Reader Richard Kline is providing a mini-series that was prompted by an anonymous reader who had observed that a complex systems theory view might raise doubts about regulatory policy. Financial overseers believe that liquidity is always and ever good, but that view may be naive:
Perhaps a lesson to be learned here is that liquidity acts as an efficient conductor of risk. It doesn’t make risk go away, but moves it more quickly from one investment sector to another.
From a complex systems theory standpoint, this is exactly what you would do if you wanted to take a stable system and destabilize it.
One of the things that helps to enable non-linear behavior in a complex system is promiscuity of information (i.e., feedback loops but in a more generalized sense) across a wide scope of the system.
Richard, in a guest post last week, posed a question he wanted to take further:
To what extent have nonlinear processes promoted the Securitization Bubble, precipitated its collapse, or prolonged the resulting instabilities in the financial system?
After a background discussion, he presented five issues:
Does innovation require untrammeled information flow across social/ economic event spaces?
Is the crisis in securitized debt the result of a ‘black swan’ event?
Was the creation of the Securitization Bubble the result of nonlinear processes in the financial markets?
Is a financial event-space optimized for propagation desirable?
If not, what structure or process parameters might improve process outcomes?
That post presented his response to the innovation question and prompted quite a few comments. Hopefully, this offering, on black swans, will elicit more reader discussion. Your comments very much appreciated.
From Richard Kline:
A black swan event, loosely described, is an occurrence that: 1) is structurally possible within a system, 2) is of very low probability within that system, but which 3) should it take place has a disproportionate impact upon the systemic order. An older name for such a systemic event is a ‘blue sky catastrophe’ (catastrophe is a technical term in dynamics), itself a specific form of bifurcation catastrophe where a system sharply and disproportionately changes its internal order following precipitating shift(s) in internal variables or order which can be quite small. Does the present financial crisis as a whole represent a low-probability nonlinear transformation of the US or the global financial system, or of major markets therein? In my view, no—or at least, not yet.
Yves here. I do have a wee definitional dispute. Taleb considered black swan events not merely to be low probability, but unimaginable. They blindside the incumbents, just as the collapse of the World Trade Center towers did. But this distinction does not affect the thrust of his argument.
Back to Kline:
It is important to consider this question closely despite the quick negative conclusion since many in and near to the financial community profess loudly that “No one could have known” that such massive losses from collateralized debt obligations were possible, or that credit spreads could widen and stay widened for months on end. The financial community does not use the black swan terminology, but the claim is the same: “Our crisis was not foreseeable.” “The system is liquid, it’s just in a stubborn panic.” “We are the victims of fraud and a lack of transparency; wider remedies would be unhelpful.”
From the other side of the financial industry’s mouth with more brass than consistency, we also hear, “Financial innovations are sound, and they did not contribute to unanticipated market instabilities.” We have ‘an accident’ where no one at the scene claims to be or know the driver. Bear in mind that the bubble in financial assets, particularly residential real estate in the US and some markets in Europe, and the collapse of that bubble should best be viewed as discrete processes with potentially distinct drivers. Both those processes embody some contributory if not primary nonlinear transformations. Even more, strictly linear and low dimensional occurrences can in combination easily produce sharply nonlinear and even chaotic outcomes. Nonetheless, on present examination neither the bubble nor its crash fit the black swan hypothesis. Many if not most of the component processes of both were linear, highly probable, and fully visible.
To consider just some contributors to the Securitization Bubble: In the period of negative real interest rates after 2001, mortgages increasingly came to be originated to borrowers at the bottom of the financial system whose capacity to pay depended upon the continuation for decades of historically very low interest rates. Similarly, hundreds of billions of dollars were lent for securitized junk bond debt speculation at the top of the financial system where the viability of that debt also depended upon the continuation of historically low interest rates. Both debt streams were extensively securitized. Concurrently with these entirely observable trajectories, bank-like speculative vehicles proliferated which borrowed, lent, and underwrote the quasi-insurance of credit default swaps to very large sums, again facilitated by historically low interest rates and by borrowing at the extreme short term via commercial paper offerings to capture the best rate spreads. No commercial bank was then permitted to operate with the severe leverage, lack of hard reserves, and term mismatches of these ‘wildcat banks,’ to resort to an historical term, since these practices, severally and jointly, have a strong association with financial failure. Wildcat quasi-banks speculated in securitized debt in very large volume; in many cases, their operating models depended upon the availability of such instruments. If, when, and as rates rose, large portions of these debts would turn sour. If these debts went sour, the value of the securitized loans as collateral would decline. If the value of the collateral declined, the terms for short term loan refinancing would shift higher in rates and requirements even if liquidity remained constant. These were linear relationships of high probability.
