Converting annual/quarterly values to daily values?

[Junaid Haider]

I am doing an event study. My dependent variable has daily values while independent variables have annual (i may find quarterly, if required) How can i convert annual/quarterly values to daily values in order to run any test?

[Clyde Schechter]

There is no completely satisfactory solution to this problem. Depending on what your variables actually are and how they normally change over time, one possibility is to simply assign the annual value to the observation for each day of that year, or to the dates from July 1 of the preceding year through June 30 of the current year. Another possibility is to interpolate. (-help ipolate-) Both of these approaches suffer from severe drawbacks: first we know that the data values being used are, except for rare coincidence, wrong. Second, the variance of those variables is severely underestimated which, in turn, biases the coefficient estimates you get when you use these data in any regression. One might overcome the second limitation by using multiple imputation instead, but with only a tiny fraction of the data non-missing, doing so is of dubious feasibility and validity.

Perhaps a better solution is to go the other way. Convert your daily outcome variable to annual (resp. quarterly) by aggregating in some way. Again, the best way would depend on the meaning of the variables and how they vary over time. You might want to use the mean value for each year, or the median, or perhaps the largest, or smallest. Or a weighted mean of some sort. Or some other rule might be best. The drawback to this approach is that it evidently discards information.

The only really satisfactory solution is to obtain daily data on your predictor variables--but I presume if that were feasible you wouldn't be asking this question in the first place.

[Junaid Haider]

Thanks for very useful insights. Yes, you are very right. First solution is not appropriate in my case while second has it's own limitations. Seems like i need to reconsider my approach