Thursday, December 8, 2016

A Look at Non-Financial Debt in Bezemer and Hudson's "Finance is Not the Economy"

I searched the B&H article for the terms "non-financial" and "nonfinancial", copied the interesting sentences all to one place, organized them, and decided I like what Bezemer and Hudson are saying more than I thought. I'll go through some of those excerpts now.

Bezemer and Hudson:

Federal Reserve economists note that many contemporary “[a]nalysts have found that over long periods of time there has been a fairly close relationship between the growth of debt of the nonfinancial sectors and aggregate economic activity” (BGFRS 2013, 76).

These correlations suggest a one-on-one ratio between bank credit and the non- financial sector’s economic activity (Figure 1). Growth in credit to the real sector paralleled growth in nominal U.S. GDP from the 1950s to the mid-1980s...

Conveniently, FRED offers a series for nonfinancial business credit. The blue line on Graph #1. The red line is GDP. Remember, B&H say "Growth in credit to the real sector paralleled growth in nominal U.S. GDP from the 1950s to the mid-1980s". Do you see parallel lines?

Graph #1: Comparing Non-Financial Business Debt (blue) and GDP (red) indexed to 1951
The red and blue start out the same. Or at least they look the same. (The lines are so close to zero it's hard to say anything with confidence.) If they're the same, that's the "parallel" that B&H were talking about.

These "parallel" lines separate by 1969. That's well before the mid-80s. And at the default line width (which is narrower than shown here) the start-of-separation moves to the mid-1960s.

You might be willing to say the lines are parallel, depending on how much you want Bezemer and Hudson to be right, and how rigorous you are about the meaning of words.

I just thought of something not quite relevant, but I'll quote it anyway:

The classical theorists resemble Euclidean geometers in a non-Euclidean world who, discovering that in experience straight lines apparently parallel often meet, rebuke the lines for not keeping straight ...

If we pin the lines together in 1960 instead of 1951, the separation is still visible by 1969. If we chop things off at the mid-80s to get a better look at the early years, the separation is visible back to the mid-1950s.

Non-financial debt growth is faster than GDP growth for the whole period shown on the graph. The lines are not parallel. B&H take too many liberties: Their words lack rigor, and they offer wishful thinking as evidence.

I am not pleased to say so.


There are ways to compare two lines, other than just eyeballing them to see how parallel they seem to be. We can take a ratio: Divide one line by the other. Then, for example, we can see non-financial business debt relative to GDP. Or we could multiply the ratio by 100 and call it non-financial business debt as a percent of GDP.

If we do this and the ratio runs flat (at 40%, say) it would mean that GDP and debt run "parallel". In other words, they are growing at the same rate. Even if the ratio is only pretty flat, for the years that it returns repeatedly to the same [40%] level and doesn't wander far from it, we can say GDP and debt are growing at the same rate. I'm willing to take liberties that far.

Here's non-financial business debt as a percent of GDP:

Graph #2: Non-Financial Business Debt as a Percent of GDP (blue)
I marked it up a little. Nonfinancial business debt increased from 30% of GDP in 1952 to 50% of GDP in 1975 and to 70% of GDP in 2015. From 30% to 50% in 23 years; from 50% to 70% in 40 years. The increase is substantially faster in the earlier period, perhaps because there was less debt to begin with.

I was going for round numbers: 30%, 50%, and 70%. I could have eyeballed-in (or even calculated) better trendlines if I wasn't constrained by my round numbers. So I'm not showing Graph #2 as a perfect presentation of relative growth trends. I've... taken liberties. But the graph is good enough to give you the idea. The increase slowed after the 1974 recession.

If you relocated the red lines to suit your best estimate of trend, you'd still see a change somewhere in the mid-1970s, and a steeper slope before the change than after. Nonfinancial business debt was growing faster than GDP for all the years shown, and faster in the early period than in the late.

This presents a problem. For in the B&H paper they say that

growth of bank credit to the real sector and nominal GDP growth moved almost one on one, until ... the early 1980s.

