Econometrics and statistical software | INOMICA

Economists very often work with statistical software which is used to build economic models and conduct econometric analyzes. Learning to work and analyze data is therefore an essential skill for young economists. To be competitive as an economist in the job market, you need demonstrable skills and experience in using some of the most popular analytics and forecasting software environments.

Broadly speaking, there are two types of software: proprietary and open source. Some people and organizations choose to use proprietary packages developed and copyrighted by a single company. Others rely on free open source solutions, such as the popular “R” project. Within these two categories, there are dozens if not hundreds of alternatives, which vary in price, complexity, capacity, ease of use and popularity.

This post from Bob Muenchen is a great resource if you’re looking for a survey on the most popular packages. It uses a variety of different measures to evaluate different software options and keep them up to date. Keep in mind that this survey is for data science in general, not just economics. That said, all common software used in economics are represented, such as Python, R, Stata, SAS, SPSS, MATLAB, and others.

This is quite the list of names and it can seem intimate. To help you understand the software tools landscape, in this article we will shed some light on the most popular software packages for economists and offer some details on how you can learn more about them.

1) Which software should i use?

Before purchasing a package or downloading an open source option, it is a good idea to talk to colleagues, visit forums or online communities related to each product of interest, review course curricula if you are a student, and take free trials.

It is also important to consider your skills and / or the level of experience available within your team (if applicable). If there are several software packages available that meet your needs, options with better support could save you time and stress in the long run.

That said, business students may tend to focus on the main open source and non-proprietary tools in their studies; many courses use open source tools to teach. Python and R are two examples. R is used very often by economists. Python, meanwhile, is used somewhat less often in economics, but is used frequently in data science, which increasingly overlaps with econometric analyzes that economists may want to perform. There are many statistics packages for Python that allow it to conduct the same analyzes that other popular packages like R can do.

Hence, familiarity with both packages pays dividends. There is a large amount of free educational material on both R and Python available online; knowing how to perform regressions and manipulate data in these programs is a big plus for a serious business student.

However, there are other benefits to proprietary software platforms with paid licenses such as Stata, SPSS, and MATLAB. Often, the support and training opportunities behind these platforms are likely to be wider than with open source software. Furthermore, these tools may be easier to use for economists with little programming experience, with well-designed user interfaces that allow economists to manipulate graphs and variables where open source tools might require lines of code to do so.

2) What kind of license should I buy?

If you are a college student or company employee using a proprietary software tool, you probably already have access to the software without having to purchase it for personal use.

Otherwise, if you wish to use these options, you will have to pay a license to do so. The criteria for different types of licenses vary from one software manufacturer to another. Typically, student licenses will be the cheapest and are offered for shorter periods (weeks or months, rather than one or more years).

Many software packages also have an academic and / or non-profit license available. Business licenses are normally the most expensive, but you probably won’t need to purchase a business license yourself. Check the criteria for each package and consult your local distributor to see which license type and corresponding level of training and support is right for you.

3) Do I have to buy directly or from a distributor?

If you’re going the buying path, it’s almost always possible to buy directly from the software manufacturer via their website. But, consider looking for a local distributor instead. Pricing is usually more or less the same (with exceptions), but distributors are advantageous for the “extras” you receive as a customer, such as internal support, newsletters, and local user group meetings.

In addition, local support and training for you and in your language can be an important benefit. Another advantage is that if you are purchasing several software packages, you can have one point of contact for all of them.

4) How much programming or programming mastery do I need?

Regardless of the software package you choose, you may be wondering how much programming experience you will need to be successful as an economist. Do you need a deep understanding of data science, programming or coding, as well as economic theory to be successful?

Most likely, the answer for you will be “not really”. Most economists won’t need to be expert programmers too. Expertise in defining one’s own functions, manipulating data, and understanding the algorithms underlying common functions will make the life of economic analysis easier. More in-depth skills are probably not needed, since software tools are built to help you get your job done, but of course it will give you an edge.

For example, most software packages contain short, easy-to-execute built-in commands that allow you to immerse yourself directly in statistical analysis. For example, to perform a linear regression in R, simply type a command and tell the program which data to use. After typing “summary” a linear regression analysis will be released complete with t-statistics and p-values, the R2 value, F-statistics and many more, all calculated in a split second. Many other more complicated regressions can often be performed just as easily.

Running these simple commands doesn’t require advanced coding experience, but basic skills with your chosen software will often come in handy. For example, suppose you receive an error message. Familiarity with your favorite software and the algorithm used to output your model will help you clear the error quickly so you can get back to the interesting stuff.

Also, if you have some coding skills, you will often be able to write your own functions and figure out what existing functions are doing. This gives you more flexibility when you need to implement a very specific probability distribution, re-parameterize something, adjust a graph to highlight your results, etc.

Overall, the key to being successful in such economic analysis lies in understanding the model’s output and interpreting it in the context of economic theory, which does not require advanced coding experience. However, if something goes wrong or you need a particularly complicated setup to conduct your analysis, some coding skills are an advantage. Studying some of the software packages mentioned in this article will help you get started on your economic analysis journey. Good luck!

Photo credit: Gilad Lotan