Date of Award


Degree Name

Doctor of Philosophy



First Advisor

Dr. Michael Ryan

Second Advisor

Dr. Susan Pozo

Third Advisor

Dr. Kuanchin Chen


Transportation network companies, gasoline, digital economics, social media, Twitter, Uber


The Digital Age is defined by advances in big data, low-cost computing, and modern analytical methods. In three chapters, I apply microeconomic thought to the influence these digital technologies have on business and consumer decision-making.

Chapter One considers how Transportation Network Companies (TNCs) like Uber and Lyft impact the relationship between the quantity of car service trips and fuel prices. I find that a single day 1% increase in fuel prices is associated with a 0.367 to 0.486% decrease in the number of TNC trips and a 0.033 to 0.088% increase in the number of taxi trips, even though the drivers under both systems pay for their own gasoline. I attribute this difference to the flexible nature of TNCs and the fixed behavior of taxicabs. TNCs allow drivers to reduce their supply when operating costs are higher while taxi drivers are often restricted within a regularly scheduled shift due to higher city regulations during the period of study. Uber drivers who are unwilling to pay higher gasoline costs will leave the market, thus reducing competition for the taxi drivers who are stuck paying for the higher gasoline costs but benefit from more trips, in what I call a “rigidity dividend.”

Chapter Two considers how social media influences the information sharing and decision-making processes for customers deciding to visit monopolistic competitive brick and mortar stores. I use hierarchical linear regression to account for the random effects of brand- and store-heterogeneity and demonstrate this method is an improvement to ordinary linear regression. I find that online behavior influences offline store visits, especially for changes in the popularity, sentiment, and disagreement around a brand on social media. For example, when social media mentions of a brand increase one standard deviation either in per-like popularity, sentiment, or disagreement, then next-day foot traffic to stores of that monopolistic competitive brand will increase by 0.02 standards deviations. The manner social media activity precedes retail foot traffic is consistent with a causal pathway. This modest yet meaningful effect, however, fully dissipates within one week.

Chapter Three considers how to incorporate a large dataset of demographic and economic geographic variables into business decision-making using mobile phone data and machine learning methods. The case study demonstrates how including zip code-level tax data reduces prediction error in models of customer visits to retail stores. I consider three approaches that also reduce the effects of multicollinearity: linear regression with variable reduction though a model-dependent variance inflation factor threshold, ridge and lasso regression, and random forest regression. I find store-specific variables to be the most predictive with additional, valuable insights from economic and demographic variables relating to the population around a store. The results demonstrate the importance of economic geography in the store location decision for multiunit retail companies.

Access Setting

Dissertation-Open Access