Arnold Kling  

Help Me Understand

Still Jackasses... The Ethics and Etiquette of St...

this story:

What killed AIG was much more likely the financial and managerial collapse within AIG Global Investment Corporation's securities lending program...

It was the pure economic loss of capital in its most natural state. It represented the decline in value of assets AIGGIC purchased with the cash Wall Street's banks and brokerages gave as collateral for borrowing stocks and bonds from its life insurance investment-management portfolios.

In other words, financial crisis junkies, it was not all Joe Cassano and credit default swaps. AIG had this "securities lending" unit that loaded up on mortgage securities.

What I need help understanding is what "securities lending" means. It sounds like, oh, we just happen to have lots of securities sitting on the shelf--would you like to borrow some for a while? But the way I think of it is in terms repurchase agreements. In repo, one side lends cash and the other side is said to lend securities. I would say that you are lending securities in the sense that you are lending your ring to a pawn shop. What you really are doing is borrowing, using the securities as collateral.

So, I interpret this story as saying that AIG had a "securities lending" unit which carried long-term, risky assets using very short-term repo funding. Is that interpretation correct?

Comments and Sharing

COMMENTS (8 to date)
Milton Recht writes:

Generally, the risk in securities lending is not mismatch maturities. It is usually collateral and counterparty risk.

AIG probably lent securities, such as stocks, corporate bonds, other bonds, etc., to brokerage firms, banks, investment firms, etc.

In return, the borrower paid AIG a fee to lend, and AIG received a safe security, usually government bonds, and in this case probably mortgage related bonds, of equal or greater value, with a minimal maturity mismatching, as collateral to back up the loan.

The collateral value declined substantially (especially mortgage related securities) and the counterparties, such as Lehman, Bear Stearns, etc., were unable to repay AIG, return the collateral or increase the amount of collateral.

AIG had to write off the lost value and decrease its capital. With enough security lending and enough mortgage related collateral with steep declines in value, the capital hit to AIG could be huge.

Most of the damage to the financial system was caused by the unexpected decline in value of mortgage-backed securities used as collateral, whether for repos, securities lending, other financing and fee enhancement programs, or as investments.

The financial system shock of a couple of years ago was not due to maturity mismatching and a sudden change in yields. It was due to the unexpected (by those involved) change and decline in value of purportedly 'safe' mortgage related securities.

Chris Koresko writes:

@Arnold, @Milton:

Thank you both for showing me once again why this is such a good blog!

RSF677 writes:

Actually, AIG was investing cash collateral from the securities-lending business in subprime-linked assets and had to unload these illiquid assets at a loss when the borrowers of securities wanted their collateral back. That suggests that the problem was a combination of non-matching maturities and also investing in some really crappy AAA-rated mortgage securities that became worthless.

Richard Squire writes:

RSF677 has this right. AIG id not suffer "counteparty" risk here. AIG *was* the risky counterparty. This was because AIG was skirting state insurance regulation, which requires insurers to hold safe assets. AIG did this by lending out its safe securities (blue chip stocks, T-bills, etc.), receiving cash collateral in return, and then investing the cash collateral in subprime mortgage-backed securities. When the market for those securities tanked AIG couldn't convert them back to cash and thus ended up owing its counterparties cash collateral it couldn't repay. And the same market crash caused the contingent liabilities on the credit default swaps sold by AIG's Financial Products gorup to be triggered at the same time.

I recently published an article in the Harvard Law Review that explains why firms have incentive to take on correlated risks as AIG did:

TA writes:

As I understand it, RSF677 has it right. They were investing cash collateral. The securities they were lending were probably decent stuff owned by the various AIG insurance businesses; they were lending mainly to broker/dealers. That's a normal thing to do.

Their problem was, they invested the cash collateral in horseshit.

Elvin writes:

Securities lending programs had huge problems in 2008. I'm surprised there isn't more discussion about what happened because it clobbered liquidity for a lot of institutional investors.

