Tuesday, December 6, 2016

A Look at Causation in Bezemer and Hudson's "Finance is Not the Economy"

Revised after posting

Bezemer and Hudson:
In Figure 2, based on calculations by Dirk Bezemer, Maria Grydaki, and Lu Zhang (2016), we plot the correlation of income growth with credit stocks scaled by GDP. This provides a proxy for the growth effect of credit over time. The trend is downward from the mid-1980s, and from the 1990s the correlation coefficient is not significantly different from zero. Credit was no longer “good for growth,” as many had for so long believed (from King and Levine 1993 to Ang 2008).

A major reason for this trend was that credit was extended increasingly to households, not business.

The growth effect of credit has been downward since the mid-1980s, they say. The reason they give for the decline is that "credit was extended increasingly to households". The argument makes sense. I mean, households are consumers, not producers. It's almost intuitive, now they point it out.

I hate arguments like that. Dangerous arguments. Arguments that make so much sense you think you don't need to actually look and see if they are supported by the data.


Consider household debt relative to other, more comprehensive measures of debt:

Graph #1: Household Debt as a Percent of Private NonFinancial (red) and Total Debt (blue)
The blue line shows household debt as a share of total credit market debt: drifting downward from a 30% share in the 1960s. Yes, up a little just before 1980, but that didn't cause any Global Financial Crisis. Yes again, up a little after 2000. By the timing of it, you might say this was the increase that caused the crisis.

And yet, household debt runs higher on the graph from the early 1960s to the early 1980s than it was at its pre-crisis peak. Household debt was a smaller share of total debt on approach to the crisis than it was in the early 1960s and the two decades that followed. Given this fact, my view can only be that the increase in total debt was the real cause of the crisis. Household debt just went along for the ride.


The red line on the graph shows household debt as a share of private non-financial debt. Private non-financial consists of non-financial business debt and household debt. For half a century, from the early 1950s to the early 2000s, household debt was about half of private non-financial debt. The other half was non-fi business debt.

It varied, of course. Household debt ran as low as 46% and as high as 54% of private non-financial. But until the 2000s it was always within four points of the 50% level.

Looking at that red line on Graph #1, you could say that the household share has increased since the mid-1980s. There was a brief dip in the latter 1990s, but the trend was upward from the mid-80s to 2006.

Bezemer and Hudson say the "growth effect of credit" has been falling since the mid-1980s. They say that the growing household share is responsible for this decline. But in the mid-80s, household debt was at the bottom of its normal range. It had nowhere to go but up.

Household debt was on the low side of private non-financial for two decades, from 1970 to 1990. After it escaped the low side, climbing above the 50% level in 1991, it remained within its normal range (below 54%) for a decade.

Based on Graph #1 we can say Bezemer and Hudson seem to suggest that the increase from low normal to high normal was the problem, and that the fall of the growth effect began the moment household debt started rising from bottom in the mid-80s. But that view just doesn't make sense.


The household share was at bottom in the mid-80s. Are Bezemer and Hudson saying it was an insufficiency of household debt that caused the "growth effect" to fall? Obviously not. But perhaps they should.

When household debt reached peak share in the mid-90s, it was no higher than it was in the mid-60s. It is interesting to note that both of those highs occurred during good economic times.

Graph #5: Household Debt relative to Private Non-Financial Debt
This coincidence should raise doubt about Bezemer and Hudson's dangerous argument that a high household share of debt is the cause of our economic troubles.

If our troubles started in the mid-80s, as Bezemer and Hudson say, then perhaps the source of troubles is not that household debt is a high percentage of private non-financial, but that household debt is just plain high, like private non-financial debt, and private financial debt, and credit market debt in general.

Monday, December 5, 2016

Suddenly pervasive?


Yesterday I quoted Bezemer & Hudson from Finance is Not the Economy. Here again is that quote:
Growth in credit to the real sector paralleled growth in nominal U.S. GDP from the 1950s to the mid-1980s — that is, until financialization became pervasive. Allowing for technical problems of definitions and measurement, growth of bank credit to the real sector and nominal GDP growth moved almost one on one, until financial liberalization gathered steam in the early 1980s.

Credit growth paralleled GDP growth until financialization became pervasive.

What's the meaning of those four bold words?

