Mapping My Academic Journey
Almost a decade ago, I began working on the paper that would later become my job market paper. Up to that point, I had little idea of what I was doing. I was in the third year of my Ph.D., having just completed most of my coursework as the research year officially began. I was visiting CEMFI, where Ph.D. students presented their research twice a year.
Early in the academic year, I presented my master’s thesis, now thirteen years old and soon to get published. It was my first-ever presentation in the Ph.D. program. I worked until the very last minute, making changes to my slides based on feedback from my coauthor just before the presentation. One may think I did a great job given the effort. I didn’t. Quite the opposite. I was tired and confused and managed to confuse almost everyone attending.
With the Spring presentation approaching, I was in a pressure cooker kind of situation. The date was set, yet I had no idea about the field I would work on.
It was like I had invited people for dinner at home, but without a plan or a recipe of what I would cook, and I even lacked some essential ingredients to prepare a meal. Let alone a nice meal.
Coming up with a research idea
For several months, I immersed myself in extensive literature, across fields, desperately searching for a suitable research field/topic. Initially, upon arriving in Madrid, my plan was to focus on Econometric theory, or so I thought. Many of my peers in the master’s program in Rome (the second one) also gravitated toward Econometric theory, which seemed inevitable given the training.
However, during those months, my mind kept circling back to the recent courses I had taken in Stockholm before leaving for Madrid. These courses covered topics in Development and Political Economics. They were the only subjects I could think of at the time.
These fields connected with my personal experiences, particularly what I had witnessed growing up in Albania and my father’s involvement in politics. The past became the driving force behind the present (research).
The birth of the project
The idea came roughly a month before the presentation. Upon revisiting a seminal paper in the field by Ferraz and Finan (2008), I started to think of corruption spillovers. Do they spread? How far? What are the consequences?
I talked about the idea with a colleague and friend, Felipe. He said it was a long shot but worth the try. So, we continued to talk. Initially, it was supposed to be a joint project. Things turned out differently.
When I finally presented (the recipe) to my dinner guests, they thought it had good taste. Researchers at CEMFI were hungry for knowledge and understanding. They thought the idea was promising. It was an intellectual feast for them, sparking further questions and suggestions.
Eventually, the project evolved, but the core idea remained intact.
Skills needed
It was the right project at the right time. It really taught me a lot and sharpened my data skills. I learned GIS and ArcGIS. Because of the project, I even said no to the goldmine of administrative Swedish data. (A whole system was in place, and I would have had easy access.) I thought this one would give me the flexibility to learn skills I always craved to have.
The part of the project for which I needed GIS skills suddenly became too extensive. I had thousands of FM antennas spread across the Brazilian territory, and I needed to determine their locations and the municipalities they covered. I couldn’t manage this manually; I had to acquire programming skills to write the necessary code and automate the process. So, I did.
I taught myself Python through resources like Codecademy and other online material. Additionally, my ex-partner, a telecommunication engineer turned programmer, played a crucial role. He provided me with small tests to help me practice and even offered guidance on coding concepts when I needed it.
More skills
After returning to Stockholm, I learned more about machine learning and text analysis. My advisor, David, offered a new course on these subjects precisely the year I was back. (David himself is a pioneer in the field of Political Economy of Media and the application of text data in Economics.)
That course was centered around Python as the primary language and also covered aspects of version control and project management (including Jira). I’m grateful that I had to take this course for credit because it provided the perfect opportunity to apply what we had learned and approach the methodology taught in class from a different perspective.
Over time, the project taught me (many) other skills. It taught me about academic publications and the tastes of the referees and editors. It taught me how to write a scientific article and how to communicate it to a scientific audience.
During the first major rewrite and restructuring of the project, I focused on improving my writing skills, both in academic and non-academic contexts. For the first time, I read with the perspective of a writer, learning from interviews and podcasts with journalists (which in turn reignited my long-lost passion for writing and journalism).
Additionally, the project taught me how to teach, particularly research-related courses. My teaching methodology is a blend of what I observed as a student and what I found essential while engaged in solo writing and data analysis.
Continuous learning
I am grateful to the project: many other ideas were born because of it. I got a job because of it. Every time I return to it after putting it aside due to teaching, rejections, or involvement in other projects, it teaches me new skills including how to handle coding mistakes late in the game.
I did wonder if we had discovered the error earlier if Felipe would have been part of the project. Perhaps, we would have trusted each other and not discovered it at all.
[I’m still torn. On one hand, solo coding gives you the freedom to experiment and troubleshoot at your own pace, and you don’t have to explain every line of code to someone else. On the other hand, duo debugging can bring a fresh pair of eyes and a different perspective. What gives me peace, is that at least I caught the error before the article was in print.]
This project has really been like a constant companion, shaping me into the researcher I am today. However, I look forward to the day the paper finds its place in a (reputable) journal and also contributes to the ongoing discourse in the field.
Roughly ten years after the conception of the paper, there is new research looking at similar questions from a different angle. There are many more papers in economics creating measures from text data, including corruption measures. I am glad that directly or indirectly this project inspired that research too.
