those black boxes

plus a bonus! predictions from the past

I got distracted for the last couple of weeks from returning to the issue of “muddy boxes” and the question of when it is or isn’t a good idea to try to open up a black box.  This is another area, like that of “POSIWID” where it’s easy to get confused while reading Stafford Beer, because he presents a cool and catchy slogan which looks like it is an axiom of underlying reality but is actually part of the instructions for applying his system.

In this case, the slogan is “it is not necessary to enter the black box to understand the nature of the function it performs”.  It looks like it’s saying “don’t open black boxes” or “you won’t gain anything by opening black boxes”, but it’s not.  Just as “The Purpose Of A System Is What It Does” is really meant to answer the question “how do I know what kind of information flows I need to consider when drawing my diagrams?”, the black box principle is really just meant to reassure you that it’s OK to draw black boxes at all.

To put it another way – in fact, you can almost always open a black box, either to gain full knowledge of its internal workings, or to split it up into subsystems.  In fact, Beer’s own Viable Systems Model effectively tells you to do this, by saying that you should always be thinking of three systems at once, the “system-in-focus”, its main operating subsystems and the higher-level system of which it itself is a subsystem.  The question to ask about opening black boxes is, “is this a valuable use of my time?”.

And the answer Beer is trying to give with the grandiose slogan is “not necessarily”.  As long as the box itself makes sense and you can reasonably reliably map its inputs to its outputs, then that’s good enough.  Another way of thinking about this might be to reverse the principle – if you can gain some important insight from opening up a black box, then maybe you hadn’t drawn enough black boxes in the first place.

If you understand the function that the black box performs, then further insight into its internal structure is likely to be “nuance”, in the most pejorative sense. 

Photolithography machines and jet turbines are very complicated, but it’s important to understand every detail because otherwise the thing won’t work.  (It is a source of concern to many that the latest generation of ASML machines have reached a level of complexity such that there is not only no single individual, but no readily identifiable small group of individuals, who know enough about their workings to reconstruct one if they were all destroyed tomorrow).  In this kind of context, “it’s more complicated than that” can be a useful thing to say.

In a lot of social, political and economic contexts, though, we’re dealing with things an order of magnitude more complicated than that, made up of millions of human beings, each of which carries around a neural network substantially bigger than anything OpenAI has ever considered building.  In that context, “it’s more complicated than that” is just an irritating truism – in order to justify a deeper level of analysis, you need to demonstrate why it would actually be valuable to do so.  You can only do that by showing that we don’t currently understand the function of a particular subsystem well enough, and so it is necessary to enter the black box.

--

Anyway, enough of that. While searching my email outbox for something else, I came across this prediction for “the next 20 years of financial regulation”.  It’s quite long, so don’t bother if you’re not interested.  It’s by no means particularly impressive – in particular, I wildly underestimated how long things would take.  But there are a few hits in there, and it’s interesting to me at least to see how this system looked from the vantage point of two years after the fall of Lehman Brothers and shortly before the Eurocrisis.

Financial Regulation - the next 20 years

I wrote this for an unusual client who wanted to take a twenty year view, and it was a rather refreshing exercise, my usual time horizon being "how soon can we get this out the door". .  Note that this is my long-term view of the global regulatory and political forces at work - it’s not necessarily the view I take when doing equity analysis and looking for the version of the world most relevant for current stock prices.  Obviously and necessarily it gets more speculative (and thus my comments get shorter) as we go further out in time:

2010 - 2015+ - The Long Implementation of Basel 3

The official timetable has set out the Basel Supervisors’ Committee’s intention to have a new global regulatory framework agreed by the end of this year, and implemented into national law by the end of 2012. However, this is a very ambitious timetable - Basel 2 published its first consultative document in 2001, did not achieve an agreed set of rules until 2005, and has still not been implemented worldwide.  The German BaFin has already commented that 2015 is a more realistic target implementation date.

