## Monday, August 22, 2005

### Updated Evidence on the Productivity Puzzle

A mid-level geek alert applies to this post as it relates to a data issue.  I promised to follow up on this post looking into why the GNP/NI ratio varies systematically through time.  In that post, I plotted the deficit/NI ratio and showed how closely it co-varied with GNP/NI.  Here’s a graph of G/NI and GNP/NI which is an even closer fit:

I still need to figure out why these move together.  I’m finding out all I don’t know about how the NIPA accounts are actually calculated by the BEA.  I hope I’m not missing something simple!  In any case, there’s still more to follow.  One way or the other, I want to understand why GNP/NI varies through time.  Repeating from the earlier post, here's how these series are related:

Recall that S = I + (G-T) + NX. Then S/NI = I/NI + (G-T)/NI + NX/NI. Replace S to obtain (GNP-C-T)/NI = I/NI + (G-T)/NI + NX/NI so that GNP/NI = C/NI + T/NI + I/NI + (G-T)/NI + NX/NI (this is just GNP=C+I+G+NX divided by NI). But this just relates the terms, it doesn't explain why GNP/NI varies, nor why GNP/NI and G/NI move together.

Update #1:  I should have put these in from the start.  Following up on PGL's comment, NNP (net national product) is GNP-depreciation.  NI (national income) is NNP-indirect business taxes (e.g. sales taxes).  Then NI = GNP - depreciation - indirect business taxes.  So the difference between GNP and NI is depreciation + indirect business taxes, and there is also the statistical discrepancy to account for.  But, again, why does this vary with G? I keep getting the feeling there's something simple I'm missing.  I hope at some point to relate this back to productivity (which measure, income or product, is used matters), but for the moment I've become interested in explaining the pattern Brad DeLong first identified here.

Posted by on Monday, August 22, 2005 at 12:33 PM in Economics, Methodology | Permalink  TrackBack (2)  Comments (6)