Could it mean that the financialization was gradual, and gradually became pervasive? That's more or less what happened, as I see it. We had creeping financialization all along, visible in the data even in the early 1950s. But it didn't become pervasive, or problematic, say, until the 1980s. Or, until the 1970s. Or, until the 1960s. It depends who you ask.

Come to think of it, "problematic" is the better word for the view that I hold. But Bezemer and Hudson say "pervasive". So now I'm not so sure they are describing the view I hold. They seem to be describing a different view, one in which financialization suddenly emerged everywhere in the 1980s and was suddenly a problem everywhere. They say finance grew "almost one for one" with GDP until the early 1980s. And then it was suddenly a problem everywhere.

I'm not comfortable with that view. I don't think it is realistic. I think it is rather like telling a story of the bad witch who suddenly gains control of the village. It's a story for children. Financialization didn't happen overnight.

Total Credit to the Real Sector as a Percent of GDP
In my view the change only seemed to be sudden; that is how it seemed, not how it was in fact. It's perception versus reality. For we could live with finance. And we could live with more finance, with financialization. To some degree, finance benefits us even now.

I think finance was always pervasive: It was always everywhere in our economy. But in the 1950s there was only a thin layer of finance, insignificant and beneficial. In the 1960s we still found it beneficial, though harmful effects arose as finance deepened. Then in the 1970s, though debt grew bigger and more problematic, the cost of it got lost in inflation. Finally, in the 1980s as inflation subsided and debt appeared to be burgeoning, we were suddenly able to see the harmful effects that had long been there.

As Bezemer and Hudson put it: "Credit was no longer 'good for growth,' as many had for so long believed".

Credit was no longer good for growth because we got too much of it and it was choking us. That much should be obvious. The more subtle point is that credit grew all thru the 1950s and 60s and 70s and 80s, and the nineties and the naughts. Credit grew because it was our policy to make credit grow, because we thought credit was 'good for growth'.

Financialization didn't happen by accident. We made it happen.

Sunday, December 4, 2016

If the growth of debt was one-for-one with the growth of GDP before the 1980s, the line would be flat


Bezemer & Hudson:
Growth in credit to the real sector paralleled growth in nominal U.S. GDP from the 1950s to the mid-1980s — that is, until financialization became pervasive. Allowing for technical problems of definitions and measurement, growth of bank credit to the real sector and nominal GDP growth moved almost one on one, until financial liberalization gathered steam in the early 1980s.
FRED:


If the growth of debt was one-for-one with the growth of GDP before the 1980s, the line would be flat before the 1980s.

Bezemer and Hudson say that the relation was one-for-one until the early 1980s. I don't see it. And this graph shows nominal values. If I showed inflation-adjusted values the increase would be steeper, especially in the 1965-1984 period that you want to say is flat even though it isn't.

I didn't make it up. It's BIS data, from FRED (copyright BIS).


Here is Figure 1 from the B&H article. It follows immediately after the paragraph I quoted above.

Figure 1 from the Article by Bezemer and Hudson
There are three lines on the graph. Forget the jiggy ones, the YOY lines. Just look at the dashed line that runs from lower-left to upper-right. It's not a perfectly straight line but it is pretty straight, and runs consistently uphill.

The dashed line shows the "cumulative difference between credit growth and income growth". It's always going uphill. This means the cumulative difference is always increasing. In other words, from 1953 to 2010 credit growth was consistently faster than income growth.

That's what the dashed line shows.


If you need a better look at change-in-debt relative to change-in-GDP, back in October I showed exactly that. It's a different, more comprehensive measure of debt than Bezemer and Hudson consider. But it shows what's going on. There is a slowdown of debt growth in the latter 1980s, and another during the crisis. But it's all uphill from 1952 to the mid-80s. Just like that dashed line of theirs.

Graph #3
Apart from "technical problems" they say, debt and GDP moved "almost" one-to-one until "the early 1980s." That's not what my graph shows.

It's not what their graph shows, either.

Saturday, December 3, 2016

Picking nits


One paragraph, two nits.

Bezemer & Hudson:
Households do not receive incomes from the houses they live in. The value of the “services” their homes provide does not increase simply because house prices rise ...

The value I get by living in my home surely must be measured in terms of the replacement cost of the home. Surely I would not get the same value, living in a cardboard box under a bridge. Oh, and putting the word services in quotes as they do, to belittle it, does not make their case stronger.

As a blogger who tries to write every day about the economy, I can tell you this: If the argument is weak, don't use it. Find a better argument, or find a better analysis.

