Friday, September 16, 2011

Credit use was growing consistently for the full post-war period

Following up on yesterday's post.

Financial Sector Debt in the 60-year period:



Financial Sector Debt in the first 30 years of 60:



Financial Sector Debt in the first 15 years:



Credit use was growing consistently for the full post-war period.



UPDATE -- Related posts:

Disputing my read of the graphs: Total Debt at Retirement Blues.

Broadening the topic: Total Debt vs. Total Wages at Illusion of Prosperity.

Wondering why debt is so *very* exponential: Explain this: P=Ce^rt at Liminal Hack.

29 comments:

jim said...

Hi Art,
Have you looked at Minsky's Financial Instability Hypothesis? He suggests that financial credit in the current era would pass through 3 phases (Hedge, Speculation and Ponzi). Each phase would produce higher levels of debt and risk. Then when things had shifted to the last stage it would all crash in a frenzy of borrowing.

Take a look at the TCMDODFS for 1970-2008 using the "change from yr ago,in $B". The graph then moves in stages. Up to 1985 there is less than $100B of additional liabilities. Then it jumps to several billion a year of added debt. Then in 1998 it jumps to around $1 trillion/year. Then around 2006 when the housing market started to unravel it jumps again for an instant and collapses.

-jim

LiminalHack said...

Hello Art.

As you observe, the growth in credit is a fairly smooth function starting right after the war. We should therefore not look for political or ideological narratives to explain that along the lines of jim's suggestion above (sorry jim!), but rather some formulation of underlying mathematical dynamics that give rise to an exponential equation of the right form that produces the curves in those graphs.

Once the underlying dynamics and intensive variables are identified, then one could consider solutions or predictions.

Would you agree?

Clonal said...

Art,

You should look at all sectors of credit.

See Household, Nonfinancial Business, State and Local Govt and Financial sectors credit market debt

The picture is quite enlightening

Jazzbumpa said...

No, no, and more no!

1) To clonal's point, I have no idea why you chose financial service sector debt.

2) You simply cannot perform a visual evaluation of exponential growth on a linear scale and get it right. EVER! You have to look at the data in the right way.

3) Put in a log scale and you see a knee in the 80's. Also, up to 80, it levels at every recession. Not always since.

4) Look at percent change from previous year and you'll see it's nowhere near constant, vacillating wildly from 0 to 30% - except for 1990 until the great recession, when it narrows considerably - which is not at all your point.

This is not in any way constant over time.

Cheers!
JzB

LiminalHack said...

JZB, the log lines are still straight, which implies an exponential whichever way to cut it.

So the form of the compound growth equation:

P = Ce^rt

applies in all cases since WWII.

What I think is missing is an explanation of why compound growth would apply to total credit in the first place. Once we have a candidate for that, we could ruminate about what caused the change in the variable r.

The Arthurian said...

"What I think is missing is an explanation of why compound growth would apply to total credit in the first place."

Liminal, that's where I come in.

At the top of my sidebar clicp the "First Visit?" button to open a list of posts. Check out the two "Venn" posts.

I figure you have your own explanation for it...

Art

LiminalHack said...

Art, regarding the venn principle I agree.

However it tells us absolutely nothing about why the growth of total credit is a near perfect exponential, albeit one that might form time to time experience a (small) change in the exponent.

Given the mathematical perfection of the TCMDO curve (or curves, to satisfy jzb), there must be a mathematical (not political) factor underlying it.

BTW, I don't have a canned answer for this, so I'm not trolling here.

jim said...

To LimH

If the growth was exponential then the YOY change in growth would also be exponential.

It is not. For instance the period 1998-2006 the financial sector added about 1 trillion every year.

http://tinyurl.com/3vfpj7n

Minsky:
http://www.levyinstitute.org/pubs/wp74.pdf

LiminalHack said...

jim, what are you talking about? The log lines on jzb's charts are straight and that means there is an e in there.

all we need for an exponential is compounding, we don't need a change in r.

The Arthurian said...

Never read Minsky. If I had a reading list he would be on it, but I don't. (Slow reader.)

