Monthly standard deviation in time series dataset

[Neha Deopa]

Hello

I am trying to find the standard deviation (i.e. volatility) of the monthly stock market returns.
Below is my data where date2 is basically date in %tm format - I dont know why it's not displaying it here properly but 336 = 1988m1, 337=1988m2, 338 = 1988m3 ... etc.
lsp is the log of the s&p 500 index
lret is the returns calculated from lsp by:

 gen lret = D.lsp

I have declared the data to be time series by

tsset date2

Whenever I enter the code

egen monsd= sd(lret), by(date2)

I get basically no values and the entire comes up as "348 missing values generated". What am I doing wrong here?

I was replicating someone else's code who converted the daily returns to monthly and then calculated the volatility by the same codes and this problem does not seem to be coming up there.

Thank you!

* Example generated by -dataex-. To install: ssc install dataex
clear
input double date2 float(lsp lret)
336 5.508376           .
337 5.585299   .07692337
338 5.602045  .016746044
339 5.570784 -.031260967
340 5.539615  -.03116894
341 5.633396   .09378052
342  5.59326  -.04013538
343 5.573294 -.019966125
344 5.624053   .05075932
345 5.676309   .05225515
346 5.627945  -.04836369
347 5.663551  .035605907
348 5.688466   .02491474
349 5.719295   .03082943
350 5.710824 -.008470535
351 5.781792  .070967674
352  5.81434  .032547474
353 5.846554   .03221464
354 5.872231   .02567625
355  5.90206  .029829025
end
format %tm date2

[William Lisowski]

Your command

egen monsd= sd(lret), by(date2)

instructs Stata to calculate the standard deviation of lret for each group of observations with the same value of date2. But your data have only 1 observation for each distinct value of date2, and there is no such thing as the standard deviation of a single number.

I suspect that the work you are replicating first calucated the volatility by month using the daily observations and second reduced the daily observations to a single monthly observation.