The financial sector does not produce goods or even “real” wealth. And to the extent that it produces services, much of this serves to redirect revenues to rentiers, not to generate wages and profits.

B&H seem to see "rentiers" as a class or caste of society. In the article they say "An economy based increasingly on rent extraction by the few and debt buildup by the many is, in essence, the feudal model". Lords and serfs. The rentiers and the rest of us.

Well maybe that's how it is today. But it's not how the problem started. And it's not entirely like that, even now. Hey, I got my nine cents interest on my savings account. That makes me a nine-cent rentier.

The thing is, it's not that some people are rentiers and others are laborers. That's not it. It's that there are different kinds of income. And most, or many, or at least some people receive a mix of different kinds of income. Adam Smith defined wages, and profit, and rent. These are categories of income, not kinds of people.

In the dog-eat-dog, decline-of-civilization world we live in, we take the categories of income and use them to define people. That's wrong. And it doesn't help.

Adam Smith wrote:
I.6.19:
When those three different sorts of revenue belong to different persons, they are readily distinguished; but when they belong to the same they are sometimes confounded with one another, at least in common language.
I.6.23:
A gardener who cultivates his own garden with his own hands, unites in his own person the three different characters, of landlord, farmer, and labourer. His produce, therefore, should pay him the rent of the first, the profit of the second, and the wages of the third. The whole, however, is commonly considered as the earnings of his labour. Both rent and profit are, in this case, confounded with wages.

I read somebody one time who wrote that Smith's thinking was "class-based". Bullshit. Adam Smith's thinking was cost-based. It is we, misinterpreting Smith, who are the class-based thinkers.

Friday, December 2, 2016

Why figure inflation-adjusted debt?


I asked when is "real GDP" useful?

Investopedia said:

Economists use real GDP when they want to monitor the growth of output in an economy.

Real GDP is useful when you're looking at economic growth.

Why?

Nominal GDP, typically referred to as just GDP, uses the quantities and prices in a given time period to track the total value produced in an economy in that same span of time. Conversely, real GDP tracks the total value produced using constant prices, isolating the effect of price changes. As a result, real GDP is an accurate gauge of changes in the output level of an economy.

Real GDP removes the price changes, in order to see more clearly the changes due to greater productive capacity.

Yeah.

I had some fun with the idea a few posts back, when in addition to holding prices constant, I held population constant in order to see more clearly the changes due to greater productive capacity. (Yeah, I know: Economists have a measure like that already.)

But if you want to see economic growth, you definitely want to look at GDP with prices held constant. I look at debt with prices held constant for the same reason.

An increase in debt indicates that some extra spending happened. "Extra" in the sense that it couldn't have happened if some new borrowing didn't happen. The extra spending contributes to economic growth. The new borrowing adds to accumulated debt. When I'm considering economic growth, I want to look at debt with prices held constant.

//

Why figure inflation-adjusted debt? Because we want to know about economic growth.

People say you can look at a graph of debt relative to GDP and see that debt was "flat" before the 1980s. Yeah, you can, but it's not inflation-adjusted. So you can't look at that graph and learn anything useful about economic growth.

You can learn something harmful. You can start thinking that debt didn't grow faster than output in those days. That's harmful, because it's not true. The "flatness" we see on that graph is a result of inflation.

Do the math.

Thursday, December 1, 2016

A new "real debt" calculation


Nominal (blue) and Real (red) Ratios of Debt-to-GDP
I thought of an easier way to do the "real" calculation for data that accumulates over multiple years. Debt is the obvious example of something that accumulates over multiple years.

Here's the calculation:

Take the change in debt as a share of nominal GDP, multiply it by the real GDP value for that period, and add the previous period's Real Debt value. That's it. That's the whole calculation.

It's a proportions thing. The change in debt is re-sized from a certain proportion of nominal GDP to the same proportion of real GDP. The purpose of the exercise is to determine the real (inflation adjusted) purchasing power of the additions to debt over time. This is a key part of the puzzle when one is considering economic growth.

The only data series needed are nominal GDP, real GDP, and the debt series you want real values for. Three columns of data. And you can do the calculation in a single step if you want, so just one column is required for that.


