The World Economic Recovery: Navigating between the Positives and Pitfalls

Patrick BOLTON
Today I will talk about some specificities of the current crisis. In many ways, this is not an unusual crisis, and is one that Japan went through around 20 years ago. It is a crisis that is built on a maturity mismatch, lax monetary policy, deregulation, a real estate boom and a stock market crash, which all led to the twin crises. The new twist in this crisis is the development of shadow banking and the specific role this played. The question of how to regulate the shadow banking sector is central in this examination.

Among the similarities between the Japanese banking crisis and the recent financial crisis are low interest rates and lax monetary policies, and real estate booms. Between 1986 and 2008 in the U.S., interest rates steadily decreased and were followed by a steady increase in home ownership. In Japan over a similar period, real estate prices were closely tied to the stock market, which was also true in the U.S.

The new twists are the development of securitization, which happened so fast that many were not even aware of it, and the growth in subprime mortgages. For much of the period from March 2005 to September 2008 issuances of asset-backed securities were predominantly made up of subprime home equity loans. The sector grew rapidly and then disappeared in the summer of 2007 and later the fall of Lehman Brothers. To put the size of the asset-backed securities market into perspective, the market overtook the corporate bond market in size in 2006. Such rapid growth in a market raises concerns over why the market grew so fast and why it is being relied upon.

Until the late 1980s, a bank issued a mortgage and held on to the mortgage, so the ultimate investors in the mortgage were depositors and bank bond holders and equity holders. With the advent of securitization, new elements were added including an originating bank which was separate from the loan servicer, asset pools that were put in separate legal entities, credit enhancers, rating agencies, underwriters and others. This created a more complex landscape for loans. When loans are originated, they are put into an asset pool, are separated from the originating bank, put into a conduit, and then the final investor becomes a pension fund or money market investor, among other fixed-income securities holders. The loan servicer collects interest payments from the original borrower and the payments get channeled to the conduit and are distributed on a flow basis to the final investor.

There is a maturity mismatch in this chain, however. After the claims are issued to the final investors, someone has to pay the coupon, but the flow of interest payments was not fast enough, thus forcing short-term borrowing to pay final investors. This was done by issuing asset-backed commercial paper and other forms of short-term lending. At this stage, another claim holder enters the chain. When mortgage-backed securities are issued, the loan asset pool is used as collateral with a bond issued against it, and then a lot of cash is procured for those claims. The cash that was procured did not go back into the conduit, but was immediately used by the originating bank to offer new loans.

As for credit enhancement, to make the MBS attractive to pension fund investors, the bonds had to be very safe, AAA bonds. To make them very safe, credit enhancement needed to be added in the form of dividing the asset pool into senior and junior tranches, and introducing credit-default swaps and cash collateral. Hedge funds played the role of credit enhancers, but also and mainly "monocline" bond insurers, whose role was mysterious. Bond insurers are unregulated companies that, before the crisis, insured municipal bonds. However, this is not a lucrative market and the low probability of default led to low premiums. They began insuring mortgage-backed securities by underwriting credit-default swaps. Because they were unregulated, the amount of risk they took on went unchecked, leading to enormous risk being taken on by them.

The question remains as to why cash raised through MBS issues did not get channeled back into the conduits. The fact that it did not suggests that securitization was not done properly. One benefit of securitization is risk diversification as the typical pool would be made up of 3,000-4,000 mortgages from various regions, thus ensuring diversity.

A second benefit is distribution to long-term investors. One of the fundamental transformations of financial markets in the U.S. over the past 30 years was the growth of pension savings channeled through the financial markets. Contribution-based pensions rose dramatically which got channeled into pension funds which created greater demand for investment vehicles. This created an incentive to securitize loans originated by the banks and sell the securities to the pension funds.

Third, the most important benefit of securitization, asset-liability matching, was not done properly. The savings and loan crisis of the late 1980s was caused by a maturity mismatch combined with high exposure to interest rate risk. In this crisis, savings and loans were financed with short-term deposits or variable interest rate borrowing, and long-term loans were issued with fixed-interest. Every time the interest rate moves, the net worth of the loan shifts creating an unstable situation. Securitization has an important part to play in reducing this risk. If done properly, when long-term loans are sold on day one, the cash is realized and the same interest rate risk is seen for both assets and liabilities. This, however, did not happen.

Another stated benefit of securitization is conserving on bank capital. Banks liked securitization because it allowed them to take risk off balance sheets and thus not hold equity capital against it. But this is a bad reason to securitize. To get around capital requirements, banks played a game of shadow maturity transformation where maturity transformations were conducted off the balance sheet. Banks were relying on higher and higher non-deposit financing starting in 1990.

