Streetwise Professor

In It’s a Wonderful Life, facing financial ruin and scandal, banker George Bailey attempts suicide.  Rescued by his guardian angel Clarence, a still bitter George laments that he wishes he had never been born.  Clarence shows him that the world would have been a poorer place without him, and George regains his will to live.

Certainly the 21st century remake of the Frank Capra/Jimmy Stewart classic couldn’t feature an executive from AIG’s Financial Products unit, right?  More likely, in the 2009 version a crowd gathering at the bridge over an icy New England river outside of Greenwich would be shouting “Jump!” to the distraught exec standing on the rail–if they didn’t rush to throw him in bodily.  The modern-day Clarence would rescue him just to torment him with a demonstration that yes, the world WOULD have been better without him because then there would have been no financial crisis.

But not so fast.  It’s not so clear by any means that this morality play would be reality-based.  It is in fact quite possible that the world would have been a worse place without AIG; that the financial crisis would have been more severe; and that taxpayer bailouts would have been bigger.

AIG has been at the center of many of the debates over the financial crisis, and possible regulatory responses to it. In particular, the company’s fate has been used to justify the creation of a clearinghouse for credit derivatives, and indeed for all OTC derivatives. It is often argued that had there been a clearinghouse, the AIG debacle wouldn’t have happened. What’s more, it is often asserted that by taking on so much real estate price risk by writing protection on collateralized debt obligations backed up by subprime mortgage debt, that AIG triggered the financial crisis when these positions went south, requiring a massive taxpayer bailout.

In my view, both views are wrong.  Although AIG’s decisions were certainly a disaster for its shareholders, the systemic implications of AIG’s original plunge into CDOs, and the subsequent collapse of this strategy are vastly overstated.  AIG’s collapse was a symptom of the underlying systemic problem, not its cause.  The cause was the real estate bubble and the huge pyramid of structured finance built on top of it.  Although AIG arguably contributed to the size of that pyramid, its shareholders absorbed a good deal of the blow resulting from its collapse.  Without AIG, the major banks and investment banks–the Citis, the Goldmans, and the other firms that traded with AIG–would have suffered even worse losses, and required an even larger bailout.  A systemic event would have occurred if AIG didn’t exist, and that crisis likely would have been more severe.

So, maybe AIG isn’t George Bailey.  But it isn’t the financial devil incarnate either.  The obsessive focus on the company, and on legislative and regulatory remedies to prevent the recurrence of another AIG, are distracting attention from the true sources of the financial crisis, and the policy lessons to be drawn from it.

I’ll first focus on the clearing issue, because (a) Treasury Secretary Tim Geithner recommended the mandatory formation of a clearinghouse as a means for preventing future financial crises, and (b) it is a focus of my research interest.  A close examination of clearing also helps to bring into sharp focus the key forces driving the financial crisis, and AIG’s role in it.

This view that clearing could have prevented the AIG problem, and the necessity of spending amounts of huge taxpayer dollars in a bailout, is based on an incomplete understanding of how clearing actually works. A more complete analysis demonstrates that it is unlikely that clearing would have made a blow up less likely, and it would almost certainly have made things worse by concentrating the risk on fewer systemically important banks.

This conclusion is obviously completely contrary to the conventional wisdom, but I think it is the right one once one takes into account the way derivatives markets and clearing actually work. The situation is a complex one, so it takes a rather involved argument to make the point. I apologize in advance, but bear with me. Hopefully it will be worth it.

It is helpful to break the argument into three parts. The first part assumes that the adoption of clearing would not have affected the positions that AIG assumed. The second part considers whether clearing would have induced AIG to take on smaller positions. The third part examines whether things would have been all that different had AIG in fact reduced its positions in response to the adoption of clearing.

