"I argue that the increase started well before the 1980s."

Yesterday I looked at a graph from J.W. Mason's The Myth of Reagan's Debt. The graph shows "the overall and primary budget balances for the federal government since 1960," Mason says.

"The balances are shown in percent of GDP," he adds.

That troubles my friend Jazzbumpa. Jazz asks Mason

How do your graphs look as straight values, not as ratios to GDP? I'm always suspicious of the denominator effect.

Following up on his own blog, Jazzbumpa says

J.W.M. is looking at surplus or deficit as a % of GDP. Conclusions based on ratios always make me want to take a different look.

And then Jazz did take a different look. He found a source for overall and primary budget balance numbers. He made a graph comparable to Mason's, but showing budget balance in billions of dollars rather than in percent of GDP.

I grabbed Jazzbumpa's source data, stuffed it into a spreadsheet, and duplicated his graph:

Graph #1: Overall (orange) and Primary (blue) Federal Budget Balance

Comparing the "straight values" graph to Mason's "percent of GDP" graph, Jazzbumpa observed:

There may not be much distortion from a denominator effect in this specific case ...

and again:

For the period in question, this [straight values graph] does not look substantially different from JWM's graph, with data as percentage of GDP.

Okay. That's why we do graphs, you know? To see how they turn out.

I had a sudden urge then to subtract the primary balance from the overall balance, so that I might see the shape of the space between the two. The subtraction would show me the portion of the deficit caused by the payment of interest, so to speak.

It was simple enough to perform the subtraction, seeing as how Jazz had already done all the legwork.

Graph #2: Take Graph #1 and Subtract the Blue Line from the Orange Line

Well, now that is one heck of a lot smoother than what we see on Graph #1.

Allow me to remind you that Jazzbumpa created his Graph 1 as a version of JW Mason's graph that is not expressed as a ratio. Jazz omits GDP from the calculation because he is concerned that the changes in GDP will give rise to changes in the ratio, and that we (the viewers of graphs) will mistakenly attribute those changes to the numerator -- to the Federal budget deficits -- when they might in fact arise from the changes in GDP.

An excellent precaution.

However, looking at Graph #2 here, at Graph #1, and then Graph #2 again, I want to say that in Graph #1, Jazz did not manage to eliminate the denominator effect. Oh, he eliminated the denominator, for sure. But the jogs and jags on his graph are intimately related to GDP, the denominator that he eliminated.

Jazz eliminated GDP from his calculation. But that did not eliminate the effects of GDP from the Federal deficit numbers. The Federal government always runs big deficits at times of recession.

Graph #3, from November 2010

Removing GDP from the debt-to-GDP calculation does not eliminate the tendency for Federal deficits to rise during times of recession. Removing GDP from the calculation does not remove the effects of changes in GDP from the Federal budget balance.

Subtracting the primary deficit from the total deficit eliminated most of those effects. That's what the smoothness of Graph #2 tells me.


Oh, and by the way: Graph #2 shows a big increase in deficits in the 1980s, an increase that began, I'm tempted to say, in the 1950s.