They say the line was flat before the change. It's not. And then

While the total credit stock expanded enormously in the 1990s and 2000, credit to nonfinancial business was stagnant at about 40 percent of GDP...

They say the line was flat after the change. It's a stretch.

Granted, the quotes apply to two different measures of debt. That sort of slipperiness is part of what makes it hard to pin down what Bezemer and Hudson really mean. Their sentences slide from one context to another almost continuously. That makes for a very good story until you sit down to write about it. Then you find that no two of their thoughts fit cleanly together.

So if you are not pleased with my evaluation of their work, do sit down and try to piece together your own orderly montage of their thoughts. Let me know how it works out.


The two measures of debt referenced in the above quotes are "credit to nonfinancial business" and "credit to the real sector". The latter includes the former, and also includes household debt. I'd say it includes government debt as well. But I won't guess what's in the debt measure Bezemer and Hudson used. I know government debt doesn't play a large role in their article.

Household debt includes consumer credit and, increasingly, household mortgage credit. Bezemer and Hudson have a strong focus on mortgage debt. They imply that mortgage debt crowded out non-fi business:

While the total credit stock expanded enormously in the 1990s and 2000, credit to nonfinancial business was stagnant at about 40 percent of GDP, while its share in overall credit plummeted. By contrast, the share of household mortgage credit issued by banks rose from about 20 to 50 percent of all credit.

In our time, arguably the most significant form that rent extraction has taken is in the household credit markets, especially household mortgages. The contrast is with loans to non-financial business for production.
B&H are concerned about the growth of mortgage debt, comparing it and non-fi business debt as shares of, well, as shares of different things. Here are the two debt series, shown as percent of TCMDO -- the series formerly known as total credit market debt owed:

Graph #3: Household Mortgage (red) and NonFinancial Business (blue) Debt as % of "All Credit"
Mortgage debt (red) runs consistently below 20% of "all credit" until 2005. It reaches a maximum of 21% in 2006, then falls. Mortgages never come close to "50% of all credit" in the U.S. data. Good grief!

Non-financial business debt (blue) climbs to near 35% in 1974, then falls consistently till the crisis. But clearly in the U.S. data, the fall in the blue line is not matched by a rise in the red line. Non-financial business debt was not crowded out by mortgage debt. And the "plummeting" in the 1990s and 2000s is curvier but not faster downhill than the plummeting between 1974 and 1990.

I should say I had to guess what data to use for these graphs, and sometimes B&H use multiple-country averages that I cannot duplicate. I'm just using U.S. data from FRED. Still, Graph #3 doesn't show anything comparable to what Bezemer and Hudson describe. Nor does Graph #1. Nor does Graph #2.

I can't duplicate the Bezemer and Hudson graphs. But I can inspect the U.S. data for the changes and levels and patterns they describe. If their argument is valid, I should be able to find some evidence that supports it in the graphs that I make, the graphs of U.S. data.

I don't find that evidence.


jim said...

In the interest of accuracy:

Bank credit is not the same thing as Credit to Private Non-Financial Sector

And it looks like the ratio of bank credit to GDP was fairly flat (around 32% of GDP) until the 1990's and then the ratio went up.
Bank credit did start to rise sharply in the early 1980's but then collapsed as a result of the S&L meltdown.

And Bezemer and Hudson are completely wrong about bank credit pushing up mortgage debt. The increase in mortgage debt was entirely due to funding by private investors - mostly thru Private Mortgage Conduits (AKA private label mortgage backed securities) If you subtract out the mortgages financed by private investors the period from 1990 to the present would have been below normal for mortgage debt.

It was those mortgages that banks would have never funded that were responsible for jacking up house prices because without that funding source almost all of the excess house sales would never have happened.

And it was those mortgages funded by private investors that have almost all now failed while the mortgages funded by banks proved to be pretty sound and have mostly endured.

jim said...

If you subtract out mortgages financed by private conduits there is a pretty steady decline in the ratio of mortgages to total debt from the 1960s to the present

Postkey said...

Interesting comments, thanks.