Everyone knows -now at least- that the government considers some banks too big to fail. But did the banks themselves know it, and did it affect their behavior? If they were unaware they were too big to fail, could a moral hazard still be present?
It has long been a presumption that the existence of a central bank that will bail out other banks in the event of a crisis creates a moral hazard, wherein the banks’ knowledge that they will be bailed out induces riskier behavior. For instance, here is Vera Smith writing in 1936 summing up Walter Bagehot in 1873:
“It is bound to happen that a central banking system being created by State aid is more likely than a natural system to require state help, and what it knows it can depend on, it will not hesitate to utilise.”
Of course that argument presumes that the banks believe they can depend on being bailed out. But did the largest banks assume this? On Wednesday Jamie Dimon, the CEO of J.P. Morgan, made the claim that Morgan’s internal analysis didn’t assume they were too big to fail, which on the face of it might imply that their may not have been moral hazard at all.
However, Felix Salmon points out correctly that even if the banks didn’t know it, if investors lending to the banks were willing to lend at lower rates because they assumed the banks were TBTF, then being TBTF was subsidizing the banks via a lower borrowing cost. This ability to borrow at lower rate than was being paid out on subprime CDOs allowed banks to profitably borrow cheap funds to invest these in supposedly safe assets. Thus the banks behavior was altered because it did not have to face the full cost of it’s risks, ergo there can be a moral hazard even if the banks are unaware they are TBTF.
The key component of this story that moves it beyond theory is the supposed evidence that investors knew banks were TBTF and were thus willing to lend to them more cheaply. Felix buys this evidence, which is the fact that large banks have a cost of borrowing that was 78 basis points (0.78%) lower than small banks. He cites James Kwak for this statistic, who got the number from Dean Baker. My problem is that I don’t think this is a very sound number, and even if you believe Baker’s method is correct, I don’t think it’s the right number from his paper.
First, as Dean recognizes in his paper, 78 bps is the spread between small banks and large banks cost of funds for the period between Q4 2008 and Q2 2009. But even from Q1 2000 through Q4 2007 the spread was 29 bps, so unless you want to try and attribute that 29 bps spread to the banks being TBTF outside of the financial crisis, the TBTF is at most the increase in the spread of 49 bps.
Additionally, as Dean also points out, the banks being TBTF is not the only reason the spread might grow in that period. It might also be due to investors preferring, completely aside from considerations of TBTF, to lend to larger firms during periods of economic uncertainty. Supporting this hypothesis is the fact that from Q4 2001 through Q2 2002 the big-bank-small-bank-spread grew to 69 bps. Thus the more recent spread is only 9 bps above the baseline spread from the 2001 recession. How do you know that the 9 bps is a result of a TBTF premium rather than resulting from the same preference of lenders to lend to larger banks during a period of economic uncertainty? The Great Recession was certainly a period of greater uncertainty than the recession of 2001, and so a slightly larger spread would be likely on that basis alone. Again, this is unless people want to argue that any spread is a result of TBTF, but I think it’s much harder to convince people that lenders were considering the probability of bank failure in the 2001 recession or from 2000 to 2007.
As a final point, I’ll note that it’s well established that larger firms in all sorts of industries have a lower risk premium than smaller but otherwise similar firms. According to Ibbotsons Valuation Yearbook from 2008, the risk premium for the largest 10% of firms is 34 bps below the CAPM rate, and for second largest 10%of firms is 68 bps above it*. So using a standard valuation handbook, firms in the largest decile should borrow at 102 bps below the rate charged for the second largest group of firms. You hardly need to appeal to a too big to fail premium to explain a 78 bps gap between the rates charged to large banks and small banks.
So what evidence is there that the spread is a result of TBTF and not a normal size risk premium? I would be curious to see how this spread changes over time compared to the spread in other industries, how they are related to the business cycle, and whether that relationship was different during this crisis. That could provide convincing evidence in favor of a TBTF premium, as Dean’s paper has not. I’m not saying banks weren’t TBTF, or that they didn’t know it, or that investors didn’t know it, so I don’t want to argue with anyone about that. I’m just saying that as far as I can tell, this spread doesn’t prove anything.
*These numbers are only meant to be illustrative of the fact that large firms are considered to have a lower risk premium than small firms in other industries, I don’t want to quibble with anyone about these numbers, or how realistic they are, or why CAPM is bogus, etc.