I saw this piece by Wolfang Munchau, “The credit revolution looks to the long-term,” last night, and every time I’ve thought about what to say about it, I’ve gotten unusually foggy headed (of course, some readers may contend that I am always foggy headed and am merely having an atypical spell of self-awareness).
I think I have come to grips as to the reason for my reaction. Muchau’s piece attempts to set forth the reasons securitization could in theory ad value and examines whether the logic holds up. Put more simply: was securitiztion largely fool’s gold, or was there some economic merit?
This is far too large a topic to cover adequately in the space Munchau had available, and similarly, one or even several blog posts would only serve as a beginning. However, this is an important topic, since regulators will inevitably have a point of view on the merits of securitization when they decide what, if any, regulatory changes need to be implemented (perhaps I am too optimistic, but I don’t see the “financial innovation is always and ever good” point of view so popular over the last decade, perisiting much longer).
So let me add some observations, with the caveat that these are far from comprehensive, and are meant primarily to stir further though and debate.
One way to look at securitization is to go back to the world before it was prevalent, say circa 1980. Banks dominated credit intermediation; only very large issuers, like corporations, utilities, and governments issued bonds, and those had semi-annual coupons and fairly simple repayment terms (sinking funds or not). Deposit disintermeidation had just started; money market funds were becoming popular and banks had recently won permission to pay interest on their checking accounts.
Part of the reason for the large role of banks was that proportionately fewer funds were held by institutional investors. Insurers were big players, as were defined benefit plans, but the number and range of institutional fixed income managers was far smaller.
So what did banks do? The collected short-term funds, some of it readily accessible (demand deposits); some of it given to the bank for longer in return for a better rate (term deposits) and lent long. But the term of much of their long term lending, unlike the bonds sold by large players in the capital markets, had some uncertainty. You didn’t know what credit card balances your customers would carry month-to-month. Loans secured by assets such as cars and real estate, might have the asset sold, leading to early prepayment. And of course you had the pesky problem of deadbeats and credit losses.
So the bank took in lumpy uncertain deposits and make lumpy heterogeneous uncertain loans of various sorts. The bank was responsible for pricing and managing the risk, collecting and crediting the payments, and managing its own balance sheet (for instance, some of its money would come from depositors, but some would often come from interbank markets). Banks held equity as a cushion against credit losses and as a reserve against withdrawals (this is simplified but directionally correct).
Now how was the world of securitization an improvement? Munchau gives three professed benefits, which he then questions to a significant degree: profiting liquidity, pooling of information, and broader sharing (distribution) of risk. While the last, broader risk-spreading, was often cited as an advantage, if you look at a speech last year by New York Fed president Timothy Geither on innovation, the advantages he cites are risk dispersion, greater issuer and investor choice, and a resulting improvement in the ability to fine-tune risk exposures.
These all sound well and good, but they are ex-post-facto justifications. When I was younger and involved in advising banks losing market share to investment banks due to securitization, there was one very simple reason: securitization is cheaper. Having a bank stand with its balance sheet between a large number of very short term depositors and a dog’s breakfast of much longer term borrowers is inherently expensive. Any pieces you can hive off to investors who will buy loans bundled in some fashion is less costly.
And why, despite the high costs of securitization (developing the legal documents, registration with the SEC, rating agency fees, underwriting fees, ongoing reporting), is it cheaper? Go back to the banks’ balance sheet. Depositors and bank bondholders all expect to get 100 cents back on the dollar (in the stone ages of finance, most big banks were rated AAA). That’s why banks have to hold equity, and equity is costly. If you cam make those pesky assets look enough bond-like via clever packaging so that institutional investors will buy them, institutional investors are willing to take a certain level of losses in a diversified portfolio in return for higher yield. And most of them, unlike depositors, have a longer maturity requirements, so the mismatch between the sort of liability profile they have and the sort of investment that the securitized bank assets could offer was often less.