The conclusions, in brief:

• Holding AIG’s positions constant, clearing would have not substantially affected the allocation of losses among its trading parties, and if these losses required a bailout without a clearinghouse, they would have required them with a clearinghouse;
• if anything, the losses from an AIG default would have been concentrated at fewer banks (the members of the clearinghouse, a subset of AIG’s counterparties);
• although clearing would have presumably raised the costs that AIG incurred to hold positions (due to margining), it is very plausible that these costs would not have been so large to have induced AIG to reduce substantially its positions, given its estimation of how profitable they were;
• even if AIG had reduced its positions, since its counterparties were trading to hedge their exposures to structured products, these counterparties would have incurred larger losses on these positions, losses that would have likely .have required a government bailout of these firms; and
• only if clearing had led AIG to scale back its trading, AND if such a scaling back of AIG’s trading had led to a substantial reduction in the issuance and holding of the securities that AIG’s counterparties were hedging through the insurer, AND if this in turn had led to a substantial decline in the amount of subprime lending, would clearing have had a material effect on the financial crisis.

That is, any effect of clearing on the financial crisis insofar as it relates to AIG would have been extremely indirect. Moreover, it is very difficult to evaluate just how extensive these indirect effects might have been.  Indeed, in my view, there were so many powerful economic forces contributing to the real estate bubble and the explosion in structured finance, that it is likely that both would have assumed almost the same proportions, and imploded with the same effects even if AIG had never been born.

Several themes recur through the analysis, and given that it can become complicated at times, it is worthwhile highlighting those at the outset.

The first theme is that there is a very limited cast of characters here: AIG and the major banks that AIG traded with. The major banks bore AIG’s counterparty risk as OTC counterparties. But crucially, these same banks would almost certainly have been the members of a clearinghouse, and hence would have borne the counterparty risk even in a cleared market! Indeed, due to the way clearing works, these banks might have had a greater exposure to AIG counterparty risk in a cleared market than in a bilateral one.

The second theme is that these very same banks were trading with AIG to hedge the risk on various products (e.g., CDOs) that they carried on their balance sheets. Had AIG not been around, or had traded less, they would have suffered losses on these products. In the event, AIG bore the losses. That is, AIG’s equity absorbed losses that otherwise would have fallen on the equity of these very same banks. They would have been in even worse shape than they are now! So, if it was necessary to bail out AIG to ensure these banks didn’t fail, without AIG it would have been necessary to have bailed out the banks directly, and paid more money to do it.

Only if the absence of an AIG to absorb risk had induced the banks not to buy these risky securities in the first place, and this led to a reduced scale of issuance of these securities, would things have played out differently in a good way. That outcome is highly conjectural. Therefore, any assertion that clearing, or any other regulatory or institutional change that constrained AIG’s risk taking is equally conjectural.

Let’s now get to the details of the argument, and compare how counterparty risk is allocated in a clearinghouse and a bilateral market. This comparison has immediate implications for the effects clearing would have had on the allocation of this risk, and the ultimate effects of the real estate collapse on the health of large financial institutions.

A clearinghouse (central counterparty, “CCP”) has member firms. These members are usually large financial institutions. A credit derivatives clearinghouse would almost certainly be dominated by large banks. The clearinghouse only has a relationship with its members. Customers do not have a direct relationship with the clearinghouse. They must trade through members.

I’ll assume that AIG would have not been a member of the clearinghouse (a similar argument would obtain if it were.) Therefore, it would have traded through member firms—major banks. AIG would have had accounts at one or more clearing members. Given its size, it would likely have had accounts at several members.

The important thing to recognize is that these members are the first line of defense against a default by a customer. That is, if a customer defaults, the firm(s) it clears through must make good the loss. Only if these default losses in turn force one of the clearing members into default do losses get passed to the clearinghouse. But then, the losses are shared among the other members of the clearinghouse. If (a) the capital of the clearinghouse, including any guaranty fund, cannot cover the default losses, and (b) the obligations of solvent clearing members to cover any remaining loss are smaller than that loss, then (c) those with positions on the opposite side of the market to the defaulter would absorb losses.

So, who bears the losses of a default?: (a) the other members of the clearinghouse, and (b) perhaps other customers (if the default loss is so large as to break the clearinghouse).

