Krugman is Astonishingly Ignorant About the Global Financial Crisis
Margin Call (2011)
The Policy Tensor has long admired Krugman's crusade against zombie ideas in economics perpetuated by the political economy of K Street. But his latest demonstrates a remarkable ignorance of the causal mechanism behind the Global Financial Crisis (GFC):
True, nobody saw the crisis coming. But that wasn’t because orthodoxy had no room for such a thing – on the contrary, panics and bank runs are an old topic, discussed in every principles book. The reason nobody saw this coming was an empirical failure – few realized that the rise of shadow banking had done an end run around Depression-era bank safeguards.
The point was that only the dimmest of free-market ideologues reacted with utter bewilderment. The rest of us slapped our foreheads and said, “Diamond-Dybvig! How stupid of me! Diamond-Dybvig!”
Sorry, Paul. But the notion that Diamond-Dybvig explains the GFC is fundamentally wrong.
The Diamond-Dybvig model is a classic model of bank runs. Banks accept deposits that the depositors can demand at any time. They lend some of this money to firms and individuals who promise to pay back the loans over time (but not on demand). In the normal course of things, only a few depositors want their money at any given time. Given that the bank will be able to pay on demand there is no particular reason for depositors to worry about getting their money back. That's the benign equilibrium. The problem is that there are multiple equilibria. In particular, there is a run equilibrium wherein for whatever reason too many people want their money back at the same time. Given that the bank may not be able to pay those at the back of the line, it makes sense for every depositor to try to get their money back before the bank runs out of money. The basic model of banking is therefore inherently exposed to such runs and is the reason why we need deposit insurance.
That's simply not what's happening in the GFC. The GFC was a systemic banking crisis; not a classic bank run. The Diamond-Dybvig model features a run on a single bank. Within that model, there is no mechanism for the run to spread to other banks, much less to engulf the whole system. Moreover, with few exceptions, depositors did not run on the banks during the GFC. Indeed, the crisis was centered not on traditional deposit-funded banks but on wholesale-funded investment banks. Furthermore, the banks at the center of the crisis did not originate loans and hold them on their own balance sheets (the traditional 'originate-to-hold' model). Instead, they sold these assets to investors willing to bear the risk (the modern 'originate-to-distribute' model).
One may claim that funding runs are similar enough to Diamond-Dybvig type depositor runs for the model to apply. One would be wrong. This is because wholesale funding is secured by collateral. If the borrower cannot pay, the lender can get her money back by selling the collateral. The secured lender therefore does not face the same incentives as a Diamond-Dybvig type depositor in an uninsured bank, who has no choice but to line up to get her money back if others are doing the same.
An entirely different mechanism generates runs in secured funding markets. Here one must distinguish between bilateral repo and tri-party repo. In the former, cash investors must take counterparty risk into consideration as well as the quality of the collateral since it can become problematic to secure the promised collateral in the event of a messy bankruptcy. In tri-party repo markets on the other hand, cash investors can be sure of getting their hands on the collateral (since it is in administrative custody of the third party and placed in a sort of escrow account that automatically delivers the collateral even if the third party gets in trouble) and the only thing that matters is the perceived quality of the collateral. More precisely, what matters is the certainty with which one can exchange the collateral for cash at par. So runs can obtain in bilateral repo markets if either borrowers or collateral are perceived as dicey. In tri-party repo markets on the other hand, runs can only obtain if asset classes used as collateral are no longer perceived as safe.
A run in the wholesale funding market generates firesales of assets which further intensifies the run in the funding market and so on. This is exactly the vicious doom-loop we observed in 2008. The reason why it is important to distinguish between the bilateral and tri-party repo markets is because it allows us to identify the causal mechanism behind the GFC.
During the GFC, there was a generalized run in bilateral repo markets as perceived counterparty risk spiked. Basically, cash investors lost confidence in dealers (ie, Wall St banks) and dealers lost confidence in each other as credit defaults mounted. No one could be sure who was hiding what on and off their balance sheets. In tri-party repo markets on the other hand, runs were confined to private-label residential mortgage-backed securities (RMBS). Even during the brutal week of the Lehman bankruptcy, the core of the flywheel continued to spin as cash investors lent trillions of dollars to dealers overnight against Tbills (obligations of the US Treasury) and agency RMBS (obligations of Fannie and Freddie assumed to be backed by the US government).
What this means is that the instability of the wholesale funding flywheel was due to the introduction of private-label RMBS as collateral. In other words, the GFC would not have acquired the virulence that it did had collateral remained restricted to state-backed assets. And the reason why private-label RMBS had to be used as collateral was the shortage of public safe assets. For it was the demand for safe assets emanating from the wholesale funding market that prompted the dealers to manufacture private-label RMBS at such a large scale using residential mortgages as raw material. That's what caused the subprime lending boom. It was not the Fed's low rates; not the result of declining standards at Fannie and Freddie; it was not a classic credit boom à la Reinhart and Rogoff; nor was it due to the global savings glut à la Bernanke. And the GFC itself was certainly not a classic Diamond-Dybvig type bank run as Krugman would have it. No, the financial boom was instead the great sucking sound of the wholesale funding market whose denouement was fittingly the seizure of this very market and of the Western financial system built around it by Wall Street.