To me this seems a simple calculation -- at least compared to my old method. Let me go over it again:

Figure the change in debt relative to NGDP, then multiply by the RGDP value, and add up the values as you go. This gives you inflation-adjusted debt. If instead you figure the change in debt relative to NGDP, multiply by the NGDP value (instead of RGDP), and add up the values as you go, you get your nominal debt series back. The calculation works as it should.

If I was looking at consumer debt I'd want to use Disposable Income and Real Disposable Income instead of NGDP and RGDP. The method is flexible that way.

Also, you can use quarterly data. With my old method I was stuck using annual data for some reason.

And you don't need to use the GDP Deflator series. The new calculation uses the nominal and real GDP values instead. That's an improvement because the deflator creates rounding errors. If you use the deflator to figure real GDP, your values will differ a bit from the real GDP you get from FRED. Using the new calculation, the rounding error would show up in our deflator, if we calculated one. Since we don't need to show a deflator, the rounding error goes away.


Here's a usage tip: Insert a blank row between the data labels row and the first row of data. This way, you can use the same "change-in" calculation for the first data item and all the rest in that column. The first "change-in" value will be the first given value minus zero, which is exactly what you want.

And remember that for the base-year part of the calc, you probably need a dollar sign before the row number, so the row number doesn't change.


For those who may be interested, my first hard look at the inflation-adjustment of debt is in a series of five short posts beginning with an unnumbered intro. (The last post in series is number 4.)

My second hard look at it (four posts) begins here.

Here is a post that links to a spreadsheet with my old calculation.

And here is an Excel file that uses the new calculation.

Tuesday, November 29, 2016

A response to Bezemer & Hudson's "Finance is Not the Economy"


I've been reading Finance is Not the Economy by Dirk Bezemer and Michael Hudson. It's really good. But we need to talk.

You know how, when you look at a graph of debt relative to GDP, the line starts out flat and then suddenly starts going up in the 1980s? Like this graph:

Graph #1: TCMDO Debt relative to GDP
Some people look at a graph like that and say there was an "explosion" of debt in the 1980s. Noah Smith said something like that a while back.

Funny, though. If you look at change in debt relative to GDP, you see a different picture:

Graph #2: Change in TCMDO Debt, relative to GDP
If you put a trend line on this one, it would start low and begin curving up immediately. It would continue curving upward all thru the 1960s and 1970s and into the 1980s. Then, after 1985, a sudden drop.

Why does the first graph run flat when the second graph shows a trend of continuous increase until the mid-80s? I mean, the second graph shows the changes in debt, and the first graph shows the accumulation of those changes. So, why does the first graph run flat?

One word: Inflation.


For near twenty years, beginning in the mid-60s, inflation pushed prices up. It pushed income up. It pushed spending up. And it pushed borrowing up.

Inflation pushed borrowing up. It made new additions to debt bigger. But inflation didn't make existing debt bigger. Inflation only affects new spending.

Inflation also made nominal GDP bigger. All of GDP is new spending, so all of GDP got bigger. Inflation increased the whole of GDP, and new additions to debt, but not existing debt.

When we look at Graph #2 we see big increases before 1980. But those increases don't appear on Graph #1 because the inflation in any given year does not increase the debt of prior years. That's why people say "inflation erodes debt".

Graph #2 goes uphill before the 1980s, but Graph #1 is flat because of inflation.


So anyway: You know how, when you look at a graph of debt relative to GDP, the line starts out flat and then suddenly starts going up in the 1980s? Everybody knows. But no one seems to know it's flat because of inflation. Everyone seems to think there was a magical time before the '80s when the economy grew fast enough to keep the line flat.

Can you see where this goes? It goes to policy. If you think Debt-to-GDP was flat because we used less debt and got more bang for the buck, or if you think it was flat as a result of inflation, it can change how you think about policy. Hey, I'm all in favor of changing policy. But I'm also in favor of avoiding wrong-thinking on the way to developing policy.

Funny thing is, we did use less debt and we did get more bang for the buck, back when there wasn't so much debt. It's true. But it's also true that the line is flat because of inflation.


So I've been reading Finance is Not the Economy by Bezemer and Hudson. Their key idea seems to be that money and credit (and debt) grew in proportion to GDP until the 1980s, and that since the 1980s money and credit (and debt) grow much faster than GDP. The pattern they describe -- a time of stability followed by a time of increase, with the change occurring in the 1980s -- is exactly what we saw in Graph #1 above.

This worries me.


