Getting Stata Files for Pollock's Econometrics Books
You usually find these when you're working through David Pollock's "Handbook of Econometric Modelling" or his time series books and the examples aren't clicking. The Stata code he published alongside his texts is scattered across a few university repositories and personal academic pages. Most of it lives on economics department sites at Reading and other UK universities. I've spent years tracking down these files because they're legitimately useful, not just because Pollock writes very clear econometric theory. The Stata commands and datasets attached to Pollock's books are the main thing people search for. His 2005 "Handbook of Econometric Modelling" has an entire companion site with .dta files and .do scripts. The 1999 "Time Series Analysis" book similarly has code available. You typically won't find a single clean download page anymore. The original hosts have been taken down or reorganized over the years. That's why you end up hitting academic archive pages and departmental file directories. Here is how I actually work with these files when they surface. First, you grab the dataset and the corresponding do-file. Open Stata, load the .dta file directly into memory with the use command. Then run the do-file line by line. I don't run the whole thing at once because the path references are often pointing to local directories that don't exist on my machine. I go through it methodically. If you hit an error on line 47 and just keep pushing forward, you waste 20 minutes chasing ghosts.
One edge case I ran into recently: the dataset for Chapter 6 in the 2005 handbook uses a variable naming convention that conflicts with a newer Stata version keyword. Stata 17 started flagging certain words differently. I had to rename the conflicting variable using rename old_name new_name and then adjust the rest of the do-file accordingly. It took me maybe ten minutes but I couldn't find anyone else posting about this exact clash online. I ended up just keeping a note of it for future reference. The counter-intuitive thing about Pollock's Stata code is that it is often more rigorous than you need for a quick exercise. He includes things like proper matrix inversion checks, robust standard error calculations, and diagnostic tests that most students skip. That is why it is genuinely worth reading through rather than just running blindly. Beginners tend to treat it like a recipe and get confused when the output doesn't match their own data exactly. The code is written for Pollock's specific datasets, not arbitrary ones. Another nuance people miss: some of the older Stata code uses syntax from version 8 or earlier. Commands like set seed values, certain graph options, and even the format for displaying regression output have changed slightly. If you are running this on Stata 17 or 18, you might need to adjust a few lines. The results themselves will not change substantially. Just the way the code is presented might flag warnings. I usually turn those warnings off during initial runs and then go back to clean them up.
I should note what doesn't work well. The companion materials are incomplete for some chapters. The 2005 handbook has full Stata support for the core chapters but a few later sections only have SAS code. If you are looking specifically for Stata answers to every example, that gap will frustrate you. There is also no centralized index of which do-files belong to which chapter numbers. You sort of have to reverse-engineer it by looking at the variable names and date ranges in each dataset. For people who just want the raw code without working through the theory first, there is an alternative. I usually cross-reference Pollock's Stata files with similar code from Kennedy's "Guide to Econometrics" or Gujarati's materials. Those sources sometimes have more complete Stata translations of the same models. It saves time when Pollock's original files are missing or broken. If you do find a working Pollock Stata dataset, here is a quick workflow I recommend. Load the data. Run the do-file in batches of five to ten lines. Compare the output against the textbook tables before moving forward. If your numbers don't match Pollock's published results, check your variable labels and frequency settings. Minor differences in rounding between Stata versions can make regression coefficients look slightly off even when the model is correct.
Get the Full Details
Most of the active links for these files now sit on cached academic pages and GitHub mirrors. Searching for Pollock Stata handbook companion files will usually surface something. I check the Wayback Machine occasionally when an old department URL goes down. It has saved me more than once.
