Tuesday, April 27, 2010

Steve Keen, Private Debt to GDP ratios, and the GFC

A little while ago I was asked my opinion about an article by Steve Keen, an Associate Professor at the University of Western Sydney and the founder of the Center for Economic Stability. Keen is apparently an expert in "the unconventional economic theory" called "The Financial Instability Hypothesis". This, in turn, was developed by the American economist Hyman Minsky, whose teaching have been making a resurgence of late. Keen apparently predicted a serious financial crisis as early as December 2005 – but, as the old joke goes, economists have predicted 7 of the last 3 recessions.

Though I'm not familiar with Minsky's work, I did read Keen's article and wasn't too impressed by his argumentation. Below is the original article, and my response.

Why the Debt to GDP Ratio matters
By Steve Keen


The main feature distinguishing my approach to economics from mainstream theory (known as "neoclassical" economics) is a focus upon the importance of credit in a dynamic, non-equilibrium framework. From this perspective I identify the ratio of debt to GDP — and the rate of change of that ratio — as the key determinant of the state of the economy, and this is what alerted me to the approaching Global Financial Crisis in late 2005. Debt to GDP levels in both Australia and the USA — the two countries for which I could get reliable data — were clearly on an exponential path that I knew had to break and cause a large economic breakdown.


This phenomenon was not taken seriously by conventional economists — which is why the crisis took them by surprise — and they continue to misunderstand why it matters, even as the crisis that they did not see coming continues. Some conventional economists have criticized me for focusing on what they call a "crude" private debt to GDP ratio; others allege that I am making a "schoolboy error" by comparing a stock to a flow. In response, I have developed a simple numerical example that hopefully illustrates why the debt to GDP ratio matters.
  • Imagine a country with a nominal GDP of $1,000 billion, which is growing at 10% per annum (real output is growing at 5% p.a. and inflation is 5% p.a.);
  • It also has an aggregate private debt level of $1,250 billion which is growing at 20% p.a., so that private debt increases by $250 billion that year;
  • Ignoring for the moment the contribution from government deficit spending, total spending in that economy for that year — on all markets, including assets as well as commodities — is therefore $1,250 billion. 80% of this is financed by incomes (GDP) and 20% is financed by increased debt;
  • One year later, the GDP has grown by 10% to $1,100 billion;
  • Now imagine that debt stabilises at $1,500 billion, so that the change in debt that year is zero;
  • Then total spending in the economy is $1,100 billion, consisting of $1.1 trillion of income-financed spending and no debt-financed spending;
  • This is $150 billion less than the previous year;
  • Stabilisation of debt levels thus causes a 12% fall in nominal aggregate demand.
What about if debt doesn't actually stabilise in the second year, but instead grows at the same rate as GDP? Then we get the following situation:
  • In the first year, total demand is $1,250 billion, consisting of $1,000 billion in income and $250 billion in increased debt;
  • In the second year, total demand is also $1,250 billion, consisting of $1,100 billion in income and $150 billion in increased debt;
  • Nominal aggregate demand is therefore constant;
  • But after inflation, real aggregate demand will have contracted by 5%.
This is the real danger posed by debt: once debt becomes a significant fraction of GDP, and its growth rate substantially exceeds that of GDP, the economy will suffer a recession even if the debt to GDP ratio merely stabilizes. This approach not only explains why the GFC was inevitable, but also gives a very different explanation to preceding crises, and it explains why, to date, Australia has successfully avoided the worst of the GFC while America has suffered from it very badly.
The next chart shows the annual change in the private debt to GDP ratio in Australia, and the changes in government in the last 65 years (blue for a conservative Liberal party victory, black for a progressive ALP victory, and green for one election that the ALP won against the odds). Note that between 1945 and 1965, the ratio rose as often as fell. Then for a decade the debt ratio only rose, before then falling very sharply. Its collapse from a peak rate of growth of 7% p.a. in mid-1973 to minus 3% in mid 1975 was the real caused the mid-1970s recession.

The same pattern repeats from late 1988 till early 1993 — from a growth rate of 7% to falling at 3.5% p.a. in early 1993 — and this gave Australia what its then Prime Minister Paul Keating called "the recession we had to have".

Now to our current crisis, which has been more severe than preceding crises, not because debt was rising more quickly (in fact the rate of growth of debt in the 1970s was much higher than now), but because the debt to GDP ratio is so much higher now. The growth rate peaked at 11.5% p.a. in early 2008, only to fall to minus 5% in mid-2009.


But notice how quickly that decline in Australia's debt ratio has turned into growth once more: whereas Australia experienced 33 months of falling debt to GDP in the 1990s recession, it had only 13 months of a falling debt ratio this time.
The overwhelming reason for this rapid turnaround was the First Home Owners Boost (which I prefer to call the First Home Vendors Boost). This government policy enticed 250,000 Australians into their first mortgage by the offer of an additional A$7,000-14,000 government grant. These new buyers levered this higher deposit into a far higher loan, driving up prices which then provided the vendors from whom they purchased with an additional $30-50,000.
These vendors in turn used this windfall as additional deposits in their own levered purchases. Mortgage debt, which was on track to fall 3% as a proportion of GDP prior to The Boost, instead increased by 6% of GDP — a 9% turnaround that was equivalent to an additional $100 billion stimulus to Australia's A$1.25 trillion economy.
This dramatic turnaround in mortgage debt more than countered deleveraging by the business sector, so that private debt levels rose. On top of that, the government's own debt-financed spending returned the debt contribution to demand to the average of 1996-2006, when the debt ratio grew at roughly 5% p.a. and financed an apparent boom.


