Matlab is very niche. I never used it, and most of my colleagues don't use it either.
Stata is easy. You can pick up the basics in a couple hours, and in most cases that it all you will ever need to do most economic reasearch. However, in some cases stata makes simple tasks complicated and complicated tasks very laborious.
R is harder to learn than stata, but it has an impressive statistical library. Unfortunately, that library also has many bugs and quality control issues are prevalent. R is also slow and difficult to read. I quit using R because I could not read my old code for past empirical projects.
Python is harder to learn than stata, but it is still not that hard to learn. It's documentation and online resources are excellent, unlike stata. It's statistical library is good and growing. I like python because it is very easy to read. You can go back to old projects and pick up from where you left off quickly. It is also the best at scraping data, which is important right now. A lot of novel data sets are starting to come from scraping websites or other sources.
I will note that R is easier when it comes to running any analysis, but python is way easier when it comes to data cleaning. Most projects are 90% data cleaning, which is why I really like python.