Tuesday 16 July 2013

Of bus crashes and open bank resolution

Matt and I seem to disagree a bit on how successful the RBNZ's Open-Banking Resolution policy might be in encouraging banks to avoid taking risks that might, on the downside, require a bailout to avoid systemic effects. [Note: substantive update below at *]

Recall that, under OBR, the bank's owners are liquidated first, then their unsecured creditors, before there's ever any move to touch the depositors.* I would expect that if depositors had to take a haircut, there would be reasonable pressure for a bailout. And so there is potential for that some downside risk is foisted on the government.

But we always have to think about things at the margin. Here's a parallel. Right now, if I pay too little attention while driving and smash into the side of a bus, killing me and hurting the people on the bus, my estate does not have to compensate those passengers for the harm I have caused them. A portion of the downside costs of my risk-taking driving maneuvers has been socialised. Does that mean that I take far too many risks while driving? Not in this case: the incentive to avoid dying in a horrible fiery car wreck is sufficient to ensure that I take appropriate care. Requiring that my estate provide compensation would be an inframarginal transfer.

If there were a lot of potential types of car accidents where I'd only be slightly injured but where I'd be doing a lot of harm to others, then the liability regime will matter a lot more. But if I'm guaranteed to die horribly in the event of any car wreck at all, I will take a lot of care. If I have a spike on my steering wheel, it doesn't matter how liable you make my estate for the damage I cause. The spike induces a whole lot of risk-avoidance. Maybe even too much.

The OBR is the spike on the steering wheel. The bondholders and the shareholders are killed in the case of a severe adverse event. That should be enough to induce due caution even if there are a lot of external parties who would suffer harm if the bank blew up. Some externalities are inframarginal.

Three conditions under which I'm wrong and Matt is right, though there could easily be more:
  • A bank under such pressure might try a bit of scaremongering to try to whet public appetite for a bailout that would protect the shareholders rather than do anything to affect the deposit-holders. If the government cannot successfully avoid that, and if the bankers know that, then all this is wrong. That's why it's important, I think, to remind everybody, and loudly, that OBR kills the shareholders first and that it's very unlikely that depositors would take any kind of substantial haircut.
  • Dysfunctional bank control structures. Suppose that the shareholders appoint a CEO whose compensation has a lot of upside variation with performance and a golden handshake in case of non-performance. And suppose further that the shareholders and bondholders aren't able to adequately monitor the riskiness of the bank's balance sheet. In that case, the one making the decisions isn't the one facing the boiling-in-oil contract and so offloads his downside risk onto both shareholders and bondholders and onto the public via the potential need for a bailout. But are shareholders really that stupid and bondholders that incapable of monitoring?
  • Bank cleverness in moving all the unsecured creditors into a preferred secured creditor arrangement so they're ahead of depositors in the queue, and somehow insulating the shareholders from OBR. I'm trusting that RBNZ can prevent this. If not, then the analysis above is wrong. 
Update: More conditions under which I'm wrong:
  • Brennan McDonald suggests (comments below) that some of the bigger shareholders might have sufficient political sway to get a shareholder bailout regardless of OBR. He could be right - it is very easy to imagine somebody like John Key looking at a statement from the NZ Superannuation Fund and reckoning that it's easier to bail out the bank than to prop up the Superfund afterwards and deal with the Kiwifund providers' lobbying. Brennan also worries that banks heading towards OCR might tunnel out all the good assets; I'd expect and hope that RBNZ would be keeping a sharp eye on such things. 
* Update 2: I'd outlined the RBNZ's OBR mechanism here. Depositors can take a small haircut fairly easily. Where I had thought that depositors had priority over other unsecured creditors, they are instead counted among the unsecured creditors. So the haircut would be larger than I had previously expected, and so too consequently would be the pressure for a bailout. While OBR attenuates bailout pressure overall by allowing banks to continue trading and by allowing depositors to maintain access to most of their deposits, the larger the expected haircut, the stronger the pressure. Now the OBR also includes provision for that small depositors could be exempted from the haircut, but it would take legislative action to give that effect. If it's done, small depositors pay less attention to their bank's security. If it isn't, bailout pressure come the crisis is stronger. Part of the difference between Matt and I could then be explained by my having mistakenly thought that unsecured bondholders were burned before rather than with the depositors. 


  1. It is good to point out that the shareholders will be wiped out - it is definitely a point that needs more of a public airing!

    But, as long as there are expectations of a bailout in the face of a "systemic" event - which the failure of one of the big 4 would be - we are in a world where depositors (and thereby banks) take on too much risk.

    Even if we introduce explicit deposit insurance and price it, banks will take on excessive risk - but this is preferable to implicit insurance that involves an implicit transfer from taxpayers to bondholders.

    I'd also note here that, in the case of deposit insurance shareholders will already be wiped out - but as Kareken and Wallace pointed out, this will still lead to a situation that involves excessive risk taking.

    If the OBR led to a situation where ex-ante expectations were that there would be no bailout, then we are all gravy - but this is not only time inconsistent in terms of the governments choice, it is also an argument everyone is avoiding trying to make!

    I think the core difference is our belief about whether the government will bail out the big banks - the same difference we've had since it came out - on everything else we agree.

  2. Let's refine this further: I don't disagree that the government would bail out the depositors if it came to that. Maybe 50/50 shot of a bailout for depositor haircuts up to 10%, near-certainty of it for haircuts higher than that.

    But I think that the incentives facing bondholders and shareholders could be sufficient to avoid that world's ever eventuating. If I expect that the government through ACC will bail out in the event of a car accident in which I die while imposing costs on others, I still take precautions against dying.

    The most likely source of failure, I expect, is where the bank's managers get a lot of the upside benefits but aren't subject to the downside burning and where shareholders are too diversified to care much about exercising control and where there aren't any big bondholders to do it instead. Then the spike is pointing at the passenger instead of the driver.

  3. "Then the spike is pointing at the passenger instead of the driver."
    1. There are big bondholders but also big counterparties - if one bank goes down, collateral values go down and OBR does not do anything about that.
    2. Shareholders include government entities - NZ Super, ACC, retirement schemes of all shapes and sizes, sovereign wealth funds - political noise for a bailout would be deafening IMO.
    3. The idea that all the IT, settlement and everyday transaction systems would change flawlessly overnight is laughable. See: any big IT project
    4. A bank headed towards getting put into statutory management under OBR would be known by the market months in advance - spiking CDS, higher collateral posting requirements, more covered bond issuance (getting last of residential loan book liquid), cost cutting/layoffs etc. - this implies the bank could flick off profitable units or find an overseas buyer far before the "fail point".
    5. I think OBR could have the opposite effect - increasing the size of the spike faced by the passenger. It's not like poor managers will face career consequences if they've already made their money...

  4. "But I think that the incentives facing bondholders and shareholders could be sufficient to avoid that world's ever eventuating"

    I'd say this is a pretty strong claim is all - and given that their is a disjoint between the risk depositors respond to (due to expectations) and the actual risk of lending on their behalf, this is understandable as an issue.

    Any difference in model we have has to come back to our view on the expectations formation of depositors as far as I can tell. I'm happy for this to be an open question - I am trying to leave my mind relative open about it :)

  5. I wonder how we could test it. I am uncertain - that's why I wanted you to get pseudo betting prices out of the folks at the meetings!