Advisor-Advisee Networks in Economics
Lee, Sang Ah
MetadataShow full item record
This dissertation describes an advisor-advisee network in economics and studies the effects of that network on academic publication and dissemination of research subjects in the field of economics. I constructed an advisor-advisee network consisting of 5,283 economists who are either authors of articles published in five general-interest journals in economics (the American Economic Review, Econometrica, the Journal of Political Economy, the Quarterly Journal of Economics, and the Review of Economic Studies) from 1990 to 2010 or who were advisors of these authors during their graduate studies. In the first chapter, I analyze the effect of the advisor-advisee network on academic papers' publication process. I first study whether an author's paper has a higher probability of publication in a journal if an editor of that journal was the author's graduate thesis advisor, in other words, if the author has an advisor-advisee connection with the journal. I assume that two aspects of an advisor-advisee connection can affect a paper's publication process: (1) editors' bias in favor of or against their advisees (bias effect) and (2) editors' better information about their advisees' papers (information effect). I find that the probability that an author's paper is published in a journal with an advisor-advisee connection is 174% higher than the probability of publication of the paper in a journal without such a connection. The driving force behind this phenomenon appears to be the tendency of authors whose advisors become editors to write higher quality papers than other authors rather than the bias or the information effect. The bias effect and the information effect tend to work in opposite directions, and in the aggregate they do not change a paper's probability of publication. However, these effects do impact the quality of published papers. Papers published in a journal with an advisor-advisee connection are cited more often than other papers, supporting that the information effect is at work. Advisor-advisee connections increase publication of higher quality papers. In the second chapter, I report descriptive measures and analyze the properties of the advisor-advisee network. The analyzed features include the degree, the level of clustering, the size of the giant component of the network, and the average distance between two economists in the network. The measures indicate that the advisor-advisee network satisfies the "small-world" property, implying that the distance between any two economists in the network is small relative to the size of the network, while local clustering is high. I also fit the dynamic network formation model developed by Jackson and Rogers (2007) to the data and find that about half of the links are formed by random meeting and the rest are formed by preferential attachment. Therefore, the advisor-advisee network has a "rich-get-richer" property. Lastly, I test which characteristics of advisors have an impact on the number of advisees in the network. An advisor who has published a paper in one of the five journals mentioned above tends to have more advisees in the network than an advisor with no publication in the five journals. This result implies that advisors with good research skills are more successful in helping their advisees to develop research skills than are other advisors. In the third chapter, I study the effect of the advisor-advisee network on the spread of research subjects in economics. The dissemination of research subjects is traced through JEL codes of the articles published from 1991 to 2010 in the five journals mentioned above. I analyze how an economist's position in the advisor-advisee network, measured by his or her "centrality" in the network, is related to the volume of future published research on a subject similar to the economist's paper after the economist's paper is published. After an economist either with high degree centrality or with high diffusion centrality with short time periods for diffusion publishes a paper, the number of published papers on similar research subjects decreases in the following year. This result implies that immediately following a paper's publication, publication of papers with very similar research subject to that of the previous paper is reduced. If an economist either with high closeness centrality or with high decay centrality with low decay rate publishes a paper, the volume of publications on similar research subjects increases two or three years after the initial publication. This result can be interpreted as indicating that when an author working on a rather general research subject that can be addressed in various fields of economics publishes a paper, subsequent publication on the subject increases significantly two or three years after the initial paper's publication. These findings suggest that the dissemination pattern of research subjects has a connection with the structure of the advisor-advisee network.