Chance prediction or superforecasters: Reflections on the regulation of the securitisation “market”

When a ‘market’ isn’t really a market

What then, are the conventional wisdoms that we would advocate that we question²? Let’s start with the “securitisation market”.

It might seem like a facile point, but it is not helpful for securitisation to be referred to as a “market”. Securitisation is not a market, but a financing technique. Securitisation is the act of taking an asset which is not readily or easily tradable as a security (for instance a mortgage loan, a trade receivable, a credit card balance or a corporate loan) and converting it into a security that is tradable. This typically involves isolating the asset in question from the credit risk of the originator of such asset (asset isolation) and then raising finance which is repaid by the income or value arising from that asset (monetisation)³. Given its US origin, the monetisation aspect of a securitisation frequently takes the form of a bond, although many “securitisations" now use other forms of monetisation instruments, such as loans or (in an Islamic finance context) sukuk. The end result should be that the investor is exposed to the securitised asset and the originator receives economic consideration (whether cash or a sharing/disposal of risk of loss) for the asset that it has securitised. Why does this observation matter?

In the first instance, it matters because the question needs to be asked whether securitisation actually “failed”. No evidence seems to have been presented that securitisation structures did not do what they were designed to do. The holders of securitisation bonds were exposed to performance of the underlying assets – if those assets performed, they got paid and if they did not, then they lost money. It is true that some investors also lost money because they sold securitisation bonds at a loss (and in some cases, those bonds then performed well in the hands of subsequent investors). But the fact that investors chose to sell at a point when the prices for the relevant securities were depressed is a different point to whether securitisation itself worked.  The question needs to be asked – are we regulating securitisation (a) to make it work as intended or (b) to try and make sure that the prices of securitisation bonds do not fall? The answer in the case of the regulatory proposals being examined by the FSB would seem to be (b) – but that in turn raises the question whether the prices of securities can ever truly be supported by regulatory fiat?

It also matters because securitisation is now regulated in the same way whether the asset being financed is (for instance) a mortgage loan or a trade receivable. Yet trade receivables and mortgage loans are entirely different assets. No one would underwrite and/or service a trade receivable in the same way as a mortgage loan (and indeed securitisations of mortgage loans have different features and covenants to those of trade receivables). Securitisation is a financing technique the same way that loans, factoring, sales and leasebacks (to give merely a few examples) are financing techniques. Yet no one has ever attempted to regulate loans, factorings or sales and leasebacks by legislatively removing from market participants a large part of their discretion as to how such financings are to be structured⁴.  Loans, for instance, may be secured, unsecured, revolving, term, full recourse, limited recourse, committed or uncommitted. Why then must securitisations all have the same level of risk retention (or indeed, any risk retention at all)?

The question of incentives

Then we come to the concept of ‘misalignment of incentives’. This is a particularly insidious piece of conventional wisdom that has acquired an undeserved patina of intellectual respectability. Others have also challenged the ‘misalignment of incentives’ argument (see for instance the Financial Times, Alphaville, 15 April 2011, citing a study by Chris Taggart, Glenn Reynolds and Karen Klapper of CreditSights). But its origins need to be understood, as it arises from a classic category error.

From the early days of the GFC, it was noted that covered bonds did not seem to be as badly affected in the market as bank bonds or mortgage securitisations⁵. If there were two things that a fixed income investor did not want to hold in the GFC, they were bank bonds and mortgage securitisation bonds. It was therefore puzzling that covered bonds did so well as they were bank bonds secured over pools of (typically) mortgage loans. In other words, covered bonds should have been seen as combining the two most unsaleable things in the fixed income market during the GFC. Why then, did they not suffer? This conundrum was supposedly explained by the fact that the banks issuing such covered bonds had to keep the mortgage loans backing the covered bonds on their balance sheets, and so had an incentive to ensure they were better underwritten than mortgage loans backing conventional securitisations. It was largely from this observation that the idea of “misalignment of incentives”, the need to curtail originate-to-distribute business models and the requirement for risk retention arose. Yet this was all based on a misunderstanding of what covered bonds are and what made them attractive to investors.

