In the wake of the crisis, the question of whether financial markets are capable of effective self-regulation took centre stage. The near unanimous verdict was that they are not. The crisis itself, following on as it did from a period of extended deregulation, seemed to provide a definitive QED. So much so that surprisingly little attention has been devoted to working out why this might be so.
It has, in short, become an article of received wisdom, rarely questioned other than at sites like The Cobden Centre.
Andrew Haldane[1] of the Bank of England did so in a recent speech (PDF). Although I’m not convinced he always followed the logic of his analysis to its natural conclusion, he clearly outlined the structural developments that led to the current debacle and offered several sensible policy suggestions.
It was a long speech: the transcript runs to eighteen closely typed pages with a further eleven of references, charts and tables. It would make no sense for me to try to cover the whole thing in any detail: for those sufficiently interested in the topic, do read the original.
What I want to do is bring forward enough of the material to enable a closer focus on some of the more critical issues, and to highlight a few areas where I think Mr. Haldane may be in error.
♦ ♦ ♦
He began with a review of changing incentive and control structures within banking stretching back to the early 19th century.
At that time, most banks operated as unlimited liability partnerships. Tightly hobbled in this fashion, “banking was a low concentration, low leverage, high liquidity business.”
“Equity capital often accounted for as much as half of all liabilities, and liquid securities frequently accounted for as much as 30% of banks assets.”
Whatever its other merits, at least under this arrangement “governance and balance sheet structure [were] mutually compatible.”
With bank investors so utterly exposed, credit growth was severely hampered. The structure also relied on the substance of bank shareholders. For both reasons, the pressure for change was intense, led, amongst others, by Walter Bagehot.
In time, therefore, unlimited liability changed to extended liability, usually composed of some mixture of reserve liability and uncalled capital:
“Under reserve liability, existing shareholders were liable for additional capital in the event of bankruptcy. By 1884, British banks had reserve liability of around three times their paid-up capital. This placed them on a similar footing to US banks, which had adopted a system of double liability in 1863.”
Although less strict, this new regime also kept risk appetites under control for a time. Two developments led to its eventual abandonment. First, as the banks grew in size and progressively consolidated, vetting shareholders to ensure their capacity to meet reserve liabilities became increasingly impractical. Second, and arguably more critically, actually calling up reserve liabilities came to be seen as self-defeating. Doing so might well exacerbate a crisis. Boards simply couldn’t bring themselves to pull the trigger when in theory it might have done the most good.
And so this second regime also gave way to something more closely resembling our own:
“[B]y the 1930s the governance and balance sheet structure of banks was unrecognisable from a century earlier. Ownership and control were amicably divorced. Ownership was vested in a widely dispersed set of shareholders, unvetted and anonymous. Their upside payoffs remained unlimited, but the downside risks were now capped by unlimited liability. The pool of reserve capital had largely evaporated.”
♦ ♦ ♦
At this point, Haldane brought in Robert Merton’s contingent claims model. Although new to me, it rang true at once. In simple terms, it tells us “that the equity of a limited liability company can be valued as a call option on its assets, with a strike price equal to the value of its liabilities.”
It’s an elegant little concept. And, while it applies to any limited liability company, clearly the greater the leverage the more it skews the incentives of equity holders. Ergo, it’s tailor-made for messing with bank behaviour.
Still, I’m not entirely comfortable with some of the conclusions Haldane drew from it.
He takes the fact that the value of an option is enhanced by an increase in the volatility of whatever underlies it and applies that directly to the much broader canvas of bank behaviour. “Because volatility increases the upside return without affecting the downside risk”, banks will naturally “seek bigger and riskier bets” to “maximise shareholder value”. There’s much truth in this, but I think there’s also a fundamental error. After all, the option represented by bank equity vanishes if a bank fails. Equally, although he’s right that the “downside risk” is not absolutely affected by increased volatility (the equity holder can lose no more than he’s invested), the probability of that risk materialising changes if “bigger and riskier bets” are taken.
His conclusion that “joint stock banking with limited liability puts ownership in the hands of a volatility junkie” therefore seems overstated, perhaps grievously so.