Sec lending is about reinvesting the collateral (typically t-bills) into something higher earning than the rebate rate on the borrowing. Thus, it's a yield spread game. Lots of cash collateral pools had exposure to Lehman, SIVs, and asset-backed commercial paper. This is partly why there was such a panic in the cash markets--the ability to preserve $1.00 NAV was in question and threatened a run. People forget how frightened this sector was. A run on money market funds, bank STIFs, cash collateral pools was a part of the nightmare scenario that the Fed was dealing with.

Example of how sec lending lowered liquidity: you could have been invested in 100% Treasuries in October 2008, but if they had been lent out (and Treasuries are the easiest security to lend--there is tremendous demand for them), your custodian would probably have let you liquidate at most 50% of them without taking in-kind some really illiquid paper in the sec lending cash collateral pool.

Just this month a major sec lending program announced finally opened up their "gates" and ended restrictions on withdrawals. It has taken 21 months to regain liquidity in this market. For awhile in late 2008, the custodians were begging pension plan sponsors to withdraw only enough to pay benefits--no other transactions were allowed. (Well, they were allowed, but in-kind transfers of the illiquid portions of the collateral pool were part of the deal.)

I would have to see what the AIG cash collateral pool looked like. If they were streching for yield, had defaults on Lehman commercial paper or Lehman backed indexed notes, bad SIVs, and bad ASB commercial paper, then, yes, this could have been a major problem at a life insurance company that is playing the spread game.

Finally, sec lending is leverage. It is leveraging the spreads in the cash market onto your underlying asset (stocks and bonds). Practically all index funds and many active managers use sec lending to help defray costs and add alpha. It added a lot of leverage to the system. This market is in the trillions of $.

My favorite example of leverage in 2000s: index fund ETFs would lend out secturies within the ETF. In turn, the ETF itself could be lent out. That meant, that the underlying securities were lent out twice!

Mike Rulle writes:

About right.

But they are more like derivatives. I would argue they have the same system impact as derivatives---not necessarily bad at all. The difference is derivatives have a longer term almost all the time. It is rare a term repo goes longer than 3 months and the overwhelming majority are much less including over night. Derivatives, particularly the ones AIG had were multiple year trades. Like derivatives, repo trades are zero sum. Like derivatives repo trades can exceed the amount of the underlying outstanding. Repos can almost be viewed as virtually identical to derivatives except with shorter maturities.

The issue is really 3 things. The underlying credit of the instrument, collateral calls (derivatives) and haircuts (repo market), and the length of the commitment.

The AIG derivatives were insuring better credits----this is why, despite their decline in market value, they have not had any defaults. Odd but true. On the other hand, they had longer terms, thus forcing counter parties to call for collateral. If they had been repos, at the end of the term they would have required a larger haircut to be renewed if lenders of cash thought risks had worsened. Since repos tend to be shorter maturities, these trades could have been unwound sooner.

In derivatives, in theory at least, if market value declined they could have called for more collateral.

I guess what I am trying to say is that the structure (derivatives or repo) is insignificant in importance compared to the 3 issues stated above. While derivatives can reference indices and repos must entail a specific security, the fact is they can be structured in such a way that the risk difference between the two are basically identical.

AIG CDS derivatives were of a much

Great post and comments. This comment isn't directly on point to Arnold's question -- I don't quite buy the "it was rational for AIG to do this" theories. Squire's correlation-seeking argument, for example, is a little too pat, a little too ex post an explation for how the crisis occurred. If it's true, then why weren't other firms selling contingent debt and buying assets likely to devalue on the same contingency just as furiously? Some were, some weren't. Gillian Tett tells the story pretty well in *Fools Gold*.

To me it seems that nobody understood how to estimate the risk of correlated defaults ex ante, but that most firms went ahead and rolled the dice anyway once the volume heated up.

In other words, even if we can justify AIG's behavior as rational in the economic sense as "correlation seeking" or "agency costs," it was still stupid in terms of common sense, and could have been avoided had more vigilant management stayed in control of AIGFP and the SIVs whose cash-flow came through te repo markets.

Comments for this entry have been closed
Return to top