The issue here is that Germany is likely to drive the process, as it is the strongest regulator with the weakest banks - no credible international agreement can be reached without Germany, and the German regulator cannot agree to any regulatory standard which would leave the Landesbanks requiring recapitalisation (or at least, it cannot under current German administrative law - the Länder do not have the resources to recapitalize them and the Federal government does not have the political ability to get involved).  However, the USA would not tolerate a regulatory regime which would allow the Landesbanks to avoid recapitalisation[1].  The solution to this impasse is delay - every year that passes, particularly in a steep yield curve environment, reduces the capital deficit and holds out the possibility of a change in the legal structure of German banking which would make recapitalisation more viable.

It is the easiest thing in the world to introduce delay into the Basel process, because the Supervisors’ Committee has to operate by consensus - it has no powers of its own, there is no international treaty or other legal basis for it and it matters only because it is a club of powerful regulatory organisations.  Every successive Quantitative Impact Study, for example, delays the process by about six months.  Even if the BSC manages to agree a set of rules by 2010, these have to be translated into European law by a directive (the Capital Requirements Directive IV, currently at its first consultation stage), and this directive needs to be implemented by Member States, providing further opportunities for delay.  In my opinion, 2015 is the absolute earliest realistic timetable for implementation.

So for the next five years, the sector is likely to operate in an environment of extreme regulatory uncertainty.  This will have a number of effects:

Higher than equilibrium capital ratios for big banks.  An implication of this analysis is that the regulatory process is being driven by the needs of (mostly) smaller, domestic and politically important banks rather than the big global banks.  This matters because the large global banks are for the most part much better capitalised.  They will remain so, due to risk-aversion on the part of managements.  The uncertainty creates an incentive for management to take a worst-case approach to capital ratios, ensuring that whatever the eventual regulatory outcome, they will still remain in control of their own destiny (we would note that the Basel 3 document includes some quite draconian new powers to interfere in the management of banks which do not have a sufficient “buffer” above minimum capital levels, including the ability of regulators to restrict the payment of discretionary bonuses).  As a result, the period of uncertainty should see higher capital ratios than either pre-crisis or the eventual equilibrium.

Stronger competition in asset markets.  At present, the supply of credit to the real economy is restricted, as the banking sector is still going through the delevering process.  As a result of this restriction in supply, asset spreads are widening.  This delevering, however, is nearly at an end.  Once a new temporary equilibrium has been reached, the banking sector (particularly the large international banks) will realise that they have substantially more equity than they had pre-crisis (our aggregate forecast for 2011e has €819bn of Equity Tier One capital in the sector, versus €532bn at the end of 2007), and that they have to earn a return on it in an environment of weak loan demand.  Clearly it will not be possible for all players to earn an acceptable rate of return on equity unless loan demand is so strong that there is enough business for all.

Stronger competition in liability markets.    The winners will be those banks who can fund their loan book cheapest.  A significant feature of the new post-crisis environment is that for the first time in many years, there is differentiation in the marginal cost of funds to banks - pre-crisis, notoriously, the credit curve for bank paper was very flat.  Given that the new liquidity and “stable funding” rules will force banks to fund loan growth out of long-term senior bonds rather than central bank or short term interbank credit, funding costs will end up being a key competitive driver.  Price wars in deposit markets are likely (these are already beginning in the UK and Nordic region).

Interdependence between banks and other financials.  The most important regulatory issue for the next five years for the banking system is its need to extend the average maturity of its liabilities (ie, to sell long-dated bonds).  The most important regulatory issue for the next five years for the life and pensions industry is the need to match the duration of its assets to the duration of its liabilities (ie, to buy long-dated bonds).  In principle, it is easy to see a solution here.  In practice, the way in which a greater proportion of the lending system ends up being financed out of long term savings (the only possible solution at the level of the whole economy) will depend to a great extent on the interaction between banking and insurance regulation - at present, the way in which the Second Solvency Directive treats holdings of non-government bonds by life assurers makes it very difficult for them to buy long-dated bonds issued by all but the very highest-rated[2] banks.  Banking and insurance regulators do not have a particularly good history of co-ordinating their activities, and this problem is exacerbated by the fact that there is no real counterpart to the Basel Supervisors’ Committee - the equivalent international umbrella groups for insurance and securities regulators are much less influential and prestigious - and by the fact that insurance regulation in the USA is delegated to state insurance commissions to a much greater extent than banking regulation.

