I am conducting econometric time series analysis using data from FRED (Federal Reserve of St. Louis), BEA (U.S. Bureau of Economic Analysis) and the U.S. Department of the Census. Ideally, my study using multiple linear regression will study data on a quarterly basis; however, much of the data is provided only on an annual basis. I am looking for an approach to interpolate the annual data to a quarterly basis without creating undue concerns over the validity of my study.
One approach is to use averages, 3-periods, 4-periods average. Such technique fits inconsistent data sets.
Use some relevant indicators for which quarterly data is available to transform annual data into quarterly. You can see my work on Quarterly GDP of Pakistan in which we have used various indicators in said manner for quarterisation of NIAs.
If you're good using eviews, it's simple! Open eviews, open a workfile with the annual data. Then click on New Page, choose specify by frequency/range to open a new workfile but this time, choose quarterly when selecting the data type and click ok. Copy each annual data variable from the workfile you created with the annual data and paste special into the new workfile which you created for quarterly data. In the new environment for paste special dialogue, choose series by value and choose linear in the specified in source. Click ok and the annual data will be converted to quarterly data. I have attached some pictures to guide you. I hope this helps!
Outstanding responses supported by relevant studies. You folks are incredible with your helpful and valuable feedback! Thank you very much.
I am rather sceptical what concerns the use of (purely) technically "quarterized" data in regressions. Especially if the dependent variable is annual, one should never transform it to quarterly data. But also for annual explaining variables one should be very cautious. I rather recommend to look at the correlations between the variables (on an annual basis and - where possible- for quarterly data, too) and to start with a regression on annual series. Quarterly data allow to include moving averages with lags to get information about the lags shorter than a year. The construction of quarterly data is only senseful, if you have relevant indicators (see Muhammad's answer) or if you are rather sure that a variable has no (clear) seasonal pattern or you have some information on this pattern (for example, if there are quarterly data for some of the years).
An excellent explanation is given by Mr. Chekwube Madichie for converting annual data into quarterly data. It's better to use the same frequency for all the variables.
Best Wishes,