Are there any numbers on publication rates for different areas of mathematics? For example, I would expect that the average in graph theory is higher than the average in analysis.

$\begingroup$ If one could interrogate MathSciNet electronically then one could look at primary MSC numbers on papers, but I don't think you can do this. Also, one might wish to filter the results to "core" mathematics journals, whatever this means, but there have been concerns the top journals are biased towards certain topics, so YMMV. $\endgroup$ – David Roberts Feb 19 at 21:18

1$\begingroup$ I think you will find this answered here: mathoverflow.net/a/233994/11260 $\endgroup$ – Carlo Beenakker Feb 19 at 21:24

1$\begingroup$ Carlo, that's a start. I would like to see information about average rates. For example, is it more common for a graph theorist than an analyst to publish more than ten or more papers in 5 years? what about 15 or more papers? $\endgroup$ – epsilon Feb 19 at 21:43

3$\begingroup$ You equally count a 100page paper singleauthored and a 3page paper with 5 authors? $\endgroup$ – YCor Feb 19 at 22:37
I'm hesitating to give this answer, but Italy has researched the question "on average how many papers per year does a professor in a given field publish", and they have made this into a selection criterion for promotions. It's a crazy system, but here is their table, for what it's worth. It will tell you that in mathematical analysis or statistics the average output of a full professor (I fascia) is 10 papers in 10 years, while in mathematical logic it is 5 papers in 10 years.
_{ These are among the lowest rates in the table, if you are in the medical profession the average output can be above 10 publications per year. In my own field, condensed matter physics, I might qualify as an Italian professor with less than 3 papers per year.}
This "bean counting" procedure was made into a national law, if you can read Italian it will bring tears to your eyes.

1$\begingroup$ Average is the mean or median? I would guess the mean, given the statistical ignorance this exercise shows. $\endgroup$ – David Roberts Feb 19 at 22:44

1

4$\begingroup$ @DavidRoberts These used to be called 'medians' (along the population of all associate/full professors) in a previous version of the procedure, and that's how they were computed (possibly wrongly, since it's difficult to gather data on the whole population). In the current version they are simply defined as 'threshold values' that one needs to achieve for habilitation to that category. I guess part of the reason for this change is that you can sue them for computing a median incorrectly, but you cannot sue them for defining an arbitrary threshold value incorrectly... $\endgroup$ – Federico Poloni Feb 20 at 0:47

1$\begingroup$ @JW Ooh thanks I misread that, it makes sense this way indeed. Anyway, I have deleted that comment and made an edit to the answer in the hope to make this clearer. $\endgroup$ – Federico Poloni Feb 20 at 6:22

1$\begingroup$ @DanFox  true, also notice that professors of computer science are expected to accumulate nearly 100 times as many citations to their papers as those working in mathematical logic! The sad thing is that this crazy system is taken very seriously. At one point I received a request if I could please cite one particular paper, even if totally out of context, just to tip the scale for a promotion in Italy. (The Hirsch index works that way, one fine tuned citation out of several hundreds can make the difference between promotion or no promotion.) $\endgroup$ – Carlo Beenakker Feb 20 at 9:04