Regulations which weaken as they drag on.  The Basel consultative paper published in December was as hostile to the industry as things get.  Most of its proposals will weaken as the debate drags on.  This is not a prediction based on cynicism or assumptions about industry pressure - it is the natural result of the regulatory process.  The Quantitative Impact Studies are an important part of the process, not a formality; the way that Basel works is that it publishes a tough standard and then carries out an impact study to see how much it needs to be weakened.

2016+ - 2020 - The years of regulatory arbitrage

When we finally have a stable global regulatory architecture (and this certainly might take longer than five years), the international banks will begin to restructure their businesses around it.  On the assumption that the eventual 2016 regulatory system will look something like the Basel 3 proposals in general form (as was the case with respect to Basel 2 - the specific regulations changed markedly in the various drafts, but the general trend toward use of internal models generally and Value-At-Risk specifically was clear even before the first official consultation document), we would be looking for the following list of developments:

Spin-offs of activities currently carried out by banks into the hedge fund sector.  When financial regulations change, it is not uncommon for banks to terminate their activities in business lines that generate an excessive capital charge, leaving the people who carried them out to restart their businesses in new financial companies not covered by the new regulations.  We would note that it should not be assumed that the activities spun out will be the riskiest business lines; potentially, quite the reverse.  This presumes, of course, that there will still be a non-regulated sector in existence - in our opinion, despite increasing pressure for some form of registration of hedge funds doing business in the major onshore jurisdictions, there will still remain a large class of institutions which it makes sense to describe as broadly “unregulated”, in that there will not be control over their levels of leverage or their trading activities.

Increased use of central counterparties and clearing houses.  The main changes brought in by the Basel 3 framework will be (via the leverage rule) to bring in a flat-rate capital requirement for balance sheet size, regardless of risk category and to significantly increase the risk weighting on counterparty exposures on trading business. The unweighted leverage ratio approach was, of course, considered and rejected (by all parties except the USA) in the original 1987 Basel Accord precisely because it was felt that it created an incentive to divest low-risk assets and acquire high-risk ones.  This would suggest that interest rate and foreign exchange swaps business might be particularly penalised - currently, these are low-margin activities reflecting their low capital consumption, which in turn reflects the low risks they pose.  The increase in capital charge on these business lines can be substantially mitigated by moving to exchange trading or central clearing, and it may be worth the banks’ while to make this move, even if it does come at a cost in terms of pricing power as centralisation brings with it increased transparency.

Fragmentation of group structures.  It also appears to be the case that in the national implementation of Basel 3, regulators will take the opportunity to ensure that local entities within their jurisdiction are ring-fenced in terms of their capital and liquidity (this was one of the main lessons taken from the Lehman insolvency), perhaps to the extent of literally requiring banks to operate through separately capitalised subsidiaries.  This imposes a cost on global firms who currently manage capital and funding on a global basis across their legal entities, while also increasing tax exposure.  This may make firms re-assess the benefits of having a presence in their more marginal jurisdictions.

2020-2025 - The return of deregulation

By 2020, the crisis will be more than ten years in the past, and the large international banks will have normalised their capital structure - gearing levels will be lower, and there will be less short-term funding than before the crisis, but the extra capital held during the years of regulatory uncertainty will have been returned to shareholders or deployed in balance sheet business.  It is also likely that the differentiation between high and low quality institutions in the credit market will not be so steep; industry profitability will have recovered and competitive pressures decreased, particularly as the recapitalisation of the industry in the preceding ten years will most likely have resulted in material consolidation (official concerns about “too big to fail” notwithstanding) and the privatisation and demutualisation of a number of institutions which currently subsidise their competitive position through not needing to meet a return-on-equity constraint.