What’s more relevant, I think, is the incentive effect on management. With the spread of equity participation through option grants in recent decades, senior bank management in effect acquired an option on an option (appropriate, perhaps, in an era that also gave us CDOs both squared and cubed).
It isn’t that Haldane ignores management incentives. Indeed, late in the speech he discusses the effect of introducing return on equity (ROE) as the principle performance metric at some length, and there’s no doubt in doing so he’s highlighting something important. Had return on assets (ROA) been chosen instead, things would have turned out very differently. As he says, it has the great virtue of “cover[ing] the whole balance sheet and, because it is not flattered by leverage, do[ing] a better job of adjusting for risk.”
Not least, it would have drastically cut the absurd remuneration. Had ROA rather than ROE been the metric from 1989 onwards:
“By 2007, their (bank CEOs) compensation would not have grown tenfold. Instead it would have risen from $2.8 million to $3.4 million. Rather than rising to 500 times median US household income, it would have fallen to around 68 times.”
♦ ♦ ♦
At this point, he briefly recaps and then highlights what he sees as a puzzle:
“The story so far. Ownership and control rights for banks are vested in agents comprising less than 5% of the balance sheet. To boost equity returns, there are strong incentives for owners to increase volatility. Those risk-taking flames have been fanned by tax [policies favouring debt over equity] and state aid [all the direct and indirect state underwriting of bank related risks]. As stories go, this one sounds grim.
But this story also contains a puzzle. Long-term shareholders in banks have not obviously reaped the benefits of these distortions. The purchaser of a portfolio of global banking stocks in the early 1990s is today sitting on a real loss. So who exactly is it extracting value from these incentive distortions? The answer is twofold: shorter-term investors and bank management.”
Reading that second paragraph for the first time, I thought “Aha. Finally.” I was sure the answer to the “who” would be bondholders and bank management.
Here’s his explanation for putting in shorter-term investors instead:
“Institutional investors in equities are typically structurally long. They gain and lose symmetrically as returns rise and fall. Many shorter-term investors face no such restrictions. If their timing is right, they can win on both the upswings (when long) and the downswings (when short). For them, the road to riches is a bumpy one -and the bigger the bumps the better. As in Merton’s model, all volatility is good volatility.”
Although he’s perhaps a bit blasé about the ease with which market swings can be timed, it’s certainly true there are winners from the shorter-term trading game. And market-makers do thrive on volatility (albeit within reason, even for them). However, these opportunities aren’t confined to bank stocks; they apply to all stocks, indeed to all markets.
The sharp fall in average bank stock holding periods he thinks might bolster his thesis is equally universal. Average holding periods for all equities have been in secular decline for at least 50 years, and the patterns look much the same.
I can’t see, therefore, that short-term investors deserve their spot on this podium of winners. Bank debtors, my first pick, still seem a far better choice.
When he turns to bank management (the other big winner), his focus is on the effect of tying incentive arrangements to ROE. It’s a most interesting and useful discussion, but I’m not sure it properly zeros in on the deeper principal-agent problem.
Unlike bank shareholders, management really doesn’t have much downside. Because they’re effectively holding an option on an option, their incentive skew was, and often still is, near absolute. Setting aside reputation for a moment (which seems to have been a tolerably safe thing to do in the last decade), it truly was all upside.
I’m not suggesting this explains the crisis: far from it. But it did provide a powerful incentive to light the afterburners on a system that was already skewed and overextended.
Before moving on, one last quote from his discussion of the ownership versus control dilemma:
“When the downswing came, the volatility of equity returns sent many banks to the wall.”
Obviously true, but I’m not sure it sits all that well with his earlier comments about the incentive structure for equity holders, much less with Merton’s “all volatility is good volatility”.
♦ ♦ ♦
Let’s turn to the missing component in this discussion of asymmetrical incentives: the role and influence of bank debt.
Haldane does a great job tracing the historical experience and analysing where things went so badly wrong. Put simply, lenders to banks should in theory provide a countervailing influence to the somewhat one-sided incentives of shareholders. After all, their risk/reward curves are near mirror images:
“This quasi-disciplining role of debt persisted up to the Great Depression. Calomiris and Mason (1997) find that debt and equity prices did a reasonable job of distinguishing good and bad banks during the Chicago banking panic of 1932. Indeed, they signalled distress fully six months prior to banks’ failure.
