Grinis, Inna
(2017)
Essays in applied computational economics.
PhD thesis, London School of Economics and Political Science.
Abstract
This thesis presents four distinct essays that lie at the intersection of economics and computation.
The first essay constructs an abstract framework for defining skills gaps, mismatches and shortages geometrically and thinking about these phenomena in a unified, formal way. It then develops a job matching model with imperfect information, in which skills mismatches influence the job application decisions of the workers, while skills gaps and shortages shape the competition for workers on the resulting bipartite job applications network. The tools proposed in this chapter could in future work be employed as the main ingredients of an agent-based model used to investigate how skills gaps, mismatches and shortages affect equilibrium outcomes.
The second chapter designs and tests machine learning algorithms to classify 33 million UK online vacancy postings into STEM and non-STEM jobs based on the keywords collected from the vacancy descriptions and job titles. The goal is to investigate whether jobs in “non-STEM” occupations (e.g. Graphic Designers, Economists) also require and value STEM knowledge and skills (e.g. “Microsoft C#”, “Systems Engineering”), thereby contributing to the debate on whether or not the “STEM pipeline leakage” – the fact that less than half of STEM graduates in the UK work in STEM occupations - should be considered as highly problematic.
Chapter 3 relates to empirical growth. It proposes a programming algorithm, called “iterative Fit and Filter” (iFF), that extracts trend growth as a sequence of medium/long term average growth rates, and applies it on a sample of over 150 countries. The paper then develops an econometric framework that relates the conditional probabilities of up and down-shifts in trend growth next year to the country's current characteristics, e.g. the growth environment, level of development, demographics, institutions, etc.
Finally, Chapter 4 studies credit risk spillovers in financial networks by modelling default as a multi-stage disease with each credit-rating corresponding to a new infection phase. The paper derives analytical and proposes computer simulation-based indicators of systemic importance and vulnerability, then applies them in the context of the Eurozone sovereign debt crisis.
Actions (login required)
 |
Record administration - authorised staff only |
Download statistics
Download statistics