Jim, your "change from year ago in $B" graph is remarkable!! My first reaction was: This is some kind of business cycle for sure. The peaks don't seem to have anything to do with recessions, though.

Same pattern appears when corrected for inflation, TCMDODFS / GDPDEF, so inflation is not the cause.

I don't know enough about Minsky to think his work might be ideological. But if he's good, it isn't, and I expect he's good. (My work seems to remind people of his, somehow.)

I agree that what's happening with debt is definitely mathematical and definitely *not* political. I said so in Deviation from Trend.

And, Liminal, I definitely agree that a satisfactory analysis of the problem must come before solutions; otherwise you just get another Reaganomics.

Clonal, I have looked at all sectors. Jazz, I chose to look at financial service sector debt in response to what TwoMinds said in my previous post: "Note the recent rise of finance-based profits." I argued that "credit use growing consistently for the full post-war period" -- *all* credit use. But TwoMinds seems to see a lag in financial-debt growth. So I looked specifically at financial debt growth in my three graphs here.

Jazz, Excel returns a 0.99 value for R-squared, for closeness of the total debt curve to the exponential trend.

And Liminal, as long as you are not putting forth an explanation for this --
What I think is missing is an explanation of why compound growth would apply to total credit in the first place. Once we have a candidate for that, we could ruminate about what caused the change in the variable r.
Then I will say again that I am trying to do so.

// Oh, more messages since I started writing a response.

Jim, the Minsky PDF is not too long for me to read. I will go do that now. Thanks.

Art

Liminal Hack said...

Thanks for the deviation link.

Great find with the sinusoid, even if I am a little late to the party!

I'm following this on my blog too now and asking my readers to offer their thoughts here:

http://liminalhack.wordpress.com/2011/09/16/explain-this-pcert/

As my first offer for a line of investigation, is their any relationship between base money-per-capita versus total debt and that sinusoid? I think you touched on this base-money-per-capita issue before on this blog.

Oh, and may I duplicate your charts from that sinusoid post on my blog with attribution?

Liminal Hack said...

Looking for something to support my money-per-capita suggestion, I find this:

http://www.investmentu.com/images/usa_dependency_ratio.jpg

Its the USA dependancy ratio (working age pop to old/young):

peaks and troughs in 1965, 1985, 1995, 2010. That roughly corresponds with your deviance wave, though I emphasize, roughly.

Liminal Hack said...

dependency ratio

The Arthurian said...

Liminal, use it!

I got nothin' else right now.

Anonymous said...

thanks, I will, manyana.

Bedtime for me now!

Clonal said...

Jazz, Art

Looking at the graph I linked to, it is very clear that financial sector debt was growing at the fastest pace, followed by the household sector, then the non financial business sector, and finally the state and local government sector.

Given the rates of growth of financial sector debt, and household sector debt, it is not at all surprising that these two sectors were responsible for the "Great Financial Crisis"

However, the questions that seems to get lost are "Why is the debt rising exponentially?" and "Why are the growth rates so different between the sectors?"

These questions need to be thought about, along with Jazz' observations.

Calgacus said...

Arthur: Never read Minsky. Read him. I've not read too much, but his outlook is very congenial to yours, his interests and focus very similar.

Jazzbumpa said...

However, the questions that seems to get lost are "Why is the debt rising exponentially?" and "Why are the growth rates so different between the sectors?"

For the first question, it is a characteristic of a system that is growing exponentially. GDP and population both have exponential growth curves. Even reasonable use of credit should follow the same pattern.

Different growth rates by sector can relate to attitudes toward credit use (which have changed drastically over time in the household sector), access to credit, and ability to quality, for starters.

Art -

How and why you can get a .99 R^2 for something that varies so wildly across the data set is an absolute mystery to me.

Cheers!
JzB

The Arthurian said...

Jazz writes: "How and why you can get a .99 R^2 for something that varies so wildly across the data set is an absolute mystery to me."

Hey, I didn't get that result. Excel got it for me.

You have Excel. Check my work. The XLS file and the graph are available at my old Google Site:

http://sites.google.com/site/arthurianeconomics/welcome/-debt

Second worksheet lists the source data by source: Historical Statistics and a bunch of Statistical Abstracts. From a time before FRED.