As the 20th century progressed, however, evidence of debt disciplining became patchier. Studies in the 1980s typically failed to find balance sheet risk having a significant impact on banks’ subordinated debt spreads (Gorton and Santomero (1990)). Evidence from the 1990s was more encouraging (Flannery and Sorescu (1996), Morgan and Stiroh (1999)). But even then the link was weaker among larger banks, with little evidence of market prices influencing banks’ risk decisions (Bliss and Flannery (2002)).”
During the lead up to the crisis, this break was complete. In fact, for a while the correlation looked negative with CDS premiums for banks falling “dramatically between 2002 and 2007″ at “precisely the time risk in the system was building.”
How could this be?
“As much as bank management and the authorities may pre-commit to debtors bearing risk ex-ante, they may be tempted to capitulate ex-post.
“Economists call this a time-consistency problem. Agents cannot credibly commit to stick to their guns in the midst of war. Private contracts for bank debt and public policies towards bank debt suffer from a severe case of this time-inconsistency problem. Having debtors assume pain is fine on paper. But crisis wars are not waged on paper. And if debtors recognise that risks in contracts will not be enforced, they will no longer have incentives to price risk and exercise discipline themselves. So it has been for well over a century.”
In 2008, for example, just when it would have been most useful, banks chose not to convert hybrid debt to equity. Why? “Because they too feared scaring creditors and making a bad liquidity situation worse.” In the public sphere, the incentives are even more heavily skewed towards capitulation.
As if all this weren’t enough, the astounding growth of banking assets in relation to GDP[2] made matters even worse:
“Big, connected firms increase the chances of a bad situation turning not just worse but catastrophic. Knowing the authorities will shoulder that tail risk, debt-holders will not price it for themselves. That is doubly unfortunate, as it means debtor discipline will be weakest among institutions for whom society would wish it to be strongest. Worse than that, bigger banks will then benefit from an implicit state subsidy, for cheaper debt means fatter profits. That might itself encourage further risk-taking.”
It’s a vicious cycle and despite all the sound and fury of recent years, I don’t know that things are any better in this regard than they were in 2007. Arguably, they’re worse.
Haldane provides estimates on the value of this implicit subsidy. For the four largest UK banks, they range from tens of billions of pounds per year up to hundreds of billions. For the 22 largest global banks, the range is “hundreds of billions of dollars per year, on occasions four figures.”
Even more remarkable, perhaps, is the doleful fact that these numbers represent “a large chunk [of], and sometimes exceed, the measured value added of the financial sector to annual GDP.”
♦ ♦ ♦
At the end of the speech, Haldane looks at what might best be done. Given his focus on the way in which ever more distorted incentives lie at the root of most of these problems, he sees “these interventions [as] best directed at incentives themselves. Otherwise risk-taking is likely to be simply displaced, rather than curbed, by reform efforts.” Amen.
He put forward four suggestions:
1. Higher equity capital: As he says, it’s the most obvious solution. It would “act on at least three of the underlying incentive frictions. It would put more skin in the game for equity-holders, thereby reducing their incentives to extract option value. It would reduce leverage directly, thereby reducing banks’ capacity to risk-up. And it would increase banks’ capacity to absorb loss, thereby reducing the probability of official intervention.”
He views Basel III is a useful starting point, but no more. He thinks the eventual capital ratio should be considerably higher, perhaps as much as 20%. He anticipates the inevitable counterargument that “equity is expensive” and acknowledges the distortions stemming from current tax policies and state subsidies. However:
“Using [these arguments] to support lower capital ratios is to argue that three wrongs make a right.”
Amen again.
He favours leveling the playing field between debt and equity, perhaps even by doing both. That is, not only reducing the tax deductibility of interest but also “allowing firms to deduct from profits an allowance for corporate equity . . . .” Radical, certainly, but he believes it may be justified because of the large negative externalities (the “deadweight costs of default and the growth sapping effects of debt overhang”).
2. Equity-like liabilities: Additional equity doesn’t entirely remove the “asymmetry of payoffs”. Nor does it “guarantee discipline by debt-holders.”
