Saturday, December 21, 2013

Between the surges

Yesterday I said it's obvious from the up-trend on the graph that debt was a developing problem already in the 1950s.

Graph #1: Consumer Debt (on a Log Scale)
What I meant was: If debt became a problem because debt got so big, then the problem arose not after debt got big, but when the increase first started.

It's obvious, I think.

See that flat spot in the recent years, in the upper-right corner of Graph #1? That's when the growth of debt finally slowed. That's when people decided they had enough. It caused quite a disturbance in the economy, that slowing of debt growth.

Here's how it looks to me: The debt growth slowdown created big problems because debt plays such a big role in the economy. I don't know the numbers, but suppose 80% of our spending involves debt. Now, imagine that the debt growth slowdown happened much earlier instead -- in the 1960s maybe -- when perhaps only 20% of our spending involved debt. The effect of the debt growth slowdown would have been much less.

It's obvious, I think.

Graph #1 shows consumer debt, the log of consumer debt. It shows no ratio. It provides no context for the debt. In context, there would be more variation in the uptrend. It wouldn't just go up and up. In context, it looks like this:

Graph #2: Consumer Debt in Context (Relative to the Size of the Economy)
It's not just up-and-up on Graph #2. It's up and down and wiggle around, that's true. But you know, Graph #1 shows a pretty consistent up-trend. And Graph #1 doesn't show anything but consumer debt. It shows consumer debt, and you can see a consistent up-trend in that debt.

Graph #2 doesn't show the same uptrend. But Graph #2 doesn't only show consumer debt. It also shows the size of the economy as given by GDP. And, you know, if consumer debt shows a consistent uptrend, and consumer debt in context does not, then maybe the disturbance of the trend comes from the context data. Not from the consumer debt data.

The changes in GDP throw off the ratio and make it look like consumer debt doesn't have a consistent up-trend. But it's GDP that's throwing it off. It's the context that distorts the picture.

It's obvious, I think.

Scott Sumner looked at a graph like my #2 here, a while back. Here's what he said about the graph:

What do you see? I suppose it’s in the eye of the beholder, but I see three big debt surges: 1952-64, 1984-91, and 2000-08.
Okay. But remember, he's looking at a disturbance imposed on the up-trend by a context variable.

If you look at the same data, total consumer debt relative to GDP, and also at the change in consumer debt relative to GDP, a more complete picture emerges:

Graph #3: Consumer Debt relative to the Size of the Economy (blue) and
Change In Consumer Debt relative to the Size of the Economy (red)
Scott Sumner identified three "debt surges" on the graph. I've emphasized in heavier red the two intermissions between those surges. The first I show from 1965 to 1983, and the second from 1992 to 1999. Both intermissions match up with relatively flat spots in the blue line, the flat spots between Sumner's surges.

During the second intermission, the red line is relatively low. The change in debt is relatively low, relative to the size of the economy. This intermission, then, seems to support Sumner's view that there was no debt surge in the 1992-1999 period.

The first intermission is more problematic. Yes, there is a low spot in the red line just before the 1970 recession. (The blue line is actually going down at that time.) But after the 1970 recession there is a peak at least as big as any earlier peak in that red line. And after the 1974 recession is an even bigger peak. This peak rivals the one at the start of Sumner's second surge; it rivals the mid-1980s peak. But the blue line shows only a very mild increase during the post-1974 peak, not a strong increase like the one in the mid-1980s.

Why this discrepancy?

The behavior of the context data is the key to this puzzle. That flat spot in the blue line from 1965 to 1983, that flat spot corresponds to the time called the Great Inflation, the time that prices were rising rapidly. Incomes were rising, too. And yes, new additions to consumer debt were growing also as prices climbed. But the old, existing debt did not grow with inflation. Your mortgage payment was the same month after month and year after year. Meanwhile, your paycheck increased along with prices, because of the inflation.

Each year's inflating prices helped to enlarge the new increases of consumer debt. But each year's inflating paychecks made existing consumer debt smaller in context.

It is the rapid, inflationary increase in the context data, the GDP data, that distorts the picture of debt growth during the Great Inflation. Despite the large increases to consumer debt through the 1970s, the increases to paychecks and incomes and GDP made consumer debt growth look like it slowed significantly.

Consumer debt growth slowed almost not at all during the Great Inflation. If you strip away the effects of inflation, consumer debt in context (relative to the size of the economy) shows a strong up-trend, comparable to Graph #1 above.

Graph #4: Consumer Debt In Context, like Graph #2 (blue) and
Consumer Debt In Context with Inflation Stripped Away (red)
Okay, there is a bit of a slowdown in debt growth during the Great Inflation. But only a bit, and really not like it looks on the second graph. It is not a slowdown that merits exclusion from "surge" status. We had rapid debt growth, camouflaged by inflation.

It's obvious, I think.


Jazzbumpa said...

Ahh, this brings back fond memories.


The Arthurian said...

From your link:

By my reckoning, Sumner is incapable of identifying either a debt surge or an economic boom.

That's a good line :)

Jazzbumpa said...

Yeah, I didn't hold much back. Maybe it's a little over the top.

He's not stupid, but his assessment was.

Blinded by ideology, rapt in confirmation bias, in the throws of Krugman derangement syndrome, or just a tool.

How can you sort it out?