To get an idea of who the members of a clearinghouse likely would have been, let’s look at the current list of members of ICE Trust, the first CDS clearinghouse to get up and running:

• Bank of America Corp.
• Barclays Capital
• Citigroup
• Credit Suisse
• Deutsche Bank
• Goldman Sachs
• JP Morgan Chase & Co.
• Merrill Lynch & Co.
• Morgan Stanley
• UBS

Now let’s look at the list of the major counterparties of AIG who received money from the bailout:

• Société Générale (France)
• Goldman Sachs
• Merrill Lynch International
• Deutsche Bank
• Calyon, Crédit Agricole (France)
• UBS
• Barclays
• Coral Purchasing, DZ Bank
• Bank of Montreal
• Rabobank
• Royal Bank of Scotland
• Bank of America
• Wachovia
• HSBC
• Barclays Global Investors
• Here’s a link that breaks out the amount owed to each counterparty.  The original document has disappeared from an AIG website.

Note that there is a lot of overlap there. Moreover, if AIG’s default had been so huge as to break the clearinghouse, the very same counterparties in the list would have borne some loss.

That is, given AIG’s positions, clearing would have had NO EFFECT on the size of AIG’s loss. Moreover, it would have had little effect on who bore those risks. Indeed, it would have almost certainly INCREASED the loss borne by the banks that would have been members of the clearinghouse.

Why? Well, exactly because the overlap between counterparties and CCP members is incomplete.  The counterparty firms receiving bailouts  not in the list of CCP members (plus other firms that traded with AIG but which didn’t receive bailouts, e.g., some hedge funds) had taken positions on the opposite side of the market from AIG. These firms would have made money when AIG lost money. In a cleared market, in the event of an AIG default, the clearing firms would have been obligated to pay these firms what AIG owed them, but could not pay. In contrast, in a bilateral market, the banks in the list would have had no such obligation. Thus, the clearing member banks’ share of the loss arising from an AIG default would have been larger with clearing, than they were in a bilateral market.

So, if AIG’s failure necessitated a bailout to save these banks without clearing, its failure would have required a bigger bailout of the CCP member banks with clearing!  That is, Goldman, Citi, Merrill, BofA, etc., would have needed more support.  Moreover, the total would have likely exceeded the amount actually distributed to all the counterparties, because (a) presumably not all of AIG’s counterparties received bailout money, just those deemed systemically important, and (b) in a cleared market, the clearing members would be on the hook to all those owed money, not just those favored by the government.

In a nutshell: given the size of AIG’s positions, the total loss from its would be the same with clearing or without.  All clearing would have done is affected the allocation of those losses among counterparties.  The most likely outcome is that clearing would have concentrated the loss on a smaller number of systemically important banks–the very same banks that the government decided could not go under, and who had to be bailed out by funneling money through AIG.

This means that a necessary condition for clearing to have made things better is that it would have induced AIG to hold smaller positions. This raises two questions: (a) is it plausible that this would have happened?; and (b) even if clearing had induced AIG to hold smaller positions, would it really have made things better (i.e., would smaller positions have been sufficient to ensure that the financial crisis would have been less severe)? I believe the answers to both questions are “no!” At the very least, they are not obviously “yes.”

First consider whether clearing would have induced AIG to hold smaller positions.

Due to its high credit rating, AIG’s counterparties did not demand that it post collateral when it created its positions (though they did include “credit triggers” in the deals allowing them to demand collateral in the event of an AIG downgrade).  A CCP would have established minimum initial collateral levels for customers, and implemented a system of margin calls based on changes in mark-to-market values of AIG’s positions as a customer.

Margins/collateral are costly.  They typically must be posted in cash or near-cash instruments, thereby requiring firms to hold more of their assets in these low yielding instruments than they would prefer.  Thus, it is possible that by increasing the amount of collateral that AIG would have had to post that a CCP would have raised the costs AIG incurred to hold its positions, thereby inducing it to cut back on them.

But I doubt they would have cut back that much.  AIG loaded the boat on the CDSs on CDOs because it believed that these instruments would be very profitable.  Yes, higher collateral would have cut into those margins, but given how big AIG’s FP unit thought the margins were, it is unlikely that they would have cut back all that much.