In fact, a covered bond does not rely (much) on the backing of the cover pool that sits behind it. While there is a requirement to maintain the value of such cover pool and it is, in extremis, available to be used to pay back the covered bonds if the covered bond issuer fails, it is not the primary source of funds to pay the covered bondholders (which remains the bank itself). Even in a default scenario, covered bondholders tend to have much more limited rights to enforce the security that they hold against the cover pool than a normal secured creditor typically would. Instead, it is usually envisaged that the covered bond issuer’s regulator will play a significant role on the default of a covered bond. The cover pool may ultimately be sold to raise proceeds to redeem the covered bonds, but other resolution options are typically available, such as the transfer of both the covered bond obligations and the cover pool to a solvent bank who will become the new issuer of that covered bond. Covered bonds therefore rely less on the credit quality of the cover pool backing them and more on the likelihood of a coordinated banking system response to ensure that covered bondholders are kept whole if their issuer defaults.  Viewed in this light, covered bonds are an entirely different instrument to securitisations. Enforcement against the collateral for a securitisation would be a private matter (i.e. not involving a regulator or other public official) undertaken at the behest of the securitisation investors, typically by a security trustee or an insolvency official appointed by it. No amount of “skin in the game” could give a securitisation the same characteristics as a covered bond⁶.

Change is not always a good thing

Another question that the FSB could usefully ask, but has not, is whether constant regulatory flux (since the onset of the GFC up until now (in the case of the UK, for instance)) has itself had a negative effect on securitisation. The decline in purchases of securitisations by “real money” investors in the years immediately following the GFC was frequently used by regulators (particularly in the EU)⁷ as a justification of the need for more regulation of the securitisation market in order to improve its attractiveness to investors⁸.  A cursory review of sections 3.2.1, 3.2.3 and 3.2.4 of the Report⁹ shows that the period of regulatory flux has (conservatively measured) run from July 2009 until November 2020. This in fact understates the period during which there was near constant regulatory change and uncertainty – first this timeline refers only to BCBS and/IOSCO recommendations and not their in-country implementation and second, it does not address non-BCBS or non-IOSCO regulatory changes. During this period, neither originators nor investors could say with certainty what the cost of or return on a securitisation would be over its lifetime, given that new regulatory requirements could change that outcome. In fairness to the FSB, the Report does discuss actual implementation, but such discussion is not comprehensive, being beyond the frame of reference the FSB has set for itself.  Nevertheless, this feels like another missed opportunity to consider whether the process by which these regulatory initiatives were introduced and implemented was itself optimal and whether other approaches to sequencing regulatory change during and/or following a future financial crisis would be beneficial.

Solving the right problem

But the really big, missed opportunity is to ask the question – should these regulatory initiatives have even been pursued at all?

Philip Tetlock, the psychologist who demonstrated that, generally, our accuracy of predictions really is no better than flipping a coin, also noticed that certain people can make predictions far more accurately than the general population. These are the so called ‘superforecasters’. The key to these superforecasters, his research has found, is teamwork, looking at probabilities, taking knowledge from a variety of sources and, crucially, a willingness to own their mistakes and take a different approach.

In this regard, it has been acknowledged on a number of occasions that pre-GFC, the financial systems of developed world countries, particularly the US, were absorbing a “savings glut”¹⁰ being generated in consumer goods exporting countries (including, but not exclusively, China) and energy producing countries. There are many instances across history of developing countries having their financial systems and economies destabilised by sudden outflows of (typically developed world origin) “hot money” - including as recently as the 1997 Asian Financial Crisis. Is it possible that the GFC was the developed world financial system being paid back in its own “hot money” coin?  The debate around the GFC picked out securitisation as the culprit so quickly (and as noted above, on specious grounds) that all regulatory initiatives have sought to “fix” securitisation. Yet what if securitisation was not the methane in the coal mine, but the canary? To quote Philip Tetlock again:

“In one of history’s great ironies, scientists today know vastly more than their colleagues a century ago, and possess vastly more data-crunching power, but they are much less confident in the prospects for perfect predictability.”¹¹

In light of the above, the question of whether we may be solving the wrong problem should never be far from our minds. The Report does highlight (possibly unintentionally) that there are some opportunities to examine alternative hypotheses. According to section 3.2.4 of the Report, Australia has not adopted any of the IOSCO recommendations (sadly (from an experimental perspective), it had no choice but to adopt the BCBS recommendations which amount to the global rule book for banks everywhere). Yet securitisation in Australia seems to have remained at a fairly stable level through and post-GFC. The fact that we have this natural semi-control (we say semi-control as it has adopted the BCBS proposals) offers us a natural experiment against which we can measure the counter-factual – i.e. whether the regulatory initiatives proposed and pursued by the BCBS and IOSCO actually make securitisation safer or not.

At the very least, there ought to be a degree of intellectual curiosity as to why Australia has not suffered deleterious effects on securitisation when it has not adopted these seemingly essential so called “reforms”. At worst, failing to ask this question smacks of ignoring inconvenient evidence. Yet surely, if the aim is truly to make securitisation safer and better, the effect of non-“reform” should also be investigated?

The value of independence