The ECB's paper makes a valiant attempt to fit the financial system into a DSGE model of the economy that previously ignored it. The maths is fearsome and I admit I skipped much of it. But they draw some important conclusions.
Very early in the paper they accept the prevailing wisdom that financial crises are endogenously determined - in other words, they happen as a consequence of behaviour within the system, not because of external shocks to it. Now this immediately causes a problem with the model. DSGE models work on the basis that shocks are exogenous - sort of like meteorite impacts on life on earth. But using the same analogy, an endogenous crisis would be CAUSED by the behaviour of life on earth. Attempting to use a DSGE model to explain the effects of the behaviour of humans on their own behaviour is enough to drive a serious economist to drink. Admittedly there are shocks in the model, but adverse ones are regarded as secondary and the causative positive supply-side shocks they postulate happen some time before the crisis itself. They cause the buildup of the behaviour that leads to the crisis, rather than the crisis itself. This I think is an important insight, and I commend the ECB economists for sticking to their guns despite the considerable difficulty in using a model that on the face of it looks inappropriate.
They outline a typical run of events leading to a financial crisis as follows.
- A sequence of favorable, non permanent, supply shocks hits the economy. They don't say what these are, but candidates might be reductions in key interest rates or falls in oil and raw material prices.
- The resulting increase in the productivity of capital leads to a demand-driven expansion of credit that pushes the corporate loan rate above steady state.
- As productivity goes back to trend, firms reduce their demand for credit, whereas households continue to accumulate assets, thus feeding the supply of credit by banks (presumably to households).
- The credit boom then turns supply-driven and the corporate loan rate goes down, falling below steady state.
- By giving banks incentives to take more risks or misbehave, too low a corporate loan rate contributes to eroding trust within the banking sector precisely at a time when banks increase in size. The credit boom lowers the resilience of the banking sector to shocks, making systemic crises more likely.
- When counterparty fears in the interbank market rise too high, the market freezes.
1) rising corporate debt to start with, then declining
2) rising productivity
3) household assets rising as corporate credit falls
4) increasing size of bank balance sheets
Did we actually see this in the run-up to the last financial crisis? Well, maybe. The UK's ONS gives a picture of rising corporate debt from 2000-2007, then a fall*:
(I've used liabilities instead of debt in this case because of the ONS's observation that PNFC balance sheets were actually more highly geared than acquisition of debt and assets during the period suggest. The whole ONS paper is well worth a read).
Household assets did indeed rise during this period. But so did their debt:
So household balance sheets were inflating overall. This doesn't necessarily invalidate the authors' conclusions, but the omission of household debt from their analysis I think weakens it. And perhaps more importantly, there isn't any evidence from this pair of charts that households continued to accumulate assets after corporate borrowing started to fall. Both household assets and corporate borrowing seem to have fallen at the same time.
However, the authors are right about the increasing size and riskiness of bank balance sheets. I don't need to produce a chart for this, since the accumulation of both household and corporate liabilities in a bank-dominant lending model such as exists throughout Europe must result in inflated bank balance sheets. And the increasing riskiness of bank lending and trading in the run up to the financial crisis is firmly established.
The other factor is productivity. Productivity did indeed rise hugely prior to the financial crisis:
(larger version here)
The shock to productivity caused by the financial crisis is very evident in this chart. But more importantly, it suggests that the productivity rise prior to the crisis was unsustainable. In which case the UK economy cannot realistically return to that rate of productivity growth: a shallower curve is both more likely and more desirable. The question is whether it can bounce back to the level it reached in 2008 and then grow more sustainably. I'm not convinced. It seems more likely that productivity growth has dropped to a new, lower (and shallower) trend. This has serious implications for output and for economic recovery.
But I digress. The ECB team suggests that declining corporate borrowing and a fall in productivity lead to unsustainable (bubble) growth of household and bank balance sheets, causing increasing risk which ultimately ends in a market freeze when counterparties think the risk is too great. I don't buy this. Neither corporate borrowing nor productivity started to fall until the subprime crisis in the US, by which time household and bank balance sheets were already over-inflated and high risk. It is difficult to establish any likely causative relationship between productivity fall and increased riskiness, except in the early stages of the crisis itself. I wonder what exactly the ECB team regard as the crisis: perhaps they are referring only to the fall of Lehman and subsequent meltdown. If so, to my mind they are only looking at half the event. Lehman happened towards the end of the financial crisis, not at the beginning.
However, the remainder of their analysis is excellent. Their DSGE modellling establishes the following principles.
- Financial crises are associated with unusually high and rising productivity.
"One striking result that emerges from this experiment is that the typical banking crisis is preceded by a long period during which total factor productivity is above its mean. In some 20% of the cases, crises even occur at a time when productivity is still above mean. This reveals one important and interesting aspect of the model: the seeds of the crisis lie in productivity being above average for an unusually long time. The reason is that a long period of high productivity gives the household enough time to accumulate assets beyond the banking sector's absorption capacity...."
So a highly productive household will save more. But Broadbent's chart shows that households are accumulating both assets AND liabilities. Assets, after all, include houses, which most people buy with debt (mortgages). And there is no reason to assume that this behaviour is homogenous, either. The bifurcation of the labour market that has been going on for well over a decade now suggests that high earners acquire net financial assets, while lower earners acquire net financial liabilities (debt). Households' balance sheets are polarised between high-debt and high-savings, with a substantial proportion of households having both - with savings mostly in illiquid form. This places their balance sheets under pressure when income falls due to productivity collapse. I think the authors miss this.