Another advantage was diversification. If you are a bank in California, all you can generate are California assets. Securitization allowed for geographic risk, meaning less exposure to the possibility that a regional economy will suffer a worse decline than the rest of the country.
Munchau points out that the advantage of liquidity proved to be illusory in many cases. I simply don’t buy the second of Munchau’s professed advantages, better pooling of information. Securitization leads to information loss. The more parties in the chain between the borrower and the end investor, the less the end investor will know. Indeed, simply being a large organization leads to information loss. The banking industry has ever shown a slightly increasing cost curve (big banks have higher costs than smaller banks) despite the fact that big banks can fund at much more favorable costs than small banks. I’ve long believed the reason is that the small bank “know your community” model is cheaper and produces better outcomes than multi-level loan approvals based on credit scoring.
And Munchau omits what I think is securitization’s fatal flaw, namely, agency problems. Geithner mentioned them in his March 2007 speech:
The response of prices and volatility to the downgrades in the automobile sector in the spring of 2005 and the recent experience in subprime mortgages and related asset-backed securities and credit derivatives illustrate different types of surprises faced by the participants in these markets. They are a reminder of the dimensions of uncertainty that exist about the shape of the distribution of potential returns. This is particularly true of what we might call the adverse tail, or the negative extreme. These challenges exist for all participants in these markets: for the institutions that underwrite, structure and distribute these risks; for those who trade or hold them; and for third parties, like the rating agencies.
These challenges of complexity are significant as well because they can exacerbate the problem of dealing with classic principal-agent problems. You can see this in the subprime mortgage market where, for example, a person may be rewarded for generating new mortgages on the basis of volume, without being directly exposed to the consequences of default; but these problems exist wherever incentives diverge and contracting is imperfect. Financial institutions typically maintain a range of different checks and balances to deal with the risk of misaligned incentives: for example, between a trader and the principal whose resources are at risk, or between the mortgage broker or underwriter and the firm that ultimately ends up holding the risk. But these checks and balances depend in part on the capacity of risk managers to observe and understand the underlying economic risk in these instruments. Where that is harder because of this combination of complexity, imbedded leverage, and short loss history, then market discipline will be weaker.
Munchau believes there will be a place for securitization, albeit likely a smaller one, because it offers some fundamental advantages. I believe it will continue, but for more cynical reasons; Wall Street has little in the way of institutional memory, and past practices that were shown to be problematic at the time have too often returned, for instance, in the LBO wave of the 1980s, investment banks made what were then called bridge loans to win mandates; some of these (the most notorious was Ohio Mattress, aka “Burning Bed”) came a cropper and the practice was abandoned until recently. Off the top of my head, the only financial innovation I can think of that was discredited at the time that hasn’t come back is portfolio insurance (any other examples very much appreciated).
From the Financial Times:
Over the past 20 years, the world economy has experienced two revolutions both of which gave rise to asset price bubbles. One was the rise of modern information technology. The other was the credit revolution – the inexorable rise of modern credit markets. Both innovations triggered boom-bust episodes. The credit crisis in particular will have nasty short-term economic consequences. But how about their long-term effects, beyond this downturn?
As far as IT is concerned, I believe that the hype at the time was fully justified – just premature. The internet has revolutionised the media, retailing, banking and our lives in general. Modern telecommunications have had a dramatic impact on our mobility similar to the invention of the automobile or the train. The fact that IT has not had a bigger effect on productivity has probably more to do with the way we measure productivity than with IT itself.
So what about the long-term prospects of securitised credit? In A Short History of Financial Euphoria, John Kenneth Galbraith made the depressing claim that finance does not lend itself to innovation, full stop. What masquerades as innovation, he said, boils down to nothing more than credit secured on some asset. You can repackage it, slice it horizontally or vertically, repackage it, call it a different name, but at the end of the day, somebody owes money to somebody else. If this market booms, you can bet your securitised dollar that this is due to exuberance. This is worth discussing in some detail.
