About Me

I was born and raised in Fairbanks, Alaska. I cross-country skied competitively in High School, before attending the University of Vermont for my undergraduate degree. At UVM I studied Economics and Computer Science, and successfully defended my honors college thesis on the effects of peer-to-peer lenders on geographic redlining. At UVM I also developed my affinity for programming, modeling, and econometrics. After graduating from UVM, I moved back to Fairbanks and began working full time for Spirit of Alaska Federal Credit Union as a Data and Business Analyst. In this role I dealt with data in all of its forms. I migrated our monthly financial reports from Excel to Tableau, and designed a variety of interactive dashboards on everything from our loan portfolios, to delinquency and charge off rates, to member demographics and member engagement. Additionally, I was project manager for the implementation of a fully online account opening solution for the credit union, and co-project manager for the implementation of a new fleet of Interactive Teller Machines (ITMs). Finally, I assisted our CEO and CFO with strategic financial and business planning, including setting deposit and loan rates, managing our investment portfolio, improving the member experience, and defining our “digital transformation.” This job was fulfilling for me because it combines my passions for data, technology, and financial inclusion. I am now pursuing my MBA at the University of Rochester, Simon Business School.

A Not-So-Random Walk Through Economics

My Cover Letter for Departmental Honors, 2021

My exploration into economics began in 2016 when I took AP microeconomics and macroeconomics during my senior year of high school in Fairbanks, Alaska. This decision was motivated by my interests in politics, finance, and math; however, I did not yet have even a passing understanding of economics as an academic field. My AP teacher was very interested in behavioral economics, and focused on group activities where we would determine the supply and demand of ‘class points’ as well as the ‘price’ for those points, among other things. These activities were my first taste of the concept of equilibrium, and peaked my interest in economics as a sort of all-encompassing puzzle. Little did I know how much more I had to learn.

I applied to UVM as an economics major and haven’t looked back since. Thankfully, during my participation in the Integrated Social Science Program (ISSP) I was able to branch out and study a wide range of social sciences, which made me confident that I had made the right choice in studying economics. Additionally, ISSP widened my understanding of the scope of economics. Capitalism and Human Welfare (EC060) showed me that economists can study welfare with more than just a passing discussion of utility. Gender Inequality (POLS095) gave me my first taste of econometrics (even though it was not an economics course), and Social Inequality (SOC032) introduced me to the very interesting question of opportunity cost in relation to the Alaska Native Claims Settlement Act. As a senior, I look back fondly on these courses as an introduction to the massive breadth of questions that economics can help answer.

Throughout my time at UVM I attempted to branch out and study a diverse range of economic topics, from Law and Economics (EC135), to International Finance (EC146), to Economy as a Complex System (EC237). Throughout this exploration I have found what I believe to be my main interests inside of economics: banking, agent-based modeling, financial technology (FinTech), and market efficiency. These topics represent all of my current work in economics including the attached research paper and my Honors College Thesis.

Money and Banking (EC120) was one of the economics courses that potentially changed my life. The course peaked my interest in banking, as well as my interest in the inequities that still persist in credit provision. Before this class I was scratching my head looking for internships, and by the end of the class I had mustered the courage to approach the CEO of an Alaskan credit union (of which I had been a member for life), and secured an internship that has now blossomed into salaried post-college employment. For this internship I used my knowledge of econometrics, data analysis, lending mechanisms, interest rates, and deposit requirements to design custom reports that get sent to the CEO, CFO and the Board of Directors every month. Additionally, the COVID-19 pandemic has offered me the opportunity to more fully understand how massive unemployment, the Paycheck Protection Program, stimulus checks, and business shutdowns have affected the (now VERY liquid) banking industry.

Professor Knodell also helped me with my pre-thesis Honors College research on FinTech, which has now blossomed into my thesis about credit discrimination on peer-to-peer lending platforms. My research looks at every loan application ever submitted on the Lending Club platform (~ 30,000,000) in order to calculate the predicted probability of loan approval for different historically redlined groups. Using geographic data that is provided by Lending Club, I created binary indicators for living in a predominately African American, predominately Hispanic/Latino, or historically redlined zip code. These indicators allow for me to calculate predicted probabilities of approval for different groups while controlling for the classic measures of credit-worthiness. Additionally, my thesis calculates the predicted probabilities of charge-off for approved loans in order to understand if groups that are less likely to be approved are more likely to charge-off. This analysis is pivotal for understanding whether discrimination on peer-to-peer lending platforms is statistical or taste-based.

My research has found significant disparities between approval rates for applicants living in predominantly African American zip codes, and slightly less significant disparities for applicants in predominately Hispanic/Latino zip codes. Additionally, applicants from predominantly African American zip codes are more likely to have their loan charge-off, but not by a large enough amount to justify the approval discrepancy. This project showed me the complexity of questions surrounding credit discrimination, and discrimination more generally. While it is relatively easy to uncover that a disparity exists, understanding what is causing that disparity, and whether the cause is discriminatory, is a whole different beast. For example, although Lending Club releases troves of data on every loan application that is received, they do not give any indication of how they use that data to approve and deny applications. This means that even though the disparity exists when controlling for credit score, employment length, debt-to-income ratio, and amount requested, it is difficult to map those disparities back to the algorithm that created them. In spite of this difficulty, I am interested in continuing to work on the economics of discrimination. This is because I believe that discrimination is a non-zero-sum game. Discrimination dramatically hurts the efficiency of our social and economic systems, and I am motivated to use the tools of economics to help solve it.