The competitive pressure that exists, however, will be coming from the non-bank sector and from non-regulated or lightly-regulated institutions.  There will be considerable lobbying and pressure from the industry to “level the playing field” (and of course, there are no countervailing interests exerting pressure to do so by increasing constraints on the non-regulated sector rather than relaxing them on the regulated institutions).

Deregulatory tendencies will be exacerbated by international competition.  Elements of international regulatory competition are likely to be seen well before this period (particularly, if the Obama administration succeeds in placing prescriptive regulations on investment banking activities, which is not our central case at present, European jurisdictions are likely to see this as an opportunity to take market share, as they have in all previous episodes of US reregulation from Glass-Steagall to Regulation Q). Gradually, some of the key provisions of Basel 3, probably on the liquidity side (the current suggested stable funding regime is quite prescriptive and a plausible case could be made for allowing internal modelling of liquidity requirements, for example) will be relaxed. New financial innovations will be created and the regulatory regime for these will be substantially lighter than for traditional banking products, encouraging their use.  Restrictions on cross-ownership will be relaxed, as they have been in every major deregulatory episode, allowing the banking industry to restore those parts of their franchise which were spun off in the previous period (compare with the rise of “captive” hedge funds in the 2004-8 period).

2025-2030 - The next crisis

Obviously there is no point in trying to forecast the date or particular characteristics of the next crisis.  However, a fifteen year gap between major crises in the financial industry would be a long one by historical standards (counting backwards and ignoring purely local crises even if they are as big as the Nordic, Spanish or French ones, we have the current crisis which began in 2007, the dot com bust in 2000, the Asia-Russia crisis in 1997, the Mexico crisis in 1995, the commercial real estate crises in 1991, the Japanese banking crisis and the Lloyd’s market crisis in 1988, Black Wednesday in 1987, the S&L crisis in 1985, the LDC debt crisis in 1982, the UK secondary banking crisis in 1975 …).  And in general, crises share a number of characteristics.

They are associated with lightly-regulated or recently deregulated parts of the industry

They often centre round structures based on closed-end investment companies with unusual or “sophisticated” capital structures which aim to offer low-risk securities with a higher return than government bonds (CDOs fit this description, as did the investment trusts of the 1930s, split-caps etc).

They tend to involve markets in which either non-insurers are attempting to write insurance (CDS), non-banks are attempting to offer deposits (money market mutual funds) or non-trading companies are attempting to carry out futures trading (Enron).

They often involve a new development in finance or economics which makes it appear to the innovators that their model suggests that assets can be bought or liabilities can be assumed from the people who originated them, systematically and at a profit.

They invariably involve an increase in leverage, either on- or off- balance sheet - this is practically constitutive of a financial crisis, the dot com bust being the only major crisis not to have involved rapid growth in debt (and even then, it is definitely arguable that the so-called “dot com crisis” was actually a crisis of highly leveraged telecom and cable companies).

A little more than half of the major financial crises of the last forty years have involved real estate in some way, and most of the rest have involved emerging markets.

[1] It should be noted that the current consultation document of Basel 3 more or less exactly reflects current US regulatory practice on most important points (specifically, the practice of the Office of the Comptroller of the Currency).  There is a reason for this - the Basel 2 experience of trying, and failing, to reach a compromise with the US regulators.  In drafting the new rules, the Basel Committee was presumably working on the assumption that the only way to reach agreement with the USA in the timetable given was to give them everything they wanted

[2] Some comment on the ratings agencies would seem to be appropriate, but although the regulators realise that there is a problem with respect to the dependence of a number of key regulatory regimes on agency ratings, nobody at present has a sensible proposal for an alternative.  It is not impossible that some regulators, particularly in Europe, will end up starting their own bond-rating organisations - IBCA (now part of Fitch) was originally created specifically to rate counterparties of the World Bank in response to a desire on the part of European countries for a non-American rating agency.