Good luck with the paper and thanks for your notes. You’re right that different methods of data interpolation will lead to different estimates of monthly figures from quarterly actuals. Here’s a good overview of two common interpolation methods — quadradic and cubic spline method — from the UK’s official statistical office:

http://www.statistics.gov.uk/STATBASE/Product.asp?vlnk=2184

http://www.census.gov/popest/national/NA-EST2007-01.html

I am attempting to estimate the short-run price elasticity of demand for gasoline and I’m going to follow your methodology with updated data (through August 2008).

I noticed an error with your treatment of GDP deflators (it may also apply to your treatment of monthly income). GDP deflators are quarterly, so obviously some extrapolation is required to arrive at monthly values for the purposes of our regression. The way you did this is (as far as I can tell) to assume the GDP deflator increases linearly (fair assumption) and then arrange your monthly numbers so that the GDP deflator for the first month of any given quarter is equal to the reported GDP deflator for that quarter.

This is unacceptable because quarterly GDP deflators are a report of the average GDP deflation over that entire quarter, not the average GDP deflation over the first month of that quarter (as you have it in your data). For example, consider Q12002. The reported quarterly GDP deflator is 103.568. You have in your data: January 2002: 103.658, February 2002: 103.691, March 2002: 103.815. These three values average to 103.721, which is well above the GDP deflator for Q12002, 103.568. Essentially, your method overestimates monthly GDP deflators.

A better method is to calculate monthly GDP deflators such that such that the GDP deflator increases linearly but the three monthly GDP deflators for any given quarter average to that quarter’s reported GDP deflator. So, I compute GDP deflator for January, February, and March 2002 as follows (bracketed periods represent the GDP deflator over that period, for example [March2002] = the GDP deflator for March 2002):

[January2002] = [Q12002] – ([Q12002] – [Q42001])/3

[February2002] = [Q12002]

[March2002 = [Q12002] + ([Q22002] – [Q12002])/3

This method produces monthly GDP deflator data which is consistent with the reported quarterly GDP deflator data. I applied the same method to calculate monthly disposable income from quarterly values (with slight variation for quarters in which monthly disposable income declined instead of rose).

Good luck with Chamberlain Economics, and I hope it’s alright if I list this post as a reference for my paper (along with the rather famous paper it mentions, obviously).