Clonal said...

Jazz,

You should know by now that my questions are not supposed to be answered by giving the "obvious" answers. I am quite aware of how exponential growth occurs, and the the arithmetic of compound interest.

If you deflate each of the above curves by population growth, the cpi, and a real gdp growth index, you will still find that there is residual exponential growth -- it is there for total debt (I am not so sure as to how the component sectors stack up - as I have not done those calculations.)

So the question is related to the residual growth in debt. Where does that come from, and to whom is that growth accruing.

beowulf said...

You're ignoring the larger question, IS Clonal the reinicarnation of Leon Keyserling, or simply channeling him?
"Clonal is saying that, far from causing inflation, increases in the quantity of money may actually subdue inflation."

Keyserling also believed that orthodox anti-inflationary policy of restricting demand was, in fact, inflationary and would, thus, increase inflation, not decrease it... As to monetarism: he agreed that the money supply must be increased over time; but that high interest rates accepted by Friedman had restricted growth below maximum; that the money supply must grow in real terms instead of nominal terms; and the money supply must be increased as needed, not at a constant rate as in Friedman's monetarist model.
http://www.allbusiness.com/government/government-bodies-offices-legislative/8902193-1.html

The Arthurian said...

THE COIN GUY WAS HERE!
THE COIN GUY WAS HERE!

C: "You should know by now that my questions are not supposed to be answered by giving the "obvious" answers."

Don't ya hate it when that happens?

C: "So the question is related to the residual growth in debt. Where does that come from, and to whom is that growth accruing."

The Where: It is our policy to grow credit-use at a faster rate than money.

The To Whom:
To creditors.

LiminalHack said...

""Clonal is saying that, far from causing inflation, increases in the quantity of money may actually subdue inflation.""

I think it depends where one currently resides on the interest rate scale.

The velocity of money varies exponentially with interest rates.

The Arthurian said...

LH: "The velocity of money varies exponentially with interest rates."

How can the velocity of money increase? We cannot spend money faster than we get it. The only way is to borrow, and spend that: To create money that doesn't count through the fractional-reserve process, and spend that.

Since the money "doesn't count" -- GDP increases but M does not -- velocity has to go up.

Interest rates, as the price of credit, would be associated with velocity in this way, I think.

Clonal said...

Getting back to the exponential rise in total debt, I downloaded Art's spread sheet, looked at the log of debt. Identified four periods, regressed each period wrt time, looked at the residuals, and at 3 sigma limits around the residuals.

I get the following graphs.

Debt vs time linear scale
Debt vs time log scale

In the linear scale, as is the case generally with exponential growth, the early year changes are visually negligible.

So we look at the log scale graph, we see four very distinct segments to growth

1945-1965 growth rate 5.72% sd 2.93%
1966-1976 growth rate 10.3% sd 1.62%
1977-1992 growth rate 11.3% sd 5.06%
1993-2007 growth rate 8.25% sd 1.34%

From 1945-1976, the debt growth was less than the sum of population, and nominal gdp increases, and thus was affordable and manageable. This was not the case from 1977 onwards.

LiminalHack said...

Clonal, at no point was the rate of US population growth 10%, or anywhere near that so I have no idea what you mean by 'less than the sum of population, so ok'.

The peak rate of population growth was about 1.75%, right after the war.

Therefore I cannot see anything that makes any these periods 'bad' or 'good'.

LiminalHack said...

In fact the 1977 to 1992 phase is that in which the baby boomers come into the full productivity, hence the higher growth rate.

I think the various inflexions are all explained demographically.

None of which sheds any light on why the exponent of credit growth is so much greater than the exponent of pop growth.

Clonal said...

1945-1965 debt 5.72% GDP 6.28% Affordable
1966-1976 debt 10.3% GDP 8.68% getting unaffordable
1977-1992 debt 11.3% GDP 8.31% more so
1993-2007 debt 8.25% GDP 5.47%

population growths are important only for the question of per capita affordability

The Arthurian said...

Liminal, for me it would take all the "interesting" out of economics if I thought it was all demographics.

Clonal, this last group of numbers interests me. See you at 8 o'clock.