I’d like to propose some structure to the policy debate around the Too Big to Fail (TBTF) concept. This arises mostly from my frustration with both the debate among policymakers and the press coverage of that debate. Many people seem to be talking past each other.

The following explanation will be too elementary for some, but I find I often benefit from getting my problem definition straight before trying to debate solutions. My thoughts on this topic continue to evolve, so I invite readers with expertise to help me improve the following. I’m thinking of this both in the context of the current legislative debate, as well as to help my work as a member of the Financial Crisis Inquiry Commission.

Too Big to Fail is catchy but imprecise. I’d like to incrementally build a more complex but more precise problem definition.

Original problem definition:

Financial firm X is too big to fail.

Some misinterpret this as meaning that really big financial firms cannot fail. Of course we have seen large financial firms collapse, so we know they can fail. Our policy problem is instead that policymakers are unwilling to allow financial firm X to fail, because firm failure could cause damage to something else valued by those policymakers. This leads me to …

Revision 1:

Financial firm X is too big for policymakers to be willing to allow it to fail, because they fear that failure could have severe negative consequences for the broader financial system, economy, or society.

Severity is a key concept in Revision 1. When a factory shuts down, the town hosting that factory may vanish. While policymakers hate it when factories close and towns vanish, they generally don’t intervene to prevent this bad thing from happening. It’s the painful but necessary result of a market-based economy. In TBTF, we should only be willing to intervene if firm X’s failure would have negative effects beyond the scope of the firm that far exceed a town vanishing (which is pretty darn severe).

The other key concept is risk aversion. I know some conservatives (including some in positions of power in Congress) who argue that the Fed and Bush Administration should have just sat tight and let market forces act on Bear Stearns, Merrill Lynch, AIG, Citigroup, Fannie Mae, Freddie Mac, and others. Whether or not you agree with this view, it won’t happen. Imagine you are the Treasury Secretary, and your staff tells you that by letting firm X fail you risk a 10% chance of a catastrophic market collapse. No one who would be chosen and confirmed as Treasury Secretary would be willing to take that chance. Thus you don’t need to know that firm X’s failure will lead to a market collapse, only that there’s a big enough chance that it could lead to a market collapse.


If you stand up one giant-sized domino far from other standing dominoes, you don’t worry too much about it falling over.

If you stand up one giant-sized domino near a bunch of other dominoes, then you worry a lot about the giant one falling over because it may start a chain reaction. The failure of a hypothetical enormous financial firm that is isolated and has no ties to other firms doesn’t concern us (imagine it just buys and sells stocks on the NYSE). It’s the financial interconnections between a big financial firm and other firms that cause us to worry, because the big firm’s failure may cause other interconnected firms to fail. This leads us to …

Revision 2:

Financial firm X is too big and interconnected to other financial firms for policymakers to be willing to allow it to fail, because they fear that failure could have severe negative consequences for the broader financial system, economy, or society.


Now what do we mean by interconnected? And while we’re at it, what do we mean by fail?

I think people use fail to mean different things. Does fail mean a firm is illiquid, in which it lacks the cash to pay tomorrow’s bills and creditors?

Does fail mean a firm is insolvent, in which its assets are less than its liabilities?

Does fail mean a firm has filed for bankruptcy, which could lead to firm survival through a restructuring, a merger, or a sale, or instead to the firm being liquidated and ceasing existence?

What is the policy problem we’re trying to solve?

In the case of a commercial bank it’s easy: we worry that bankruptcy means depositors would lose some of their money. This is why we have deposit insurance, so that if a commercial bank fails, depositors have protection for their deposits (up to a limit). This prevents depositors from initiating a run on the bank when the bank is rumored to be at risk of insolvency. They don’t worry because their deposits are insured.

Since depositors are covered, I think the core unsolved policy problem is counterparty risk. If financial firm Y loans financial firm X a lot of money, then X and Y are counterparties. If X fails, then X’s inability to repay Y may cause Y to fail. Y may have to sell some of its assets quickly to generate cash to fill the hole left by X, and this fire sale may further weaken firm Y.

The clearest trigger for firm X not repaying its obligation to firm Y is bankruptcy. This is why (I think) policymakers believe they must take extreme measures to prevent firm X from going bankrupt. Bankruptcy triggers the counterparty problem.

Some mistakenly conclude that policymakers are trying to protect the equity shareholders of firm X. I have yet to meet a policymaker who is trying to do this. Everyone has the same reaction: equity shareholders gambled, and they lost. Tough luck for them. Unfortunately, the solutions implemented so far allow the shareholders to benefit. I think this is an unintended and undesirable side effect of the solutions implemented so far.

I think what matters is “How big of a counterparty is the failing firm X to firm Y?” What matters is the size of the unpaid loan relative to the portfolio of the lender, not the borrower. (Counterparties often involve financial relationships that are much more complex than a simple loan, but if we figure it out for a loan, I think it’s easily extensible to other counterparty transactions.)

