Examination of Institutional Investment in the United States of America from 1999 to 2018

This thesis examines the evolution of spatial preference of institutional investors located in the United States of America for the time period of 1999 to 2018 using a mix of exploratory data analysis techniques and more sophisticated space-time and machine learning techniques such as ESRI Space-Time cube and Latent Dirichlet Allocation topic modeling. This thesis concludes that despite having the appearance of a footloose industry due to almost negligible fixed costs, institutional investors are attracted to highly dynamic urban centres and on the 20 year time horizon, appear surprisingly sticky in their location preference. This is consistent with the belief from previous literature that institutional investors choose their urban locations due to the ubiquity of high quality information and access to advanced professional services commensurate to the investor's role as a command and control node in the broader economy. A close examination of the data concludes that while New York State may be in relative decline nationally, it is still the fastest growing location in absolute terms and remains firmly at the top of the US investor urban hierarchy. Lastly, the machine learning classification of investment strategies shows that California-based investors have unexpectedly poor performance across a broad range of identified strategies, however this can be explained by California's venture capital culture, hinting that local knowledge and culture can influence investment strategy decisions.