Judging in real time the onset of the turnaround for such trajectories is hard; however, the rise in the Fed Funds rate from 2006 marked a fair closing bell, and indeed many large financial buyers such as pension funds by then stood well back from purchasing such (in)securities. If markets were self-correcting, instability dampening systems, origination of low equity mortgages and high leverage buyout bonds would have tapered off from that point. Instead, the reverse happened. With rising interest rates, loan criteria _loosened_ and volume if anything rose.
At the same time, securitized debt underwriters shifted to ever shorter term commercial paper funding themselves for their pass-through conduits to maintain favorable interest rate spreads and thus stay level with their quasi-bank competitors, exposing themselves also to term mismatch potentials severe by historical norms. In the course of this bubble expansion, correlated asset price rises were ongoing in residential real estate, equities, commodities, and bonded debt, a highly negative indicator for the sustainability of price increases. Again, all of these trajectories were observable; all of them had high probability negative outcomes which would drive trend reversals; all of them had repeated historical validation for such outcomes; all of them were publicized as such by knowledgeable participants and observers.
By the autumn of 06, both residential real estate prices and residential mortgage failures had diverged so extremely from long-term trends that if these shifts were sustained the changes themselves would have represented a nonlinear systemic transformation. Instead, the observable linear relationships held: mortgage delinquencies rose and home prices flattened as interest rates rose, maintaining their historical correlations if at extreme values: not New and Different but More of the Same only more so.
Considered separately, the ongoing crisis in the financial system began in June 07 with the most mundane of events: a margin call on a smallish short-term debtor as their securitized collateral declined. That debtor failed to cover and publicly collapsed; a few smallish hedge funds similarly got squeezed out. When bids for their securitized collateral came in solidly below face, holders of securitized collateral lost face generally with their creditors: if “housing prices always rise,” the debt on them was now declining. The worst managed volume purchasers of this debt in Europe promptly went insolvent, indicating both the risk and the illiquidity of mortgage backed securities. Commercial paper quotes for securitized debt holders came in a levels they could not pay, while over the counter offers for their securities came in below the outstanding debt against them. Banks and bank-like holders of large securitized debt positions massively sold liquid collateral to cover their short-term positions, precipitating extensive swings in multiple markets, provoking a full-blown liquidity squeeze. The transformation of the commercial paper market does denote a nonlinear transformation, and will be discussed next in this series. However, none of the inputs to that change were themselves nonlinear. Furthermore, even if short term debt availability had declined in a linear fashion, it would certainly have declined, coming to the same position over a few months so long as the market price of securitized debt continued to decline.
By November 07, new reporting requirements made the existing market price decline of debt securities more difficult to hide, forcing more of these assets onto the books of major financial firms. Concurrently, home price declines and unsold inventories both accelerated. None of these changes were markedly nonlinear; instead, they extended decline trends already locked in. By January, large commercial banks and primary dealers began pulling back from lending or guarantees wherever they could while frantically raising capital themselves, severely exacerbating financial system liquidity constraints. The primary driver of those choices, though, appears to be their own functional insolvency due to unstated losses in securitized debt holdings, perhaps exacerbated by mismatch losses in credit default swaps. A drop of 300 basis points in interest rates did not reinflate the mortgage, junk bond, or commercial paper markets since accelerating housing price declines at the bottom of the financial system and major unrealized losses at the top of the financial system served as ironbound negative indicators for loan risk. If the exact circumstances for the implosion of Bear Stearns in March remain opaque, it is clear that they faced a major run of customers, cascade of margin calls, and withdrawal of dealer counterparties: a very large if old-fashioned bank run performed by financiers rather than retail depositors. There has yet to be any mention of a precipitating major loss for BSC from a nonlinear event. All of these events, and likely more to come, are directly driven by linear price declines in massive quantities of securitized debt, trajectories which were largely locked in from the date of issuance for these instruments.
