상세정보
Export to Refworks부가기능
Peer Effects: Theory and Measurement
상세 프로파일
| 자료유형 | 학위논문 |
|---|---|
| 서명/저자사항 | Peer Effects: Theory and Measurement. |
| 개인저자 | Bykhovskaya, Anna. |
| 단체저자명 | Yale University. Economics. |
| 발행사항 | [S.l.]: Yale University., 2019. |
| 발행사항 | Ann Arbor: ProQuest Dissertations & Theses, 2019. |
| 형태사항 | 149 p. |
| 기본자료 저록 | Dissertations Abstracts International 81-04A. Dissertation Abstract International |
| ISBN | 9781088382363 |
| 학위논문주기 | Thesis (Ph.D.)--Yale University, 2019. |
| 일반주기 |
Source: Dissertations Abstracts International, Volume: 81-04, Section: A.
Advisor: Phillips, Peter C.B. |
| 이용제한사항 | This item must not be sold to any third party vendors.This item must not be added to any third party search indexes. |
| 요약 | Peer effects are pervasive in human society, existing across the entire socio-politico-economic spectrum. Their presence affects decisions at the individual level and in policy making at industry and societal levels. My research is concerned with explaining the effect of existence vs. non-existence of these linkages, as well as their magnitude and potential emergence over time. I focus on two instances that are relevant in applications: matching markets and social or trade networks. The thesis contributes to the development of the micro-theory of equilibrium in the presence of peer effects and to the advancement of the econometric techniques to analyze the evolution of networks over time.The first part of the dissertation studies the effects that interactions may play in the prediction and estimation of socio-economic systems such as networks. Many interactive structures are not fixed over time and may be expected to evolve in systematic ways. The growing availability of time series data for social and economic networks makes it possible to model and estimate this evolution. I develop such a modeling framework to study network evolution together with econometric procedures to estimate the parameters of the system. I build a multivariate time series system whose components can be interpreted as weighted edges of some network, and I treat the number of time periods as large compared to the size of the network. The model is nonparametric with respect to the distribution of the errors and specifies the temporal evolution of a weighted graph that combines classical autoregression (AR) with non-negativity, a positive probability of vanishing, and peer effect interactions between weights assigned to edges in the process.My main results provide criteria for stationarity/explosiveness of the network evolution process and techniques for estimation of the parameters of the model and for prediction of its future values. The asymptotic theory is much more complex than that of the classical AR model and the results are novel in nonlinear time series modeling. Due to censoring, the naive ordinary least squares estimator for the parameters of the model is unreliable. But in this network AR setting an estimator based on minimization of the absolute deviations is always consistent and is asymptotically normal in the stationary cases.Natural applications arise in networks of fixed number of agents, such as countries, large corporations, or small social communities. The paper provides an empirical implementation of the approach to monthly trade data in European Union. Computations show that the new model leads to improved performance over the most naive (but standard benchmark) prediction that "tomorrow=today" in network evolution. Overall, the results confirm that incorporating non-negativity of dependent variables into the model matters and incorporating peer effects leads to the improved prediction power.The second part of the dissertation looks at the role that peer effects play in the school matching problem. In economic theory the relationship between schools or colleges and students is usually modeled as a two-sided matching problem. Yet a challenge in matching models with peer effects is equilibrium existence. Despite much progress in the study of stable matchings in the presence of peer effects, there is still no simple criterion for existence that applies to a wide class of models.My research takes a step in this direction. I consider an economy where agents are characterized by their types (e.g. SAT scores) and schools are characterized by their values (e.g. teaching quality) and capacities. Agents and schools are divided into groups, so that going to a school outside one's group may be associated with additional costs or even be prohibited. For example, schools sometimes only accept those who live in specified areas. Students receive utility from the school per se and from classmates. A key role in the analysis is played by a directed graph that governs the possibility to move from one group to another. I find that a sufficient condition for a stable matching to exist is that the graph should not have (directed or undirected) cycles. Under these conditions, I construct a polynomial time algorithm for finding a stable matching. Acyclicity ensures that the algorithm must terminate. While non-directed cycles do not play a major role in previous work, here they are of the same importance as directed cycles. Furthermore, I show that if the graph has a cycle, then there exist other economic parameters (types, costs and so on) so that no stable matching exists. |
| 일반주제명 | Economics. |
| 언어 | 영어 |
| 바로가기 | 
: 이 자료의 원문은 한국교육학술정보원에서 제공합니다. |
