Don’t Let Them Fool You: Why Economists Makes Great Data Scientists

….and why it is recommended too!

The steps I took into making my Master’s Thesis in Economic Analysis can be summed up as: 1. Identifying an economic problem, 2. Identifying the data needed to solve the problem, 3. Gathering the data and cleaning it, 4. Exploration of the data, 5. Modelling the data (statistical modelling, predictive modelling, you name it), 6. Communicating the findings, tying it back to how it solves the economic problem asked in step 1.

Extremely similar, a data scientist needs to have a good business understanding, then the data understanding, gathering the data, EDA (Exploratory Data Analysis), Modelling (Machine Learning) and communication of the results. Similar? I say very.

STEM and Economics

The title of the article asks you to not let “them” fool you. Who “them” is, is irrelevant. Just don’t let anyone fool you. Economics is not a “dry, theoretical and qualitative” field, not anymore and probably never was. It has all reasons to be part of STEM, in that it not only covers the ‘S’ and ‘M’, but the inclusion of the ‘T’ in Economics academic degrees is growing steadfastly.

The Science aspect of STEM in my opinion, should not limit itself to physical, biological, life and natural sciences — but should include social sciences as well. I’ll tell you why. Mathematics is a key component in STEM, and is so in Data Science, and definitely so in Economics. Some of the concepts of Mathematics crucial to an Economist are differential calculus, geometry, matrix & linear algebra. But, if I haven’t proven my point properly yet, read further for you to see me vindicated.

Data is now everywhere. Without data, there is no solution that can be accepted because data is unquestionably right, and you can eliminate the entire debate of “how can I trust you?”. Data is like God, or like your PT Teacher (drawing from experience — basically something/someone you’re afraid of). Using data to solve problems is the norm now, and economists did this since the 19th century. Its what we were introduced to as “empirical studies” — using historical data to prove a point, or some pulled-a-rabbit-out-of-the-hat kind of theory.

Ideas are just a block of rock. It is with data, that you can sculpt Christ the Redeemer, put it on top of a mountain and make it into a functioning business model. Data Scientists are naturally expected of this, and well before, so were Economists. To end this section, if a scientist can be defined as someone who “observes, measures and communicates”, you gotta include economists in the lot. Anthropology, sociology, economics, behavioral sciences and psychology are often grouped, and yes, all of them are STEM too.

I’m Not A Stranger, Just A Lost Cousin

Now I can tell you why not to let anyone fool you. I graduated Master’s in Economics, and I am a Data Scientist now. Have I made a serious shift in my specialty? Have I committed a blunder? Is it still a blunder even if I’m fairly good at Economics as well as Data Science, but just because they’re not “the same”? None of those. Point is, I have not switched sides, not by any means.

An excellent argument to make against mine, would be this: most Economics degrees don’t teach programming, databases, or even machine learning — deployment of a model, training, fitting, etc. But the thing is, what Economics does is, it makes the mind capable enough to branch out and grow easily, and sit comfortably in anything closely related to data — just like that leopard.

What Data Scientists have a high propensity to lack, and what Economists are expected to have, are perfect complements — and that is critical thinking and communication. Economists are very verbose, and are generally good with words. Trust me, even if you aren’t when you start, the words will be squeezed out of you — what with academic papers, and dissertations, etc.

The key ingredients for a reasonably good Data Scientist are:

• Mathematical and Statistical Knowledge
• Programming and Technical Knowledge
• Domain Expertise and Communication

The kind of mathematical and statistical knowledge that is applied in Data Science is substantially “easy” to an Economics graduate. Picking two people randomly in the world, one being a Data Scientist, and one being an Economist, they would approach a Machine Learning (ML) problem in these ways: the Data Scientist would be efficient in answering the question, in terms of handling the data, as well as making an accurate overall predictive model; — whereas an economist, will spend most of his time framing the perfect question that is asked in the first place. Economists are taught the difference between “correlation” and “causation” much earlier.

My Master’s degree trained me to a certain level in programming with R. And, anybody that says Python programming is not easy, really need to freshen up, get a mug of coffee, and see for how easy it is. Python and R are the two leading programming languages for Data Science today. The world is moving towards no-code anyway. I “became” a Data Scientist in less than a year. All I used to train myself was resources online, and courses online, etc. and sheer will to keep going and practice. Don’t worry, I am not here to brag, I have a major Imposter Syndrome when it comes to Data Science, even if my mom asks me to believe in myself more.

There’s a unidirectional effect when it comes to this connection though. An economist interested in data and Econometrics (using data for economics) would naturally be interested in using tools such as programming languages, Excel, STATA, etc. or whatever keeps their data flowing. This enables the economist to get hands-on quickly with Data Science tools, aka Python, Tableau, Azure ML, etc. and therefore, the Technical Knowledge I mentioned above. To all readers who know basic ML, an Economist has much deeper knowledge of Linear Regression and Logistic Regression, than the average Data Scientist. So this begs the question:- is ML just a fancy term designed to sound technical, and keep away people who aren’t into “Tech” or Computer Science?

While I agree that ML Engineering, including deployment and maintenance of a model post production can typically be handled more efficiently by a computer science graduate, economists are excellent candidates for understanding the foundations of the model and problem.

Through this article, I’m not here to take sides — I’m here to promote the publicity of the potentially happy marriage between two seemingly unrelated fields.

Economists typically are people who cannot settle for less. Every time I sat in my Econometrics classroom, I could not help but come up with another possible issue with the model, and no matter how many solutions I keep coming up with, there is always another problem. It feels like a crime to settle for some level of error. This pushes the expectations higher, and makes a well-refined model. This is achieved either through tweaking the research question, or through a thousand sleepless nights. This is where the expertise in handling a problem comes in handy in Data Science.

I hope I was somewhat convincing, and hey, you never know, I might get inspired later and come post a second part to this article too, with more ideas! The main point is, don’t ostracize social scientists in the Data Science community, as they are not only qualified to be one, but are probably equally good or better strategists as well. They make excellent Decision Scientists, and that is what Data Science is leading to in the future. We all saw that just because Michael Corleone didn’t want to be part of the “family business”, didn’t mean he was not good at it. Ciao!