If you look at debt relative to GDP and you want to base policy on it, then you have to take inflation into account. You have to look at "real" debt relative to "real" GDP. Otherwise you cannot be sure what the numbers are telling you.

But every graph of "real debt" that I ever saw was wrong. Everyone inflation-adjusts debt the same way you would inflation-adjust GDP. That's wrong. It's wrong because debt is a stock and GDP is a flow. Debt accumulates over many years. GDP accumulates for one year and starts at zero again the next year.

There's nothing in this year's GDP that is "final spending" from some prior year. Everything in this year's GDP is this year's stuff and can be figured at this year's prices. If you want to figure "real GDP" using base year 2009, you take one year's GDP, divide by that same year's price level, and multiply by the price level for 2009. Done.

That calculation doesn't work for debt. This year's debt includes this year's addition to debt, and last year's addition to debt, and the year before that, and debt from all the way back to the oldest thing that isn't paid off yet. So when you want to figure "real debt" -- when you want to figure the real purchasing power at the time the money was borrowed -- you have to adjust each year's addition to debt separately. If you don't do that, your answer is just plain wrong.

If I was writing the Bezemer and Hudson article, and if I didn't have this fetish about the real purchasing power of additions to debt, I would not have thought twice about it. I would have looked at debt relative to GDP, nominal to nominal. And existing debt from the time of the Great Inflation (and before) would be falsely low, because inflation erodes debt.

If you challenged me on it, I would have said the same thing people always tell me: I would have said the graph of "real debt to real GDP" looks exactly like the graph of "nominal debt to nominal GDP" because inflation cancels inflation.

That's not right, though, because debt is a stock and GDP is a flow. The "real" calculations for stock and flow are different. Stock-inflation divided by flow-inflation does not cancel out.

If you figure nominal debt to nominal GDP, or if you use the wrong "real" calculation, you can expect debt to look flat before the 1980s. I've been reading Finance is Not the Economy and I'm worried that Bezemer and Hudson find debt growing in proportion to GDP before the 1980s because they are using nominal values for debt, like my Graph #1.

Actually, they are. They said so. Their Figure 1 shows "YoY growth in nominal credit to nonfinancial business". They're using nominals. So their key insight, that debt grew in proportion to GDP before the 1980s, may well be mistaken. We won't know until they re-create their graphs with the data correctly converted to real values.


More on the inflation-adjustment of debt:

  •  Illusion, Reality, and the Growth of Debt
  •  By the numbers (2): History is different


EDIT 30 Nov 2016: Revised the first sentence with the word "erode".

Saturday, November 26, 2016

The Archdruid on Free Trade


The Archdruid Report: The Free Trade Fallacy

It’s not always remembered that there have been two great eras of free trade in modern history—the first from the 1860s to the beginning of the Great Depression, in which the United States never fully participated; the second from the 1980s to the present, with the United States at dead center—and neither one of them has ushered in a world of universal prosperity. Quite the contrary, both of them have yielded identical results: staggering profits for the rich, impoverishment and immiseration for the working classes, and cascading economic crises.


Since wages are a very large fraction of the cost of producing goods, the overall decrease in wages brings about an increase in profits. Thus one result of free trade is a transfer of wealth from the laboring majority, whose income comes from wages, to the affluent minority, whose income comes directly or indirectly from profits. That’s the factor that’s been left out of the picture by the proponents of free trade—its effect on income distribution.


Getting rid of free trade and returning to a normal state of affairs, in which nations provide most of their own needs from within their own borders and trade with other nations to exchange surpluses or get products that aren’t available at home readily, or at all, gets rid of one reliable cause of serious economic dysfunction.


Recommended reading.

Friday, November 25, 2016

It was all over but the funeral when the red went above the blue


Graph #1: Growth of Financial (red) and NonFinancial (blue) Business Debt, relative to GDP
https://fred.stlouisfed.org/graph/?g=bRAv


That's when it was all over. But when did the trouble start? The next graph shows the blue relative to the red:

Graph #2
The blue line here is the blue (from Graph #1) relative to the red (from Graph #1).
The red line here is a Hodrick-Prescott, showing decline since the beginning.
If you look again, you might see an uptrend in the early 1960s which appears in the Hodrick-Prescott as a brief flattening. If I started the H-P in 1960 it would start with an uptrend.

Graph #3
The ratio doesn't always have to trend downward.
The early 1960s were pretty good, by the way.