So Australia has avoided the GFC by recreating the conditions that led to it. Needless to say, this is not entirely a good thing. We are potentially avoiding pain now by setting ourselves up for greater pain in the near future.
America, on the other hand, is clearly experiencing the worst of the GFC because it is deleveraging — reducing its debt to GDP ratio. Though this is painful, it at least addresses the cause of the crisis — too much private debt — by reducing it rather than increasing it once more.
Since America is even more of a Ponzi-dominated economy than Australia, this reduction in private debt — now that the Ponzi Scheme has collapsed — is the key cause of the USA's current slump. This next chart shows the relationship between the proportion of aggregate demand which is debt financed — which I define as the change in debt, divided by the sum of GDP plus the change in debt — and unemployment. The change in private sector debt has gone from boosting aggregate demand by over 20% to reducing it by almost 15%. Only a massive increase in government debt has attenuated the deleveraging.


Whether government spending can continue to counter private sector deleveraging is now a moot point. Some of my non-orthodox economics colleagues argue that a sufficiently large government deficit can return a country to economic prosperity. While I doubt this claim, it is undeniable that if the "deficit hawks" make the government reduce its own spending, then there will be no force standing in the way of accelerated private sector deleveraging.



What this Steve Keen bloke says has several problems:

In his example, the country takes up $250b of new debt in t0, while GDP "only" grows by $100b from t0 to t1. Now, in standard neoclassical theory, debt is used to invest, i.e., increase the productive capacity of a company/country. An investment of $250b for any return less than $250b is just not profitable. So, assuming the investors are rational and the investment thus yields a income of $250b + r + X in t1, then they can pay back the original debt, the interest r, and the GDP is better off by X.

"Now hang on", you might say, "this means there's $250b of (domestic) spending/demand in t0 that isn't there in t1, because that $250b is going externally to repay the debt. So domestic spending has fallen by $250b" Correct. However, this is where Keen's second trick comes in: he's used an example with a high existing baseline of debt, so that removing that debt stimulus drops the GDP considerably. HOWEVER, neoclassically, the debt itself should not be seen as the usual state of affairs: rather, t0 should be seen as a year with unusually high GDP.

The counterargument to this point is that this level of debt has crept up gradually. Well, okay: then lets reduce it gradually. Given that the investments are returning more than their face value, we should still be left with a greater growth rate of GDP than without the borrowing/investing strategy.

Let's now check my "investment for increased production" assumption: my (neoclassical) analysis has assumed borrowing by investors – what I'm going to call "good debt"; Keen's analysis has just talked about "private" debt. This could be "private" in the standard economic sense of "non-public", i.e. non-governmental, or it could be non-government and non-business, i.e. households. Households doesn't make a lot of sense, a) because it's a strange use of the term, and b) because the figures he uses in his example would be way out for just household debt. So I'm going to assume it's the standard economic non-government "private".

Okay, so, private-sector borrowing is not always used to finance investment: most household borrowing is, in fact, to finance consumption. Borrowing to consume now really means you're borrowing against future consumption (more now, less later), so excess levels of borrowing could well lead to a situation in which, to repay their debt, households have to reduce their future consumption to undesirable levels (perhaps to the extent that they declare bankruptcy).

This, while a (still existing) problem, is not the cause of the GFC. Most household debt is long-term (mortgages) rather than short-term (credit cards). The GFC was caused by problems with liquidity, mainly refinancing short-term credit by large companies & banks.
"Aha!", you might think, "Mortgages! Wasn't the whole GFC started by sub-prime mortgages going down the drain?"
Sure – but the problem with these mortgages wasn't that it was private debt, but that it was crap private debt – it wasn't being used to create any net value. Usually you buy a house with a mortgage, but the house values faster than the interest rate, so you come out in front. Sometimes you don't, but then the negative balance comes out of your consumption. (Owner/occupiers don't even realise the difference, since they don't sell their house). Now, these sub-prime mortgages are called that because they are crap - the houses didn't rise in value, and the borrowers couldn't finance their net losses by reducing their consumption, so the banks/businesses/whatever had to shoulder the burden. Lender confidence/risk preference falls, and we have liquidity problems.

The important thing is thus the character of the debt: whether it's good debt, like I described above, or crap debt. Just measuring the level of all private debt and pretending it's bad is a gross simplification – a criticism Keen appears to be aware of but does nothing to correct.

Since Keen's link between levels of private debt to GDP and the GFC were so tenuous I didn't bother reading any further. There still certainly exists the problem of servicing and, in future, reducing our levels of private debt, but, to repeat, this is a separate issue to the GFC.

Qualification: Keen's analysis is certainly more complex than he has laid it out here, and it's possible he really has discovered a relationship between private debt ratios and liquidity. If so, he's done a very good job of hiding it.

4 comments:

Tobias said...

Also: Fitting trends to time series and then reporting an R^2 is often a decent indicator of the quality of the rest of the argument.

Martin said...

Indeed!

For the lay reader: in the first figure, private debt (generally) rises as time passes, but the passage of time does not cause private debt to rise, any more than the passage of time causes people to get fatter. Both developments depend on other factors.

Fitting a trend to the figures and saying "my equation explains 99% of what you see here" is a cheap trick, especially given the equation is incredibly simple (the real world is complex), the dependent variable (time) doesn't cause the change (see above), and having an R^2 of 0.99x can usually be assumed to be bollocks in the social sciences anyway. Poor econometrics.

Martin said...

Oh, what is, in fact, ever more awesome is the way Keen regresses to get this exponential path and then says "I knew [it] had to break."
Brilliant!

Martin said...

A final note on Steve Keen: http://www.smh.com.au/business/economic-doomsayer-reaches-physical-peak-20100423-tj2m.html