This is puzzling because while maturity transformation is done off the balance sheet, there is no deposit insurance to guarantee there are no runs on short-term debt, and there are no subsidies that come through deposit insurance. The maturity transformation happens without a subsidy from the regulatory system. There was, in fact, no benefit from this, but there were some distortions.

Subprime lending was a form of financial innovation which was disastrous. Normally, banks would never lend to the borrowers who received subprime loans because the risk was too high. Banks getting into this market is closely tied to securitization and large bets people were taking on the real estate market. Subprime mortgages were structured so that a mortgage would be issued knowing that the borrower could not repay the mortgage. The borrower is told that they will get low interest rates for the first two years followed by a significant increase, but by then house prices will have increased so much that they will have no trouble getting another mortgage to refinance the first.

The form mortgage-backed securities took for subprime mortgages is proof of this arrangement. They had to take account of the fact that there was an expectation of a sudden inflow of cash in a year and a half to two years when the original borrower would refinance with a new mortgage. The system was designed to handle those sudden inflows of cash.

Another problem with securitization is the incentive it creates to originate bad loans, without properly assessing the credit risk of the borrower. Here is one piece of evidence that is strong and easy to understand about this "moral hazard in origination." Looking at a pool of mortgages that were securitized, some were prime and others were subprime, assuming that a Fico score of 620 or higher is prime, and one lower than 620 is subprime. The number of loans around the 620 Fico cutoff in the mortgage pools that had been securitized and had low or no documentation was much higher for prime mortgages than subprime. This shows that because having a score above 620 led to no questions being asked and no requests for documentation. Banks knew they could securitize prime loans very easily. In contrast, subprime borrowers had a lower probability of being able to securitize the loan, and more of a background check was done. This figure shows that a loan with a high credit risk but a Fico score just above 620 would be originated without any documentation, while the same loan with a Fico score just below 620 would only be originated with documentation. This is evidence of minimal standards being applied at origination. As long as the borrower's Fico score was above 620 no further research was conducted.

The other problem with securitization was the failure of credit rating agencies in foreseeing the problem. Looking at the average number of downgrade notches hitting loans between the third quarters of 2001 and 2007; until the middle of 2006, there were a low proportion of downgrades. This was shortly followed by a huge spike in downgrades showing that in a short interval, rating agencies changed ratings very dramatically in a short period of time. This suggests that they had previously missed something very big in rating these mortgages.

The biggest mistake the rating agencies made was not taking into account the possibility of a downturn in real estate prices. All their scenarios were built on continually rising real estate prices even though in many markets real estate prices began peaking in 2006. The logic behind this was that there had not been a nationwide decline in house prices in the U.S. since the Great Depression, and since the securities were pools of diverse markets, they moved with nationwide trends.

Another thing the rating agencies failed to see was that subprime mortgages were extremely sensitive to movements in house prices because they were all built on refinancing. Rating agencies relied on imperfect historical data of past prime lending and applied it to subprime lending. Moral hazard in origination was ignored and they invited manipulation by issuers by tranching, credit enhancement and other measures. The statistical models they used were volunteered and issuers began to game the system.

Knowing what happened, the question of regulation of shadow banking arises. The most important issue in this context is the question of the treatment of conduits, as to whether they should be off the balance sheet or not. There is, in fact, very little gained by having the conduits off the balance sheet. The regulation for this would be for rating policies or direct regulation of banks to prevent, or make it harder, to do this kind of financing off the balance sheet. Credit rating agencies should be regulated, and there finally seems to be some action on that front. What was driving the shadow banking maturity transformation was a special subsidy in the current bankruptcy laws that give special treatment to derivatives, swaps and repos, which is an area where close attention must be paid to regulation.

The discussion on whether to keep conduits on or off the balance sheet is a relatively simple discussion. Many countries have securitization where the legal structure is done on the balance sheet, the covered bond institution. In this system, an issuing bank with a low credit rating wants to issue a claim that is very safe. The bank takes a pool of assets and uses them as collateral against the covered bond that it is issuing. The difference is that the holders of the bond have recourse on all the other assets of the bank, unlike with off-balance sheet securitization. If the collateral is not high enough, bond holders have access to the bank's equity. There are other arrangements to make the asset pool safe. Many covered bonds have collateral that is worth more than the face value of the bonds, and collateral can be adjusted over time. Finally, if the bond trustee requests a change and the bank fails to undertake it that is a default event and triggers debt collection.