The macroeconomic function of a modern financial market is to provide liquidity, pool information and share risk. During the credit boom, proponents argued that products such as collateralised debt obligations would help distribute credit risk through the economy. While that was technically correct, they distributed it in a perverse way. Instead of channelling risk to those best able to absorb it, they channelled it to those who amplified it the most – banks.
Nor did the pooling of information work according to plan. Most of the credit market products are notoriously opaque. And the liquidity argument appears to have worked only during periods of exuberance. We must therefore conclude that the securitised markets, as they have been operating until today, could not have brought many macroeconomic benefits. Since boom-bust episodes are hugely damaging to an economy, the net macroeconomic effect of the modern credit market may well have been negative until now, and be about to get more so as this year progresses.
But I still suspect that Galbraith is wrong in the long-term. A properly functioning financial market with proper risk-sharing and information flows should be positive. Without it, there would be less finance for venture capital and private equity. Nor is there anything wrong in principle with the idea of a subprime mortgage, a product that allows poor families without a credit record to finance their homes, as long as all sides in the transaction understand the risks in full and as long as the mortgage distributor shares some responsibility in case of default.
While none of these conditions existed during the credit boom, these problems are fixable. One suggestion is to force banks to retain parts of the equity tranche of special-purpose vehicles. More hands-on regulation of mortgage distributors is also desirable. Another intriguing idea, from Stephen Cecchetti of Brandeis University, is to move parts of the credit market on to proper exchanges. Once this market is reformed, it could produce those elusive macroeconomic benefits.
What would be the magnitude of any such benefit? My guess is that it would not be massive. Even if the credit boom was a real innovation, it was probably not in the league of the IT revolution. Many of the innovations in the financial markets are not quite as new as they appear. We have been repackaging mortgages into debt instruments for a long time – either as mortgage-backed securities or covered bonds. What is new is that we are re-packaging those securities into more complex products, which in turn are repackaged into even more complex structures. With every step in the cycle, the marginal macroeconomic benefit is likely to fall.
Perhaps the most important macroeconomic effect will come from credit default swaps, instruments through which an investor can insure against payment default. We have yet to see how this market copes with a severe downturn and a sudden rise in bankruptcies – the kind of event we are likely to witness this year. The modern mathematical valuation models are probably all wrong, as they have a built-in tendency to produce risk assessments that are too optimistic. For risk-sharing to be macroeconomically beneficial, a minimal precondition is that all parties fully understand those risks.
So while the short-term outlook does not look very favourable, the future looks a lot brighter beyond this bubble – though not bright enough to warrant another boom episode.
I share your doubts about Munchau’s article and I agree that this is a very large topic. A few quick points:
The risk-sharing argument gets a bit stressed when paper rates at AAA and defaults like junk.
The “pooling of information” benefit is disingenuous. No one used to invest in MBS who couldn’t do a loan-level analysis. For the last several years, the idea has been that _no_ understanding of the underlying credits is needed to assess the creditworthiness of a bond. Problems of asymmetric information and adverse selection in securitization disappeared like magic thanks to ratings.
The retreat from securitization is going to have serious implications for the business models of many banks: loss of fee income, slower capital turnover, reduced protection of interest margins, and the death of regulatory arbitrage,
whereby a securitized credit requires less regulatory capital backing than the whole loan. The impact on the banks is easier to guess at this point than the extent of credit rationing that will follow from greatly reduced securitization.
Your last point is vitally important, and not recognized widely enough. Pretty much everyone is treating the drop-off in trading and origination as a liquidity and/or solvency problem. But no one has seriously looked at the implications of a long-run fall of securitization to, say, 20 to 40% of its former level. I imagine there will be a host of misguided measures, ranging from more monetary stimulus to heavier reliance on government guarantees, to shore up a business model that was applied far beyond its natural market. And we still wind up with the same end game: a smaller credit market.