I think it's time to do some looking at the Milton Friedman argument for deficits that conservatives have scrambled to embrace of late. You know, "deficits tend to decrease spending; specifically, they tend to constrain non-defense domestic spending," and so on.

If this is true, you'd expect to see surpluses correlated with spending, and deficits inversely correlated with spending.

I loaded 1962-2001 data on government spending as a proportion of GDP from the tables of section 8 into Excel and ran some correlations. I deleted the bizarro TQ, because I'm too lazy to fix up the results for that correctly, and it doesn't change much anyway. A few items of note:

• The correlation between surpluses and non-mandatory non-defense spending is weakly negative.
  
• There is no correlation between revenue and non-mandatory non-defense spending - the "big government" conservatives complain about.

These both contradict Freidman. Here's one that doesn't:

• The correlation between interest outlays and non-mandatory non-defense spending is weakly negative.

So a fallback hypothesis, that interest payments constrain spending, matches the evidence pretty well.

On closer inspection, interest rates, which have little to do with the matter under discussion, have extremely strong effects on interest payments. Year-to-year changes in the annual deficit don't do much to the size of the total deficit, after all, so annual interest payments are barely affected by short-run changes in the deficit; it's almost entirely interest rates. These rates are mostly the result of inflation and long-run growth expectations; altogether, this means interest payments probably aren't a very good measurement of the delta in congressional spending from revenue changes.

I don't think the evidence is with Friedman here.

Update: Megan McArdle makes some good points about consumption smoothing and forward-looking interest rates, but they don't really change anything.

To see why, let's assume that interest outlays constrain domestic spending. Interest payments are spending, too (they're certainly not free), so I added them to the spreadsheet. Interest payments + domestic spending starts at 3.7% of GDP in 1962, crawls up to 6.6% of GDP in 1980, and then stays there until 1995, after which it slowly falls to 5.2% in 2001. Does that look like cheaper government through deficits to you? The correlation between deficits and interest+domestic is strongly negative, already - what are conservatives trying to do? Make the government spend just as much tax money as before, except on bond payments?

Welfare for rich people!
- posted by Jason @ 5:30 PM

 
Something you won't hear in all the neocon blathering about how "everyone hates Germany": after the 20 point drop of the last year, 71% of those polled have a favorable opinion of them.

Coincidentally, 29% is close to the proportion of strongly identified Republicans out there.
- posted by Jason @ 4:35 PM