that identifies both the average treatment effects and the average spillover effects between individuals. My approach is grounded on an endogenous model of network formation with spillovers on the outcome. This methodology can be used to identify causal effects for individuals with similar observables, analogous to the propensity score. I then propose estimators that are consistent and asymptotically normal for settings with multiple networks. I apply my methodology to two empirical examples. First, I study the effects of an intervention on political participation in Uganda where I find evidence of spillovers on non-participants. Second, I evaluate a microfinance adoption intervention in India, and find large treatment effects but limited spillovers effects. In some extensions of the method, I show how to conduct robustness checks and how to interpret the network propensity score in stratified multi-stage experiments.
- Resource Sharing in Endogenous Networks
Philip Solimine* - Florida State University, Luke Boosey - Florida State University
- An Equilibrium Concept for Multi-Network Formation Games
William Walsh* - Pennsylvania State University
- Propensity to Trust and Network Formation
Juan Camilo Cárdenas - Universidad de los Andes, Danisz Okulicz - National Research University, Davide Pietrobon* - Université de Genève, Tomás Rodríguez - Universidad de los Andes
- Inheritance and Evolution of Networks across Generations
Sumit Joshi - George Washington University, Ahmed Saber Mahmud* - Johns Hopkins University, Hector Tzavellas - George Washington University, Sudipta Sarangi - Virginia Tech University
How, and why, do people share resources with each other in situations when it may seem irrational? In this paper, we examine behavior in a voluntary resource sharing environment that incorporates endogenous network formation. Using a laboratory experimental implementation of repeated play in this information-rich decision setting, we examine the effects of a simple reputation system and discuss information design. Reduced-form estimates find significant effects of the information treatment on a number of key outcomes such as efficiency, reciprocity, and decentralization. To further understand the driving causes of these observed changes in behavior, we develop a discrete-choice framework to identify the structure of social preferences in this setting, and use unsupervised machine learning methods to disentangle the effects of the information environment on behavior and substitution patterns between strategies.
A multi-network, or multi-layer network, is a set of nodes that can be connected via multiple types of links. In this paper I define a notion of pairwise stability for multi-networks that is a trivial extension of pairwise stability of a single network. When the utility of each agent derived from the multi-layer network is additively separable across networks, then pairwise stability of all individual networks implies pairwise stability of the multi-network. This notion of pairwise stability needs to be refined when different network types are complementary. Pairwise stability does not account for the simultaneous addition or removal of multiple link types. As a result, a multi-layer network may be pairwise stable, even if two agents could rearrange the links between them and achieve a Pareto improvement. I define a new notion of network stability, multi-link pairwise stability, which captures this possibility. Finally, I illustrate the implications of link type complementarity for efficiency and stability in a symmetric connections model. The insight arising from this analysis is that link type complementarity generates multi-layer network configurations that are efficient and stable when the same configurations of individual networks are not.
We study how trust affects network formation using an empirical strategy that is immune to reverse causality. We employed a standard trust experiment and survey questions to obtain measures of trust for 72 freshman students before they had significant chances to get to know each other and socialize. After four months, we elicited five social networks capturing different relationships between the students. We estimate network formation models to identify how trust affects link formation for the networks elicited. We find that the effect of trust on link formation is insignificant, and small relative to the impact of several pairwise characteristics. In particular, the effect of homophily in socio-economic background is significant and 3 to 4 times bigger than the effect of trust.
We consider the evolution of a network across generations where each generation lives only for two periods - youth and adulthood. At the beginning of their youth, each generation inherits their parents' network and accumulates human capital. The inherited network induces human capital formation in childhood. In their adulthood, professional life begins, and individuals form another network to produce goods and services. These two networks are interdependent as human capital created in the first period positively impacts the outcome of the second. The present endeavor explains why inequality can be entrenched in a particular society across generations while not in others.
- Understanding Spillover of Peer Parental Education: Randomization Evidence and Mechanisms
Bobby Chung* - University of Illinois, Jian Zou - University of Illinois
- The Peer Effects of Persistence on Students‘ Cognitive and Non-cognitive Outcomes
Jian Zou* - University of Illinois at Urbana-Champaign
- Understanding Peer Effects in Educational Decisions: Theory and Evidence from a Field Experiment
Karen J. Ye* - Queen's University
Paper Link | Additional Link
- Network Against Cancer: Science
Adhen Benlahlou* - UJM
We utilize random assignment of students into classrooms in China middle schools to study the mechanisms behind the spillover of peer parental education on student achievement. Analyzing the China Education Panel Survey, we find a causal relationship between classmates' maternal education and student test score. In addition to the conventional peer effect and teacher response channel, we identify the adjustment of parenting style by mothers as another significant mediating factor. The changes in parenting style and parental investment on time also differ by the student or family background, leading to heterogeneous spillovers on test scores.
Little is know about the effects of student peers’ noncogntive skills on human capital formation. This paper studies the peer effects of persistence, one facet of personality traits reflects the ability to persevere when facing challenges and setbacks, on students’ cognitive and non-cognitive outcomes. By exploiting student-classroom random assignments in high schools and a nationally representative sample in China, I find substantially positive impacts of peer persistence on students’ academic and cognitive outcomes. Moreover, further results show that having peers with higher persistence enduringly leads to improved academic and cognitive outcomes, and increased mental status one academic year later. While there is no impact of peer persistence on students’ own persistence, I find suggestive evidence of increased disciplined behaviors and friendship formation with “good” but not “bad” peers as potential mechanisms. I use a simple model to illustrate that the friendship sorting, as an equilibrium outcome, could be driven by the fact that students sharing similar persistence are more likely to make friends and is socially efficient in maximizing the total human capital production.
While a large literature documents the presence of peer effects in teenage decision-making, researchers know very little about the underlying mechanisms. In this paper, I focus on the decision by high school students to participate in an educational program. I develop a theoretical model based on Brock and Durlauf (2001) with two channels of peer effects: social learning (where a peer’s decision is informative about the value of a program) and social utility (where a peer’s participation directly changes the benefits or costs of a program). I conduct a field experiment in three Chicago high schools to disentangle the two channels. In the experiment, I measure students’ sign-up rates for a college application assistance program where I randomize (a) whether a student sees a peer’s decision, and (b) which type of peer’s decision they see. I find large peer effects in the participation decision that are entirely driven by seeing a peer choose not to participate – seeing a peer choose “No” decreases the sign-up rate by 26.9 percentage points. The peer effects are driven by social utility, and seeing a peer choose “No” informs students about the social norms of participation. In this context, smart students’ decisions are especially influential. Further, while students want to conform to the social norm, they have very biased beliefs about (they drastically underestimate) their peers’ participation. I estimate my model and combine the structural estimates with collected school social network data to run a policy counterfactual. I find that when there are negative peer effects and costly initial adoption, programs targeting smart students may have decreased sign-up rates compared to programs targeting highest need students.
I study the impact of research collaboration in coauthorship network on total research output. Authors's links create direct and indirect spillovers across the network. I characterize the interior equilibrium efforts level in research when agents spend efforts in multiple projects for every network structure. In work in progress, I structurally estimate this model using a unique dataset of publications linked to thoracic cancer. I will use my estimates to study the impact of different subsidies schemes, and empirically rank author according to he production-maximizing subsidies they should receive.