Moreover, based on its modeling, AIG perceived that the risk on these positions was extremely low, in large part due to its belief (based on analysis of mountains of data by Gary Gorton) that the correlations between the risks of the assets underlying the CDOs were very low.  It is almost certainly the case that a CCP, given information available at the time, and no doubt influenced by AIG’s efforts to press its views and research, would have arrived at the exact same conclusion.  The rating agencies did.  Monoline insurers did.  Why would a clearinghouse been have been any different?

If as I surmise the CCP had determined that the risk of these instruments was low, it would have set low initial margins.  Again, given AIG’s perception of the risk-reward profile of these deals, it is highly likely that it would have been willing to pay these margins, and continue to load the boat.  Perhaps not all the way to the gunwales as it did, but put it pretty deep in the water nonetheless.

Furthermore, it should be noted that by allowing it to trade without collateral, AIG’s counterparties effectively extended credit to the firm. They presumably would have been more than willing to extend credit to AIG as clearing members too, thereby allowing the firm to borrow to meet its margin requirements.  Now this borrowing would have been on balance sheet, and could have constrained the firm’s financing in other ways, whereas the implicit credit in the OTC deals did not.  Nonetheless, given AIG’s stellar credit rating (as farcical as that seems in retrospect), it could have readily funded massive CDO CDS positions; and given its appetite for this risk, likely would have done so.

In sum, although clearing, and the margining it entails, would have increased AIG’s costs of trading derivatives, I seriously doubt that it would have caused the firm to reduce its positions substantially.

But, if I am wrong, and collateralization via a clearinghouse would have caused AIG to cut back dramatically on its positions: would things have been better? Not likely.

Note that those trading with AIG were typically doing so to hedge—to lay off—the risk on positions in structured products they held on their books. And remember, who were these firms that were trading with AIG? The very same banks considered to be so essential to the survival of the financial system.

Assume for a moment that these firms’ holding of these structured products built on subprime loans would have remained unchanged even if they had not laid off the risk on AIG. Then, if AIG had taken smaller positions, these firms would have borne more risk. When real estate prices tanked, and took the value of these securities along with them, these firms would have taken a bigger loss than they did in actual fact because in the event AIG absorbed some of the losses on these positions. What this means is that the big banks the government has felt necessary to bail out indirectly through AIG would have required a bigger direct bailout because they would have suffered larger losses on their CDO portfolios.

Given the size of the CDO positions, the amount of loss would have been the same regardless of how many CDSs AIG bought.  The only thing that AIG’s trading did was affect the allocation of those losses; the losses were shifted from the banks that held the CDOs to AIG.  Recall that Treasury and the Fed deemed that the losses these banks would have suffered after AIG had absorbed a huge portion of the loss would have jeopardized the financial health of these firms.  Their health would have been in a more parlous state had AIG Financial Products never been born, as they would have had to absorb all the loss, rather than share some of it with AIG’s shareholders. Sure, an AIG bailout wouldn’t have been necessary without AIG, but an equally large, or larger, bailout of the banks holding the (unhedged) CDOs would have been required.  In effect, AIG’s shareholders bailed out the banks holding the CDOs, but didn’t have enough capital to do the job itself.  Uncle Sam kicked in the rest.

Now we’re approaching the end of the argument.  Holding AIG positions constant, clearing wouldn’t have reduced the size of the bailout, and likely would have increased it.  Even if clearing would have reduced the size of the firm’s positions, or even if AIG FP had never been born, holding constant the size of the underlying CDO positions that banks hedged through AIG, the size of the bailout would not have been decreased, and likely increased.

So, this means that AIG’s existence, or the scale of its activity, would have affected the severity of the financial crisis  only if they  had affected the amount of CDOs outstanding, and ultimately the amount of subprime debt issued.

