Why Do "Efficient Markets" Economists Think a Speech by George W. Bush Caused the 2008 Financial Meltdown?
Where Is My Chinese Consumer Price Inflation?

Leverage and Its Discontents

Apropos of http://delong.typepad.com/20091222-delong-picking-up-nickels.pdf:

Joe Gagnon writes:

Monetary Policy and Asset Bubbles in 2010: DeLong's model cleverly captures many key features of an asset bubble and yet remains simple enough to draw clear conclusions. He shows that overconfidence about being able to get out before the bubble bursts and expectations of a bailout make bubbles costly. The cost of bursting bubbles discourages the central bank from lowering the interest rate as much as it needs to stabilize the economy. These results reflect the dilemma that many observers currently see in monetary policy--how to walk the fine line between easing too little to fight unemployment and easing too much to cause a new and harmful bubble. The main drawback of DeLong's model is that it overstates the welfare costs of bursting bubbles in two ways. First, it assumes that all declines in long-term asset prices are costly even if they are not associated with bubble-like behavior. Second, it ignores the role of unleveraged or partially leveraged investors. Modifying his model to correct these shortcomings leads to a very different conclusion for economic policy. Bursting bubbles are economically harmful only to the extent that they bankrupt investors or raise fears that investors will go bankrupt....

In DeLong's model, the welfare cost of a bursting bubble is related to the square of the losses faced by investors. This welfare cost, however, implicitly assumes that investors are fully leveraged and thus are forced to default on their short-term loans whenever long-term asset prices fall, even when there is no bubble. This feature of the model is clearly unrealistic. The way to reduce the cost of a bursting bubble is to reduce leverage. In a world of unleveraged (or lightly leveraged) investors, falling asset prices would not bankrupt anyone and thus would not raise fears of bankruptcy. In such a world, there are no welfare costs of a bursting bubble.... [T]he technology bubble of 2000 burst with no apparent ill effects because it was not leveraged to any significant extent and there were no government bailouts....

As I have argued elsewhere, monetary policy actions to support and even increase the prices of long-term assets are warranted now to speed economic recovery and avoid deflation.... The global financial crisis demonstrates the need for reforms to greatly reduce the leverage of financial institutions and to make that leverage respond to the credit cycle in a stabilizing manner.... The bottom line is that regulators need to be vigilant in maintaining the process of deleveraging... preventing any new buildup of leveraged asset purchases... greatly reduce the degree of leverage in our financial system and it may be a good idea to make leverage respond inversely to asset prices and to put stabilizing mechanisms in the tax system...

In my model the economy's degree of leverage--the vulnerability of employment to an asset price crash--is governed by the parameter α. Gagnon says: do everything you can to reduce α, and then don't worry that employment-aiding monetary policy might produce bubbles because bubbles don't matter. He has a powerful point. In my model the optimal amount of monetary accomodation is:


If you can make α small, then indeed the desirable amount of monetary accomodation is large--and our social welfare function also goes with 1/α as well.

The problem is that leverage is in our financial system for a reason. Equity holders have to (or ought to) spend their time watching and overseeing management. Debt holders can simply clip their coupons and rouse themselves to spend time, energy, and money on corporate control when there is a default. It's not efficient to force everybody to monitor and oversee their investments: better to take advantage of economies of scale and have those who are best at the business of corporate control have risky equity and let the rest not have to dissipate their resources and have debt claims instead. That is leverage. Moreover, different agents and organizations differ greatly in their taste for risk, and it is plain that there are many in asset markets who greatly value (I would say overvalue) relative risklessness in their asset portfolio. To redistribute risk to the risk-tolerant and away from the risk-averse is another source of leverage.

So low α comes at a cost: too many resources will be devoted to oversight and corporate control, and too many risk-averse people will wind up bearings risks they don't want and greatly fear. How big are these costs of structuring institutions to lower α? I don't have a clue. If I worked for the Treasury or the Fed I would probably now have 48 hours to come up with a number and a justification for it. But fortunately I do not.