One consequence is that the corporate form and the nature of the business of firm X are irrelevant. If X is a commercial bank, an investment bank, a hedge fund, a pension fund, or even a university endowment, the same logic applies. The only thing that matters in this problem definition is how large of a counterparty X is to other financial firms. Another consequence is that we also don’t care what caused firm X to fail, only that it cannot repay a big loan to firm Y, and that we’re worried about a potential cascade of dominoes. If I’m right on these points, it will significantly affect how we think about possible solutions.

Example: A kid owes you $10. Donald Trump owes you $100,000. If the kid goes bankrupt, you don’t worry too much, even if the kid started out with only $11 of assets and liabilities. If Trump goes bankrupt and can only pay you $5,000 of the $100,000 he owes you, the $95,000 loss may force you into bankruptcy, even though the $100,000 obligation was only a small fraction of Trump’s balance sheet. And you don’t care what business he was in, or whether he lost his money investing in real estate or gambling in Atlantic City. You care only that he’s not going to repay you in full and on time.

The interconnectedness between firms X and Y also depends on how highly leveraged firm Y is, for two reasons:

  1. The more highly leveraged is firm Y, the smaller of an equity cushion it has to absorb losses.
  2. Highly leveraged firms generally increase their leverage not through simple loans or bonds, but instead through more complex transactions like collateralized loans and “repurchase agreements.” More complex transactions means more counterparties, more dominoes nearby that you could knock your firm over when they fall.

Bankruptcy means that some creditors won’t be fully repaid. That’s why policymakers draw the line there, and it’s why in Revision 3 I’m going to substitute go bankrupt for fail.

Revision 3:

Financial firm X is too big and interconnected to other financial firms for policymakers to be willing to allow it to go bankrupt, because if X is a big counterparty to other firms, X’s bankruptcy might cause those other firms to go bankrupt, starting a chain reaction.


Finally, I think we need to distinguish between a solvency crisis and a liquidity crisis. A solvency crisis is when your net worth is negative. A liquidity crisis is when your net worth is positive but you lack the cash to pay today’s bills.

A confidence crisis can lead to a liquidity crisis. Imagine if firm Z has given firm X a rolling one week $100 M loan and renewed that loan every Wednesday for the past year. Assume firm X is solvent with a strong balance sheet, but that it relies on that short-term loan to stay cash-positive each week. If firm Z loses confidence in firm X and refuses to renew that loan, then firm X could face a liquidity crisis until it can find another source for short-term cash. Firm Z may be relying on good information, bad information, or rumors. Either way, X’s short term financing dries up, especially if other potential sources of short-term cash make the same decision as firm Z. (A critical question: why does firm X’s management risk its existence on such an ephemeral source of short-term financing?)

A liquidity crisis can lead to a solvency crisis. If you are desperate for cash to pay today’s bills, you can sell some of your assets at fire sale prices to raise the cash. But this weakens your balance sheet. If you lack cash tomorrow as well, your balance sheet will get worse each day as you sell assets on the cheap to pay each day’s bills. Eventually your short-term solution to your liquidity crisis can cause your net worth to go negative, in which case you’re now insolvent.

In 2008 we worried mostly about “disorderly failure.” Suppose firm X loaned $100 M to firm Y for one week every Tuesday for the past year, and suppose firm Z loaned firm Y $100 M for one week each Wednesday. Suppose you know that firm Y has a strong balance sheet, but that it needs these two $100 M short-term loans each week to remain liquid.

Suppose firm X suddenly goes bankrupt on a Monday. As a result, it cannot loan firm Y $100 M on Tuesday. Firm Z knows that firm Y is strong, but fears Y may now face a liquidity crisis. Is firm Z willing to loan firm Y $100 M on Wednesday? Firm X’s failure, combined with firm Z’s decision, can trigger a liquidity crisis at firm Y.

Y’s problem is not simply the failure of X, because Y can find another short-term lender, given enough time. Y’s problem is the sudden and disorderly failure of X, one of its big counterparties. If Y has either time to prepare for X’s failure, or sufficient time after X’s collapse to fix its problem by finding another lender, then Y will be fine.

So our final revision narrows the problem definition a bit:

Revision 4:

Financial firm X is too big and interconnected to other financial firms for policymakers to be willing to allow it to go bankrupt in a sudden and disorderly fashion, because if X is a big counterparty to other firms, X’s sudden and disorderly bankruptcy might cause those other firms to lose liquidity and go bankrupt, starting a chain reaction.

Example: In March 2008, Bear Stearns ultimately ceased to exist as a free-standing entity, despite a loan (“bailout”) from the Fed. The Fed loan bought time for Bear’s failure to result in a purchase by another firm (JP Morgan). The Fed loan was not to save Bear, it was to buy sufficient time and to facilitate (subsidize) the JP Morgan transaction, so that the failure didn’t risk causing a cascading series of failures.

I invite comments, especially those suggesting specific language improvements to this problem definition. I will turn next to looking at the classes of solutions being proposed.

(photo credit: Domino Theory At Work by r o s e n d a h l)