Experiments in the armchair: a history of microeconometrics and program evaluation at Princeton

• Ano de defesa: 2021
• Autor(a) principal: Brackmann Netto, Arthur
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Avaliação de programas; Experimentos naturais; Guerra à pobreza; Microeconometria; Microeconometrics; Natural experiments; Observações passivas; Passive observations; Princeton; Program evaluation; War on poverty
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/12/12140/tde-10012022-122433/

This thesis joins the recent efforts of econometricians and historians to tell the history of microeconometrics. Because economists mostly do not have any control over the collection of the data they use in their applied analyses, econometrics became a tool of passive observations a concept that epitomizes the accountability of econometrics for nonexperimental data. In the thesis I add to that literature by connecting the rise of program evaluation in the US government with Princeton University. To do so, I discuss how, since the 1950s, economists knew that they cannot intervene in the data they analyze and have to be prepared to deal with any inherent problem of such data. Using biobliometric methods and secondary sources, I argue that microeconometrics emerged from a change in the understanding of passive observations as simultaneity to passive observations as omitted variables. Although this change may seem purely internal to econometrics, I argue that a micro history of US governmental institutions for program evaluation and Princetons Industrial Relations Section is necessary in the history of microeconometrics. I show how, in the 1970s, poverty was on the rise in the United States and economists and econometricians were more concerned with it than with the theoretical and philosophical aspect of passive observations. Within the government, program evaluation went from a qualitative topic in social sciences to a quantitative problem inside economics. In order to analyze this change, the thesis uses natural language processing methodologies, archives from the US government and oral histories of the members of the Office of Economic Opportunity. I then show how program evaluation traveled from the government to Princetons Industrial Relations Section. Looking at the grassroots of the department, I contrast the solutions of two Princeton young scholars to the problem of omitted variables, James Heckmans selection model and Orley Ashenfelters difference-in-differences estimator, to demonstrate that the clash between randomistas and structural modelers is an artifact created after those two solutions to the same problem. Evidence from bibliometrics (using the innovative methodology of related networks), primary and secondary sources demonstrates that the solutions were not concurrent in the early days of omitted variables, and that the present-day clash resulted from a change in the contexts of the two main authors. Finally, using nine interviews conducted with actors in the Industrial Relations Section of the 1970s and 1980s, the thesis analyzes how natural experiments developed gradually inside the department, without revolutions, turns or shifts. The thesis concludes showing how the ordinary business of the academic lives of Ashenfelter and his students ended up transforming economics at large.