Is this realistic?  It is certainly plausible that if AIG had traded less, or not traded at all, that (a) the cost of hedging CDO risk would have increased, (b) hence, banks would have held smaller CDO positions, (c) this would have reduced the demand for subprime debt, leading to (d) a less severe financial crisis.  But the magnitude of the impact of AIG’s actions on the size of the CDO and subprime markets was arguably not very large.  Myriad economic and political forces were behind the housing bubble and the growth of subprime and the associated explosion of structured finance.  Lax monetary policy, massive savings by China and Japan, tax subsidies for housing, policy initiatives to boost home ownership among those who could not afford it via conventional financing methods, the implicit government guarantee of Fannie and Freddie, pervasive belief that subprime finance and the associated structured finance were not risky, among other things,  all contributed to the boom and subsequent bust.  Disentangling AIG’s effect from all those other factors is difficult.  But it is likely that the AIG effect by itself, against the background of all these other factors, was small.  This is especially true when one considers that given the widespread beliefs at the time, that if AIG hadn’t done it, somebody else would have.

So, contrary to Time Magazine (surprise, surprise), AIG was not the real WMD.  It was, in effect, the particular channel through which the financial flood traveled.  The underlying causes of the flood lie elsewhere.  AIG is a symptom, not primary cause.

Indeed, given the underlying causes, the damage likely would have been worse for other financial institutions, and for taxpayers, had AIG FP never been born.  Moreover, the “solutions” du jour, namely clearing, would have probably made things worse by concentrating the impact on a smaller number of systemically important financial institutions.

AIG is insolvent.  That is not a good thing.  But consider this analogy.  A flood far more extensive than anyone believed possible occurs.  Believing the risk of a massive flood to be small, many people had built homes and businesses in an attractive area that they believed to be safe, only to have them inundated by the historic deluge.  A company had insured these houses, also believing the risk to be low.  The losses were so large, that the company was wiped out.  It had enough capital to cover some of the losses suffered by those it insured, but not all.  The government stepped in, and bailed out (financially) the homeowners, covering the losses the insurer could not.

So the questions: Would the world have been better had the insurance company not existed?  Would the taxpayers pay more, or less, had the insurer not existed?

The answers to these questions seem, to me, to be “No.”  Only to the extent that the existence of the insurance company encouraged building that, in retrospect, was imprudent would the answers be different.  And, if belief in the impossibility of such a flood was widespread, it is highly unlikely that the absence of the one insurance company that drowned in it would have had a significant impact on the number of houses built, and the economic damage suffered.

I think that the analogy is a good one.  AIG is the “seen” effect of the financial crisis.  It is the tangible evidence of its existence.  It is a conduit by which the crisis spread.  But it is not the cause.  The unseen, or the difficult to see (because of the complexity of the reality), is the true cause.

There are several meta-lessons here.  One relates to the role of derivatives.  Derivatives are, first and foremost, means of allocating risk.  They do not create it.  They shift it.  Now, the existence of a risk shifting tool, by reducing the costs of bearing risk, may lead people to make different economic decisions than they would in its absence, and this effect could influence the size of losses that occur.  But, for the most part, derivatives merely transfer the burden of economic losses, rather than affecting the magnitude of these losses.  Indeed, this shifting of risk typically reduces the cost of bearing it.  Even if a company, like AIG, fails spectacularly because of the losses it incurred in the derivatives market, that does not mean that the derivatives are bad.  That, again, is the seen.  The unseen is who would have lost, and how much more damaging that loss would have been, if some of it had not been shifted to the company that failed.  If the company had not born the loss, who would have?  Would they have been better able to bear it? The balance of the argument, to my mind, weighs strongly on the side that says the effects of the loss would have been more severe had AIG not existed.  The unseen alternative not chosen would have been worse than the seen.

There is a related meta-lesson.  I often say that one of the great lessons of the Great Depression is that people learned the wrong lessons from the Great Depression.  We are in the process of repeating that experience today.  By focusing on the observed spectacular failure of one company, pundits, and more importantly policymakers, are devising regulatory and legislative responses that do not address the true causes of the ongoing crisis, may make future crises worse, and which are likely to reduce the efficiency of markets going forward.  Mandatory clearing is just one example of that.  Geithner’s and Bernanke’s demand for sweeping discretionary powers over non-bank financial firms is another.  AIG is the poster child for these efforts.  But the focus on AIG distracts attention from the real issues.  As a result, as in the 1930s, we are likely to create a vast regulatory structure that will not make our financial system any safer, but which will impede the ability of markets to work effectively and efficiently.