A mauling Minsky moment: comment on Martin Wolf

Martin Wolf’s call to arms is a stirring read, and I agree with much of its conclusions on the need for simple and very much tighter regulations on bank leverage.  However, I don’t agree with how they are reached.  Somewhat caricatured, the logic is:  1) as Minsky said, stability and financial systems breeds instability, which explains why our global financial system collapsed.  2)  It follows that we need drastically tighter regulations on bank leverage.

Minsky’s reading of economic systems was a stroke of creative, imaginative but informal genius.  But it’s highly conjectural.  It may be part of the story of the financial collapse that financial system stability is inherently destabilising, or it may not.  Just because intermediation was cheap and plentiful before it became expensive and scarce does not validate Minsky.  There are plenty of other preceding and coincident events which don’t fit.  To go through a few examples:  Technological advancement in intermediation and financial transactions, accompanied by deregulation:  innovation (the destabiliser) did not advance because over some prior (stable) period it hadn’t. (The regulatory story fits, which I’ll return to later).  Calomiris-like tales of political interference in the financial sector:  banks didn’t lobby Governments (destabilising) because previously they hadn’t.  According to that story, they always have and always will.  The export of savings ‘uphill’ from emerging Asia, (destabilising) looking and hoping for safe haven:  this happened because barriers to that capital export were lifted, not because for no other reason capital had been invested locally.  None of these competing narratives of the crisis have anything to do with Minsky’s hypothesis that stability breeds instability. The simplest model of bank runs, which many use as a metaphor for the closure of wholesale funding markets, Diamong and Dybvig, simply says:  runs can happen on healthy banks which borrow short to lend long.  It doesn’t say ‘runs happen because they didn’t happen previously’.

Besides, as a theoretical conjecture, it needs working through.  Before we accept Minsky’s verbal conjecture, what would an artificial economy do if you peopled it with agents with imperfect apprehensions of risk?  Would such artificial systems be characterised by a ‘stability is destabilising’ dynamic?  There are lots of counter-examples in models of less-than-rational expectations.  For example, heuristic models of inflation-forecasting, a field I know more about, often have multiple rest points.  These rest points are not necessarily inherently destabilising.  If shocks are small, they attract the system back.  If shocks are large, the economy might be sent off to another attracting rest point.  There is nothing inevitably cyclical about the volatility of systems like these.  If you hit these systems with shocks so that every now and then a large one arrives, then you will see repeated movements between the rest points.  But you would not describe such models as exhibiting a stability-is-destabilising dynamic.  You would describe them as exhibiting a ‘large shocks cause the economy to move from one rest point to another’ dynamic.  Models of learning would go by the opposite description.   In such models, stability is stabilising;  instability is destabilising.  What causes agents’ forecasting models and the system at large to explode is large shocks:  instability.  There are probably hundreds and hundreds of mathematical systems with potential lessons for economics that don’t fit with Minsky’s hypothesis.   If you follow Paul Krugman’s reasoning, the financial crisis, at least as manifest in spreads, is over and done with.  And it’s lasting consequences for demand and inflation are down to insufficiently stabilising fiscal policy.  There’s no room for a ‘stability is destabilising’ dynamic here.  The future would be bright if Governments simply spent more money raising the natural rate.

But suppose we accept the idea at face value.  One reading of it suggests that we restricting bank leverage as Martin Wolf urges won’t work by itself.  A long period of stability breeding complacency would presumably erode the support for these restrictions.  In a democracy like the UK’s, where by convention no Parliament can bind any of its successors, there would be nothing to stop this complacency leading to future regulatory relaxation.  Is this an argument for something that does bind future Parliaments?  It would be, if we found solid theoretical and empirical support for the stability breeds instability idea as the dominant cause of the crisis.  But we haven’t yet.