Standards, gentlemen, standards February 4, 2013 at 6:29 am

I want to distinguish two issues, as a prelude to another post on central capital calculations.

First, there’s behavior. If everyone, or many folk, do the same thing, then you get herding and the possibility of phase changes. In financial stability terms that’s bad. You want people to do different things, as a diverse ecology of financial institutions is more robust than a mono-culture.

Second, there are standards, like regulatory capital. You want them to be simple and uniformly implemented, because that creates investor trust. There is a massive information asymmetry between bank managers and bank investors, and regulatory capital ratios help to level the playing field. It doesn’t matter if they don’t measure risk very well because you care much more about false negatives (well capitalised bank is not solvent) than false positives (solvent bank does not pass regulatory standard).

Thus, recent objections to my proposal to centralise model-based capital calculations are wrong because they confuse the first issue with the second. SWP says:

We want to diversify model risk, not concentrate it. Centralized calculation concentrates it. I actually see an evolutionary benefit in the wide range of model risk weights that DEM bewails. That represents a diverse ecosystem of entities with divergent views that is less vulnerable to a single shock. Yeah, that shock will crater some banks, but not all of them. When I think of any-ANY-centralized calculation, I think of the Socialist Calculation Debate. I also think of monocultures that are dangerously vulnerable-systemically vulnerable-to a single shock. Think of the devastation that smallpox wreaked on native American populations. A single shock can crater everybody all at once.

That would be true if all banks did the same thing, but there is no requirement, or even strong incentive, for that. All I am asking is that they all, in their diverse ways, meet a simple standard: that centrally calculated market risk RWAs are less than the amount of capital they set aside for market risk.

Will this central calculation be wrong? Yes, clearly so. Will it be differently wrong for different banks? No, it will be wrong in the same way for everyone. But does that introduce an incentive for all banks to behave the same way? No, it doesn’t. This is partly because many banks don’t care about market risk RWAs (think Wells Fargo or Lloyds) and partly because even the ones that do don’t base their behaviour simply on what is cheapest in capital terms.

In short, letting banks use their own models for market risk RWA calculation reduces investor trust without adding much if any robustness. Root them out, cut costs*, and set a standard that is believable and whose information content is easily understood.

*Which are enormous.

6 Responses to “Standards, gentlemen, standards”

  1. Variation in mRWAs is not the same thing as variation in VaRs, which are only one input to mRWAs.

    The headline “order of magnitude” stuff that got Matt so excited refers to IRC; forget it. The BCBS deserves what it gets from that ill-conceived and ill-defined monstrosity. Much of the variation in mRWAs was driven by regulatory idiosyncrasies, and not by VaR at all. When you get down to VaR, much of the variation is driven by the choice of historical period used for calibration, just as Ginger Yellow said in you previous thread. The remainder is mostly due to the scaling approach used to get 10-day horizons and the way in which general and specific risk is aggregated – all this is summarized clearly in figure 7.

    These issues just don’t seem that hard to fix. Particularly if the Committee wants to avoid pro-cyclical capital charges, it has to mandate the calibration period. So regulatory VaR measures that are comparable across banks are not that far away. But the Committee has to give up its cherished “use principle”; banks generally *want* the VaR they use operationally to be sensitive to current conditions. You need one VaR for capital, and another for trading.

  2. In the original concept, the mRWAs were not designed to be “point-in-time” equivalent, but rather to be equivalent “on-average, over-time”. This would allow for some variation based on observation periods used to calibrate the mRWA.

    The last two sentances of Mr. Koop’s comment get to the central point – capitalization is different from day-to-day risk management, and calibration needs to be different. I may want a short-term exponentially weighted moving average for risk management that is very sensitive to current conditions, but this would be inappropriate for longer-term capitalization.

  3. […] Ex Macchiato has responded in a way that surprised (if no one else) […]

  4. Been a while since I commented on here. Sorry DEM for my ab

    Linking risk weight debates and TBTF, if you could make banks that are easily resolved with no spillover (the ultimate goal of quadratic taxes and the like) then who cares about how risk weights are calculated? Banks that get their model wrong and suffer a capital wipe out can fail and be removed from the system. Caveat emptor and the market makes a mental note that the failed firm’s models should not be an approach used at other firms.

    I admit that in the limit, this kinda removes the need for any regulatory capital and prudential regulation (you’d still need conduct regulation) but hey, think of the resource savings.

  5. PK – I agree completely with you. The use test doesn’t pass muster any more and dropping that would allow us to recognise the reality that reg var is nowhere near management var. It would also allow those firms who don’t really believe in var to use their favoured alternative, expected shortfall say, instead, without it having regulatory consequences. Reg var could become less procyclical, for instance by crossing out the var term in var + stressed var + IRC and going to stressed VAR + IRC alone.

  6. QM – Ha, interesting. The fragility literature would suggest that your proposal is sound, and I think it is, with the caveat that ‘if you could make banks that are easily resolved with no spillover’ is really hard, and capital regulation is a good risk mitigant in the meantime, or if you are unsure if you have succeeded in ending TBTF. Disclosure also has a big place I would suggest in that this can perhaps let the market give a bank a signal that is going down the wrong path before failure. The experience of the SEC’s CSE program does however suggest that three out of five broker/dealers can get their model wrong more or less simultaneously, so one needs to be conscious of that when assessing resolution plans.