The surge and purge of the Securitization Bubble has not been a black swan event; it has been a cooked goose event. In that respect as in prior historical examples, the faces change but the fools remain the same. A salient hypothesis from this summary, to which I’ll return, is that to the extent that nonlinear processes figured in these trajectories, they were in the making of the bubble more than in the baking of it. The ‘black swan’ metaphor fails 2), its low probability condition. In fact, we have not even seen 3), a real shift in systemic order in the financial system—yet.
Despite well-publicized failures of some exposed small and mid-size operators, there has been no default cascade to this point, either of short-term obligations of securitized debt holders or swap counterparties. Few banks or bank-like entities have failed outright, though only due to massive public lending. Over the counter securitized debt transactions can and do take place, though they have become rare. Liquidity is in the system for mundane commercial loans, though money is far more expensive. Mortgages are still issued, though far fewer and none at all for higher valuations; the volume declines are linear expressions, however. Local government revenues are declining, but service collapse, bond defaults, and bankruptcies are not widespread. Yet; not yet. And so long as price declines for securitized assets are not realized faster than asset holders acquire offsetting capital, such outright market failures may not happen at all. Those asset price decline trends?: they appear to be accelerating, and have a long way to run; a long way in relation to historical prices; a long way in relation to inventories; a long way in relation to solvency. We haven’t seen truly non-linear changes in the financial system—yet, i.e. (sustained) turbulence, catastrophic order transformation (say of market volume, market participants, or currency), or chaos.
In my view, there has been a black swan event in the US financial system within the last ten years: the budget surplus for the Federal government in the few years through 2000, and I don’t mean this facetiously. Sovereign pubic debt in the US is not intended to be retired overall but rolled over, since this debt, as the best quality available, highly liquid, and copious, is the fulcrum for all large-volume private financing. With the budget surplus, however, the potential for the contraction of outstanding Federal debt was real—even while it was an event that ‘could not happen’ in the financial system in the US functioned ‘as designed.’ Moreover and concurrently, external demand for US Treasury debt, especially from China if often indirectly, began rapid and large increases due to trade flows and their related policy responses. This external demand has proven to be somewhat insensitive to price, effectively making China the winning high bidder for Treasuries whenever it chooses to be, although this factor was not as evident eight years ago and more.
Thus an unheard of budget surplus together with major new Treasury buyers indicated that the asset basis for large-volume transactions in the US was going to contract sharply, putting pressure on top tier banks and bank-like entities to find the next best alternative, and fast. These were . . . securitized GSE instruments. Which as slightly less favorable assets carried slightly more charming rates. It was this experience which in many ways set the feet of large US financials on the slippery slope of asset backed security speculation. Thus an event structurally possible within the US financial system, but of very low probability (hadn’t happened since your grandfather was younger than your children are now, and wasn’t intended to happen at all), refocused major capital flows at the top of the system. With disastrous near term results as we now see. That the budget surplus appears to have been largely generated by capital gains thrown off by the dot.com equities bubble, and so not sustainable not to say illusory, is secondary since the effect of the surplus on the system as a whole was real at the time.
Did anyone at the time worry regarding ‘a destabilization of the financial system’ in consequence of the ephemeral budget surplus? People scratched their heads, and then applauded; no one saw ‘an event that cannot happen’ as indicative of a systemic problem—which it was beyond the summary above. This is what really happens when people see a black swan: nothing, as they have no context. This is another reason why markets heavily dependent upon the competence and reactivity of participants cannot reliably make stable adjustments to low probability events.
Christopher Zeeman. 1989. A new concept of stability. In B. Goodwin and P. Saunders, ed. Theoretical Biology.
Hyman Minsky. 1986. Stabilizing an Unstable Economy.
[Zeeman greatly developed the catastrophe concept, of which this paper is a fine distillation in a theoretical text itself splendidly rich. Minsky is a voice worth hearing.]