So, in addition to higher capital ratios, he favours contingent convertibles (CoCos). They’re not new, and they are controversial, but he suggests two conditions to enhance their effectiveness.
“First, no discretion on the part of bank management or the authorities about when and how conversion takes place. Such discretion undermined the effectiveness of uncalled capital in the 20th century and hybrids in the 21st.”
“Second, conversion needs to take place well ahead of bankruptcy. Doing so avoids the deadweight costs of default which, for too-big-to-fail institutions, are likely to be too large to be tolerable by the authorities. That is what undermined reserve liability in the 20th century. It is also what risks jeopardising the effectiveness of so-called bail-in debt in the 21st.”
Basing the trigger for their conversion on “market-based measures of capital adequacy” (in other words, on market capitalisation) would satisfy both conditions.
3. Control rights: Here he considers extending voting rights beyond equity holders, such that “governance and control would then be distributed across the whole balance sheet.”
4. Performance and Remuneration: We covered this earlier. His suggestion, if you recall, is to shift the performance metric from ROE to ROA.
♦ ♦ ♦
Back at the start, I said I wasn’t convinced Haldane had followed the logic of his analysis to its natural end. Let me try to explain why.
Early on, he said:
“Under the assumptions of Modigliani and Miller (1963) (MM), the pricing of risk by debt-holders ought to neutralise fully any effect of increased leverage on the value of the firm. Provided shareholders maximise firm value, and the world behaves according to MM, debtor discipline ought to defuse completely incentives to gear-up.”
If that’s right, and absent distorting interventions I think it is, then most of the problems we’ve experienced (together with the seemingly endless complexity of the proposed solutions) stem from just one thing: an unwillingness to allow bank debtors to suffer loss. Or, put another way, to allow markets to function and contractual arrangements to be honoured.
The question is why.
Haldane puts it down to the time-consistency problem. “Having debtors assume pain is fine on paper. But crisis wars are not waged on paper.” I see the argument of course. I even see the force of the argument. Still, it strikes me as an assertion rather than something truly axiomatic.
Much of the unwillingness, I’m sure, is grounded in fear. The size, complexity and interwoven nature of modern universal banking is a powerful deterrent to serious meddling. What if it goes wrong? Mightn’t the whole damn thing just collapse if bank investors get scared? Then where will the funding come from? Banks of course play on this uncertainty and ignorance, sometimes subtly and sometimes with the crassest of threats.
Then too, much of the non-deposit, non-secured bank lending comes from pension funds and other institutional investors who are for the most part investing on behalf of households. No prizes for causing them pain either.
The incentives, as Haldane suggests, are indeed all skewed one way.
Still, if banks weren’t so highly leveraged, many of these concerns would melt away. Convert a sizeable chunk of bank debt to equity and the nailbiting would be done by bank management and shareholders (and, from that point on, by bank debtors) rather than regulators, politicians and the rest of us. Given that non-deposit, non-secured bank debt often represents a quarter or more of total bank liabilities, it’s not as if there’s a shortage of available capacity.
It wouldn’t be easy. I understand that. But then again, nor are any of the alternatives. The day when pleasant solutions were on offer is long past.
At any rate, I would have liked to see Haldane worry at this issue a good deal more. He’s one of the few who grasps how vital it is to get at causes rather than simply suppressing symptoms. In the hierarchy of causes, gutting the risk management contribution of the greater part of bank balance sheets must surely rank reasonably high.
_________________________________________________________
1 Executive Director, Financial Stability and Member of the Financial Policy Committee.
2 In the UK, bank assets to GDP increased from 50% to over 500% at its recent peak. This, as Haldane sees it, is the “bigger bet” strategy at work in pursuit of higher equity returns. So too with the increased variation in returns on bank assets (“two and a half times more volatile at the end of the 20th century than at the beginning”), which he takes as evidence of the “riskier bet” strategy, once again encouraged by equity’s asymmetric incentives.
All true, but there were also larger influences at work. Without government backstopping and constant central bank additions to base money, the rapid relative expansion of the banking sector would have halted long ago and at much lower levels. As for the heightened volatility, isn’t that mostly an offshoot of this extraordinary growth, not only due to the increased leverage itself but also the spread of speculative trading that’s followed in its wake?
