Read The Use of Market Price Information in Financial Supervision: Framing the key questions.
Please post your comments by clicking the Reply link.
Claudio lays out the premises of his macro-pru thinking clearly, a view of endogenous systemic risk based on his work as well as on the work of the SRC and the FMG, including the idea of pecuniary externalities whereby prices (perhaps through mark-to-marking and risk, collateral, regulatory etc. constraints) lead to imperatives that can reinforce the price movements and create positive feedback loops and domino effects. This is now a view commonly held, but like Borio, I remember clearly how such a view was not accepted before the GFC, and even for a number of years thereafter.
Borio argues now convincingly that price information - cross-sectional this time - can also be used for micro-prudential and supervisory purposes. He mentions in particular the price to book ratios of banks and how a low P/B reflects concerns about asset valuations and above all longer term profitability, as in the case of Credit Suisse where liquidity and solvency measures looked OK.
Maybe this micro-pru measure is especially powerful when other banks do not have low P/B ratios. If all banks have low P/B ratios, this may be because the banking sector as a whole is undervalued, such as in the depth of a crisis, and the informational content for one bank is low, and the other metrics like solvency and liquidity may also signal the same anyway.
I would like to suggest now that P/B ratios might be useful also for systemic risk and for macro-prudential authorities as an ingredient in measuring the built-up of risk. In that sense, P/B ratios as macro-pru indicators may well be countercyclical. Prior to the GFC P/B ratios of banks were way above 1 globally and only tanked to lows well below 1 in the aftermath of the GFC and the Greek crisis. So the micropru measure and the macropru measure are opposite, as so often. Perhaps high overall bank P/B ratios signal forward looking risk as in such high valuation periods banks may feel invulnerable, bank share prices are high perhaps because of unsustainably high expected profits due to regulatory arbitrage or excessively large risk positions due to low cost of capital, low vol, high leverage etc.
Furthermore, and in support of the view of a useful cross-section, P/B as a macro-pru indicator may be even more informative if its cross-section is considered in addition to the overall level. Is it impossible that high overall P/B ratios are signalling an even larger risk when dispersion is low than if dispersion is high, for instance because large dispersion may indicate that not all banks have the same portfolios and therefore that coordinated selling rounds in a crisis may be less pronounced?
As usual, Claudio’s views deserve to be carefully listened at. I enjoyed his line of reasoning and found it enlighting, but it also set in motion some additional thoughts that I would like to share.
Regarding how market prices evolve over time, I guess one has to keep in mind that they depend on two different variables: the amount of risk in the market and the unit price of risk (which in turn reflects the investors’ risk appetite). If the latter factor can be accounted for, then the former can be worked out from market prices in a more reliable (or less unreliable) way. Sure, this is hard to do, but not impossible. On the corporate bond market, e.g., the shift to and from covenant-light issues is a reliable measure of how eager investors are to take on risk. As far as equities are concerned, Gordon’s model provides a simple way to work out the “long term growth rate” in earnings, g, that makes certain valuations possible. When g skyrockets, as in the US now, this is another sign that risk appetite is going up, rather than risks going down.
My second (and final) thought concerns the fact that risk mispricing may also follow - rather than from jumps in the market’s overall risk appetite - from the fact that some specific risks are poorly measured and therefore are more vulnerable to biases and bubbles. Just think of correlation pricing ahead of the GFC: rating agencies were generously paid to assess it, but their models were young and mostly fed with recent data, a recipe for disaster. My point here is that, if risk mispricing over time does not apply across the board, but rather in a selective way, then it may prove much harder than expected to use market data even to carry out cross-section valuations. Let’s say we are currently underestimating climate-related risks: some banks are likely to be more deeply affected by such bias (should I mention large US lenders leaving the Net Zero Banking Alliance?), so our idea of where vulnerabilities lie in the system could also be misguided.