Covered bonds are the best way of aligning incentives for origination and servicing of loans. They require higher capital commitments, and equity capital aligns incentives at origination. Covered bonds make it easier for regulators to see the extent of leverage, and tranching and allocation of risks are still feasible. If bankers are opposed to this system, I would like to know what value is added by keeping loans off the balance sheet.

Regarding regulation of ratings agencies, they are not currently regulated in the U.S., but there is action coming. The White House's white paper on financial reform only briefly mentions ratings agencies by saying that attention must be paid to them. However, there the Financial Services Committee in Congress has put up a bill regarding regulation of ratings agencies. What is known about the bill is that it touches on disclosure issues, how ratings agencies conduct risk assessment and liabilities. Currently, ratings agencies in the U.S. are not liable for their ratings, and the legislation will limit their First Amendment protections. Other key issues include the pay model, competition, and Nationally Recognized Statistical Rating Organization (NRSRO) accreditation.

Many believe the problem with the rating industry is that there are too few raters, but it can be argued that competition may not be good because competition magnifies the inherent conflicts of interest and it gives more options to the issuers to shop for good ratings. Regarding accreditation, currently, if the issuer gets a AAA rating, a pension fund manager or a money market fund can buy that issue by an NRSRO accredited agency, but some say that NRSRO accreditation should be lifted to decrease the importance of the rating and force investors to conduct independent research on their purchases. This is in my view a bad idea because this takes away from the value added by rating agencies.

Securitization can reduce the risk associated with maturity mismatch for the originator, but this has not happened. While taking maturity transformation off the balance sheet took away the subsidy of deposit insurance, there was, in fact, another subsidy in place, and there is an implicit subsidy in the current bankruptcy code in the U.S. and Japan. Lobbyists for this subsidy argue that it is very important to have a netting agreement for OTC derivatives markets, which means that there is no automatic stay applied to derivatives, swap agreements and repos. The netting agreement gives super priority to derivatives contracts in bankruptcy, but these instruments create massive systemic risk, first observed in the collapse of LTCM, and the Fed did not want to take the risk of immediate unwinding of derivatives positions, and thus bailed out LTCM. The wisdom of excluding derivatives from a stay in bankruptcy must be reconsidered.

Q: Japan also had a housing bubble, but Japan did not have the complication of shadow banking. What shadow banking did was to increase lending to dangerous levels given equity ratios. One of the governors of the Bank of England said that "casinos" and "utilities" should be separated. Can you please comment on this?

Patrick BOLTON
The argument for that separation can also be read as a call for separation of investment banking and commercial banking. Another important voice in this debate is Paul Volker, who has also argued for a return to separate regulation of commercial and investment banking. This will be an interesting debate going forward and there are many good reasons to call for separation. A basic reason, besides trying to ring a fence around the risks, is that financial institutions are so complex nowadays that they are too big to manage, with CEOs having a poor picture of the risk the firm is taking. Lack of understanding on the part of CEOs was part of the subprime crisis. It may not be wise to continue running huge banks that no one can truly understand.

Q: Is it possible to make the two types of institutions separate? There are so many implicit guarantees. For example, there was no explicit legal contract between Lehman and its structured investment vehicles (SIV). Is the demarcation possible?

TPatrick BOLTON
It is partly a matter of the size of the institution. It is less clear as to whether investment and commercial banking must necessarily be separated, or whether that is even feasible. Simply addressing the issue of being too big to fail from the point of view of being too big to manage would entail asking companies to break themselves up in any way they see fit, but to become smaller institutions. Even if the risks are mixed, if a smaller institution falls, it will not have the same systemic implications as a larger firm collapsing.

Q: One of the features of the global crisis is when too small a country sustains too big a bank so that even if the bank is operating globally, in the case of a crisis, the country must save the bank. Ireland is close to going under and Iceland is a case in point. Do you think there is something that should be done about country size and global operations?

Patrick BOLTON
That is a matter for the G20 as it is an area of fragility in the world financial system. If, for example, Citi and Merrill Lynch were incorporated in Iceland, there would be no way for the government to bail them out. This is an important relative scale issue, that came into play when Fortis collapsed. It needs to be addressed but the solution goes beyond national banking regulations.

Q: Rating agencies should thoroughly disclose their rating process and work towards thorough accountability of the ratings outcomes. What do you think about the recent discussion in the U.S. on rating agency regulation? Competition is an important issue in this regard.

Also, regarding rating agency liability, if rating agencies make mistakes regarding the rules to prevent conflicts of interest or if they manipulate data, the rating agency should be liable. What are the most optimal methods to conduct oversight of rating agencies?