- Sustained periods of above-average productivity feed credit booms, creating large financial imbalances that are ultimately unstable and cause crises to break out without exogenous shock.
"We measure financial imbalances by the distance between banks' absorption capacity and banks' core liabilities. The smaller this distance, the larger the imbalances, and the less resilient to adverse shocks the banking sector. Typically, a distance of less than, say, 5% reflects large fi nancial imbalances..... 70% of systemic banking crises break out in the year after fi nancial imbalances have reached this threshold. This result con firms that most crises in our model break out endogenously, without an adverse exogenous shock happening at the same time."
Or putting it another way, high gearing in banks increases their risk. Well, who'd have thought it.
- High levels of household savings in good times are destabilising.
....in our model, crises are more likely to occur in good, rather than in bad, times. This is due to the asymmetric e ffects of permanent income mechanisms on fi nancial stability over the business cycle. Bad times in the model are typically times where productivity is low and the household dis-saves....the fall in savings makes (all things being equal) crises less likely. Hence, in bad times, the dynamics of savings tend to stabilize the financial sector. In good times, in contrast, productivity is high and the household accumulates assets, which....makes crises more likely. In this case the dynamics of savings tends to destabilize the financial sector. This asymmetric e ffect of savings is the basic reason why credit-boom led crises are so prevalent in our model.
So their argument is essentially that high levels of household savings (presumably including physical assets such as houses, which are used as collateral against household borrowing) increase the riskiness of the financial system. This is particularly interesting, because it is the opposite of what is normally argued. But household savings are the liabilities of the banking system, and secured lending depends on good collateral, so a high level of household savings increases banks' fragility when also associated with low levels of loss-absorbing equity - the second bullet point above.They go on to draw the following important conclusions from their model:
1) Risk averse economies are more prone to crises.
T|his is really important. Prudential saving by households fearing a crash actually makes the crash more likely, and makes its effects worse. The quest for safety perversely creates the very disaster that the savers fear. Moneyweek readers, please take note.
2) Economies with a highly elastic labour supply are more prone to crises.
This also is really important, especially given the prevalent belief that flexibility in the labour force is a good thing and countries should undertake labour market reforms to increase flexibility. This research suggests that flexibility in the labour force increases the amplitude of the boom-bust cycle, as households work more in good times and less in bad times. That suggests that there is a trade-off between business efficiency and economic instability. Some loss of business efficiency due to labour market inelasticity might be a good thing.
3) Contract enforceability and efficient banks reduce the likelihood of crises. (Though if crises do occur they can be worse). If the bulkheads on the Titanic had been higher the ship wouldn't have sunk.
4) Higher uncertainty is conducive to crises.
This bit is really interesting. They suggest that technological shock influences the likelihood of crisis: adverse technology shocks increase it, because households save prudentially and corporate borrowing falls, making the financial sector more fragile. Positive technology shocks, on the other hand, decrease the likelihood of crisis. The question in my mind is how to distinguish positive and negative technology shocks: "robots ate my job" is a negative shock to households but potentially a positive one for firms if their return on capital improves as a result. The authors don't develop this enough and there is room for lots more research on this subject. Intuitively it seems likely, if the primary cause of financial instability is high levels of household saving reducing banks' loss absorption capacity, that general uncertainty about the future would make crises more likely unless government applied pressure to banks to maintain or increase loss-absorbency.
And that I think is where we are now. We are in a period of uncertainty about the future and fear of technological change. Our economies are highly risk-averse, as is evident from the crashing yields on "safe haven" assets such as high-quality government debt and currencies. Elasticity is increasing in the labour force as self-employment, temporary and casual working increases and permanent full-time employment falls. And corporates, households and government alike are paying down debt and trying to save. The only ingredients missing are bank lending and productivity growth. Sadly the authors don't model the present stagnation as a repressed financial crisis.
Booms and systemic banking crises - Boissay, Collard & Smets (ECB)
Corporate sector balance sheets and crisis transmission - ONS
Decoupling of wage growth and productivity - Resolution Foundation
Bifurcation in the labour market - Coppola Comment
The deadly quest for safety - Coppola Comment
The end of Britain? - Coppola Comment
Perverse incentives and productivity - Coppola Comment
The self-employed will not save the economy - Flip Chart Fairy Tales
Productivity and the allocation of resources - Broadbent (Bank of England)
Deleveraging - Broadbent (Bank of England)
The ECB researchers' analysis amounts to a flawed re-engineering of Minsky's financial instability hypothesis (which they briefly mention at the beginning of the paper) and Goodwin's growth cycle model (which they don't mention at all). I've therefore added a link here to Steve Keen's paper on the same subject. He uses chaotic system dynamics rather than a DSGE model.
Finance and economic breakdown: modelling Minsky's "financial instability" hypothesis - Keen
*I'm using UK data throughout this post because the paper is written from a European perspective and the UK is the principal financial centre in Europe. US lending markets are very different from European ones because of the dominance of capital markets. Europeans, including the UK, rely much more on banks.