{"id":{"repo_id":"toronto-retro","oai_identifier":"oai:utoronto.scholaris.ca:1807/129792"},"canonical_url":"https://search.dev.ndltd.org/etd/toronto-retro/oai:utoronto.scholaris.ca:1807/129792","repository":{"repo_id":"toronto-retro","name":"University of Toronto","base_url":"https://utoronto.scholaris.ca/server/oai/request"},"display":{"title":"Essays in Computational Econometrics","abstract":"This thesis consists of three chapters, each in the form of a self-contained essay. In contemporary econometric research, computational limitations are often binding. This can be true in theoretical work, where restrictive assumptions are imposed to maintain computational tractability, or in empirical work, where large real-world datasets present computational challenges. Each essay in this thesis presents methods that relate to computational problems arising in econometric theory or applications. In the first essay, an improved Bayesian method for probabilistic record linkage in large matching problems, with an accompanying implementation in R/C++, is presented. The method refines the modelling of comparison data relative to previous methods, allowing the distribution of disagreement among non-matched pairs to be record-specific, leading to dramatic performance improvements in a large application, with accompanying computational improvements. The second essay considers partial identification of counterfactuals in a broad class of models with discrete outcomes and covariates and develops a procedure for computing the identified interval using several new computational tools, including a new algorithm for enumerating the cells induced by hyperplane arrangements. The third essay examines the problem of inference on the value function of a linear program where the right-hand side parameters are random. A tractable method for developing confidence intervals is presented that has asymptotically exact coverage and shows good performance in finite samples.","abstract_html":"This thesis consists of three chapters, each in the form of a self-contained essay. In contemporary econometric research, computational limitations are often binding. This can be true in theoretical work, where restrictive assumptions are imposed to maintain computational tractability, or in empirical work, where large real-world datasets present computational challenges. Each essay in this thesis presents methods that relate to computational problems arising in econometric theory or applications. In the first essay, an improved Bayesian method for probabilistic record linkage in large matching problems, with an accompanying implementation in R/C++, is presented. The method refines the modelling of comparison data relative to previous methods, allowing the distribution of disagreement among non-matched pairs to be record-specific, leading to dramatic performance improvements in a large application, with accompanying computational improvements. The second essay considers partial identification of counterfactuals in a broad class of models with discrete outcomes and covariates and develops a procedure for computing the identified interval using several new computational tools, including a new algorithm for enumerating the cells induced by hyperplane arrangements. The third essay examines the problem of inference on the value function of a linear program where the right-hand side parameters are random. A tractable method for developing confidence intervals is presented that has asymptotically exact coverage and shows good performance in finite samples.","abstract_has_math":false,"creators":["Stringham, Thomas Kent"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Economics","school":null,"contributors":[],"advisors":["Gu, Jiaying","Eli, Shari"],"committee_chairs":[],"committee_members":[],"year":2023,"date_issued":"2023-11","date_published":"2023-11","updated_at":"2026-07-27T21:28:07Z","subjects":["Computation","Econometrics"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/1807/129792","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Gu, Jiaying","Eli, Shari"]},{"key":"dc:contributor.department","label":"Department","values":["Economics"]},{"key":"dc:creator","label":"Author","values":["Stringham, Thomas Kent"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2023-11"]},{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2023-11-13T16:07:11Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2023-11-13T16:07:11Z"]},{"key":"dc:date.issued","label":"Date","values":["2023-11"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Computation","Econometrics"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://hdl.handle.net/1807/129792"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This thesis consists of three chapters, each in the form of a self-contained essay. In contemporary econometric research, computational limitations are often binding. This can be true in theoretical work, where restrictive assumptions are imposed to maintain computational tractability, or in empirical work, where large real-world datasets present computational challenges. Each essay in this thesis presents methods that relate to computational problems arising in econometric theory or applications. In the first essay, an improved Bayesian method for probabilistic record linkage in large matching problems, with an accompanying implementation in R/C++, is presented. The method refines the modelling of comparison data relative to previous methods, allowing the distribution of disagreement among non-matched pairs to be record-specific, leading to dramatic performance improvements in a large application, with accompanying computational improvements. The second essay considers partial identification of counterfactuals in a broad class of models with discrete outcomes and covariates and develops a procedure for computing the identified interval using several new computational tools, including a new algorithm for enumerating the cells induced by hyperplane arrangements. The third essay examines the problem of inference on the value function of a linear program where the right-hand side parameters are random. A tractable method for developing confidence intervals is presented that has asymptotically exact coverage and shows good performance in finite samples."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Ph.D."]},{"key":"dc:title","label":"Title","values":["Essays in Computational Econometrics"]}]}],"canonical_facts":{"dc:contributor.advisor":["Gu, Jiaying","Eli, Shari"],"dc:contributor.department":["Economics"],"dc:creator":["Stringham, Thomas Kent"],"dc:date":["2023-11"],"dc:date.accessioned":["2023-11-13T16:07:11Z"],"dc:date.available":["2023-11-13T16:07:11Z"],"dc:date.issued":["2023-11"],"dc:description.abstract":["This thesis consists of three chapters, each in the form of a self-contained essay. In contemporary econometric research, computational limitations are often binding. This can be true in theoretical work, where restrictive assumptions are imposed to maintain computational tractability, or in empirical work, where large real-world datasets present computational challenges. Each essay in this thesis presents methods that relate to computational problems arising in econometric theory or applications. In the first essay, an improved Bayesian method for probabilistic record linkage in large matching problems, with an accompanying implementation in R/C++, is presented. The method refines the modelling of comparison data relative to previous methods, allowing the distribution of disagreement among non-matched pairs to be record-specific, leading to dramatic performance improvements in a large application, with accompanying computational improvements. The second essay considers partial identification of counterfactuals in a broad class of models with discrete outcomes and covariates and develops a procedure for computing the identified interval using several new computational tools, including a new algorithm for enumerating the cells induced by hyperplane arrangements. The third essay examines the problem of inference on the value function of a linear program where the right-hand side parameters are random. A tractable method for developing confidence intervals is presented that has asymptotically exact coverage and shows good performance in finite samples."],"dc:description.degree":["Ph.D."],"dc:identifier.uri":["http://hdl.handle.net/1807/129792"],"dc:subject":["Computation","Econometrics"],"dc:title":["Essays in Computational Econometrics"],"dc:type":["Thesis"]},"updated_at":"2026-07-27T21:28:07Z"}