1
$\begingroup$

I want to do a regression. However, I have quarterly data for the dependent variable and only annual data for the independent variable. Is it useful to interpolate the annual data on quarterly data or are the results then not to be used?

$\endgroup$
1
$\begingroup$

You're better off transforming the quarterly response data into yearly response data (probably either sum or mean, depending on your goal) than creating predictor data out of thin air with interpolating the annual into quarterly. Interpolation is invalid for time series data; the results would be useless.

$\endgroup$
  • $\begingroup$ +1. I agree on the main point. The last sentence seems a little strong: the key point is, I suggest, that interpolation can't add information and that any supposed increase in sample size or degrees of freedom from extra data points is utterly spurious. However, there are many reasons for interpolation of time series, not least being specifically interested in values at unknown times. $\endgroup$ – Nick Cox Feb 7 '17 at 21:54
  • $\begingroup$ A word of caution: "interpolation" would here assume that the annual data is snapshots. If the annual data is yearly averages, then the appropriate term would be "downscaling". (Here one solution is to interpolate the "cumulative series", like here ... but this is far from unique!) $\endgroup$ – GeoMatt22 Feb 7 '17 at 22:13
  • $\begingroup$ Thanks for the answer. The problem is, I have done an event study. The events were responses to quarterly results. Some events had to be deleted from the sample. So I do not have every quarterly results of every companie. Averaging or summation would thus distort the variables on the other hand. I shall have no choice but to find another indicator. Or just work with a dummy variable. $\endgroup$ – Macjackson Feb 8 '17 at 9:27
  • $\begingroup$ @ GeoMatt : I would say that the annual data are snapshots. They come from a yearly revolving survey of companies regarding their research expenditures. $\endgroup$ – Macjackson Feb 8 '17 at 9:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.