My interests in market efficiency and agent-based modeling were born when I took Economy as a Complex System (EC237) this fall. However, I am now seriously considering graduate education in complex systems. I am a computer science minor, and the modeling side of economic research seems to be the perfect bridge between these two interests. Economy as a Complex System (EC237) allowed me to use the logic that I learned from computer science to tackle economic problems. My classmates and I did this by designing simulations that reveal how the micro-level properties of individual agents trickle up to macro-level properties that are observed in the system as a whole. Before this class I thought that the only way to answer an economic question was to collect real-world data and use econometrics. Now I realize that even if the real world data offers useful results, those results are often entangled with unobserved causalities that can lead to misleading conclusions. The agent-based modeling approach allows for the economist to design the system from the ground up, so that complicated systems can be reduced to a ‘lowest common denominator’ that is more easily studied. Economy as a Complex System (EC237) also introduced me to the concept of the “outside observer,” and how economic systems can often improve efficiency through a non-participatory system in which one outside entity controls or incentivizes behavior. While this is not an argument for economic dictatorships, I am interested in the bridge between the outside observer problem and classic game theory problems such as the Prisoner’s Dilemma.

An Agent-Based Model of Random Walk Pricing in the Equities Market is the research paper that I have submitted for your review. This paper leverages the benefits of agent-based modelling listed above in order to avoid the common assumptions required by the classical approach of equity market analysis. Moreover, using an agent-based model allows me to drop assumptions such as perfect rationality and homogeneity, and allows me to code different preferences, emotions, and beliefs into the traders in the model.

This paper was motivated by my desire to understand Random Walk Price Theory and the Efficient Market Hypothesis. I have been investing as a hobby throughout college, and have always wondered the degree to which all available information is already priced into a stock. While the model in this paper is based loosely on Hessary and Hadzoladic (2019), I expanded the model to include more types of traders and a more realistic price equation. The model represents the trading of one stock, and includes fundamentalist traders and two different types of momentum traders, who all rely on vastly different information to decide whether to buy or sell. Additionally, the model allows for different traders to ‘learn’ which strategies are the most profitable over time, and attempt to replicate those strategies. The variety of trader types, the presence of emotions and preferences, and the ability to learn are all benefits afforded to this research by the agent-based modeling approach. This research found that both Random Walk Price Theory and the Efficient Market Hypothesis are backed up by data from this simulation, which answers my original question by suggesting that all publicly available information is immediately reflected in equity prices (and that it is therefore impossible to make consistent short term profits in the stock market).

The biggest challenge of this research was identifying a way to repeat the Dickey-Fuller test for stationarity on several hundred runs of the simulation. For the first draft of this paper I used data from only one run of the simulation, which made the results less reliable because random parameters of the model change from run to run. I tried running a Dickey-Fuller test on the panel data for 100 runs of the simulation, but that test only gives one result (stationarity or non-stationarity) for the entire panel of runs, which does not reveal as much as the percentage of runs that are stationary vs. non-stationary. With assistance from Professor Gibson I used an Excel document full of STATA commands to autofill the code for 100 individual Dickey-Fuller tests. Once this process was automated, I was able to repeat it with constant and variable fundamental value, which allowed me to expand the scope of the paper.

I have chosen this paper for submission because it includes the design and specification of an agent-based model, as well as the econometrics and analysis that would normally be done on real-world data. Moreover, by using the Dickey-Fuller test and fixed-effects regressions, this paper exemplifies my use of econometrics, and specifically time series data. Finally, this paper expanded my knowledge of economics by introducing me to the dangers posed by the assumptions that often riddle economic theory. For example, the ‘perfect rationality’ assumption that is often taught in lower-level economics courses is also present in much of the literature that I read while developing this model. By acknowledging that ‘bounded rationality’ is a much more realistic assumption, my model captures the real-world scenarios where traders of the same type can have different opinions or misconceptions about the fundamental value and future prices of the stock. Disentangling economic theory from some of its more burdensome assumptions is an important goal of my future work.

As I wind up my time at UVM I can’t stop reflecting on all of the ways that I have grown as an economist. I have learned how to turn a question into a thoughtfully designed economic study by looking to the theory for guidance. I have learned how to generate data through simulation in situations where real-world data is not up to snuff. I have learned how to take a large dataset and quickly identify key characteristics and trends. I have learned how to quickly and effectively find relevant literature on almost any economic topic. And I have found economists who I look to for inspiration (Levitt, Dawkins, Becker, Epstein, Fama, Malkeil). In addition, I have come to understand that my role in economics is not just to study, but to contribute. Through my time at UVM, and the writing of the paper included here, as well as my thesis, I have come to view economics as an academic field that needs to be grown and expanded. A field that needs to break away from rusty theories and assumptions, and adapt its findings to the huge scope of problems that it can help solve. I hope that as I step into my career and graduate work, I make meaningful contributions to this field.