Patrick BOLTON
I support relatively light regulation. For example, concerning liability, the big concern is that if rating agencies are exposed to too much liability risk, they will respond by being very conservative and think twice before giving a good rating, and no one will pay attention to ratings anymore. This may destroy the service itself, and thus minimal interventions should be put in place. My paper advocates disclosure of any rating that has been demanded by an issuer, whether it has been purchased or not. Second, as already is the case in New York, payment should be separate from whether you buy the rating or not. These are minimal interventions which I think will help.

Concerning competition, I believe that simply extending the list of NRSRO accreditation will not help much. New entrants are facing enormous hurdles, and the temptation is great for them to be friendly to issuers. What must be done is to increase accreditation, but also intervene in other ways to facilitate competition. A system that is similar to the system for auditors might make sense in that rotation by issuers will be required. This will make it easier to enter the market and will guarantee diversity of analysis.

Q: There are two aspects that have not yet been explored. First, you talked about why securitization takes place, mentioning that one of the positive reasons is to reduce the interest rate risk that was seen in the savings and loan crisis, however, in quite a few banking systems that risk may become larger. Second, what lessons should have been learned about how to structure the markets or instruments to deal with that?

Patrick BOLTON
Many of the lessons that need to be learned from what happened is that securitization should not be done away with, as securitization done well is a financial innovation that is useful. Regarding concerns about maturity mismatch, the biggest gap in the process was channeling funds. Nothing is done to regulate what is done with the cash inflow from selling securities. What has happened is that banks immediately use that cash to make new loans. This acceleration of the circulation of money is worrisome as it fuels a lending boom. Slowing down the rate at which cash can be redirected to new loans is necessary.

Q: You assume that there is excess speed which causes overcirculation of money generated in the repo and derivative markets. What is the optimal speed of circulation for this money?

Patrick BOLTON
I do not have the answer, but the point is that there is an optimal velocity, and nobody has asked how this can be addressed yet.

Q: Regarding the wisdom of excluding derivatives from stays in bankruptcies, there are many variations in derivatives. Some of them are rather new, but others are substantial in economic transactions like foreign exchange rate contracts. What kind of measures can be put in place to change the current situation?

Patrick BOLTON
I do not have the answer to this question either, but it is a question which we should be asking. I will speculate as to what may happen if we go in a certain direction. Supposing that the automatic stay is removed from OTC derivatives, the effect of that will be that we will rely less on OTC derivatives, and push institutions toward organized exchanges because netting questions will become more important in the OTC market if you know you do not have an automatic stay. This is speculation, but if this issue is addressed, maybe it will facilitate a migration toward organized exchanges.

Q: One thing that was not mentioned was compensation. This has been called a "skewed incentive" for bankers to get as much as they want, but when everything goes wrong they only get their salary. What do you propose in terms of compensation?

Goldman Sachs, for instance, has already returned public money so that they think they are free to pay as much as they want, but there are implicit guarantees so that they can raise capital at a very low interest rate. What are you thoughts on this?

Patrick BOLTON
The implicit guarantee is a huge problem and is an important reason why the government is entitled to step in and regulate compensation. Currently, the system is that for very large institutions, such as Goldman Sachs, because there is an implicit guarantee, they can borrow at very low rates and it is very easy for them to make money. They are thus making money with the help of the government. If this is not a free market, there is an a priori reason for the government to step in. This is true whether or not TARP money is repaid.

Regarding compensation, the only way to go is to structure compensation to be more long-term and there is a lot of movement in this direction. Institutions are now putting in place vesting requirements, clawback provisions and other measures, which is a relatively good development. It will not be a perfect solution, but it will help.

Q: Regarding the loan chain, did you say that on day one the cash is bypassing SIV or moving from hand to mouth?

Patrick BOLTON
That is exactly right. The SIV issues a bond which is backed by these assets and all the cash that is coming back is not going into the SIV, rather it goes to the originator, which is the problem. The SIV funds its debt obligations on the MBS through interest payments that are attached to the mortgages, though these payments are slow. The gap is closed by issuing asset-backed commercial paper or doing a reverse repo.

Q: The roles played by individual players in the loan chain, with the exception of the loan servicer and the rating agencies, seem to be interchangeable with each other. There is anecdotal evidence that banks such as Citigroup have played different roles, which can be conceptualized as financial incest. To what extent could this presumption be justified by statistical evidence, or could it even happen?

Patrick BOLTON
You are correct. The largest banks were involved in all of these areas. There is no reason why these functions should be separated as it would be very artificial. There is an incentive issue where if one is a holder of a mortgage-backed bond, one worries about the incentive on collection of interest payments. You want to structure the incentives in a particular way, but that does not mean that the role needs to be separated. The main systemic problem was the shadow maturity transformation, which should not have happened.