Saturday, October 8, 2011


Matt Rognile takes aim at people claiming “deleveraging” is an important reason behind the prolonged downturn:

In failing to understand this core logic, most commentary about “deleveraging” is rather bizarre. At some level, it’s the same cluelessness that we once saw from central planners: they’d trip over themselves in the complexity of fixing a shortage in one market or a glut in another, never quite realizing that the price mechanism would do their work for them. Right now, historically low inflation expectations and below-potential output are prima facie evidence that real interest rates are too high. That’s what every macro model tells us is associated with contractionary policy by the Fed. Yet we see pundits lost in all kinds of complicated, small-bore proposals to stimulate the economy—when the fundamental, overriding dilemma is getting the price (in this case, the interest rate) right.

Elsewhere, he argues that many households, despite the fall in home prices, do have substantial assets. 

The deleverage hypothesis argues that aggregate GDP will remain weak as long as household consumption is held back through the presence of debts. Rognile retorts that the balance of savings and investment is adjusted through interest rates; and that the changes in net assets can’t really support economically meaningful drops in consumption corresponding to the output declines we’ve seen. 
I think there’s a lot true here. David Beckworth has also made  a number of powerful arguments connecting deleverage to monetary policy:
For every household debtor deleveraging there is a creditor getting more payments.  Yes, household debtors have cut back on spending, but so have creditors.  The creditors could in principle provide an increase in spending to offset the decrease in  debtors' spending.  They aren't and thus the economic recovery is stalled. In other words, the problem is as much or more about the build up of liquid assets by creditors as it is the deleveraging of debtors… 
The key problem is that there are households, firms, and financial institutions who are sitting on an unusually large share of money and money-like assets and continue to add to them.  This elevated demand for such assets keeps aggregate demand low and, in turn, keeps the entire term structure of neutral interest rates depressed too.

I think Beckworth actually makes a more powerful case than Rognile. Even if households had good reason to save more; it’s not obvious that lower household consumption would necessarily lead to lower aggregate output, so long as the banks and creditors receiving those payments went out and lent the money. Then the usual money multiplier arguments would ensure a rapid circulation of credit throughout the economy, raising output. After all, we typically think that higher saving and investment is a good thing for economies; as opposed to thinking that money saved is wasted. 

There are two reasons that isn’t going on, relating to an excess demand for money: households desperately want to hold more liquid assets and hold fewer debt commitments; while creditors also are inclined to hold liquid assets rather than lend those out. Logically, the only way that a flow of money from households to creditors can have any aggregate effects is if agents in an economy simultaneously have an excess demand for money not met by the central bank. 

When you think about it this way, it becomes easier to diagram how to think about deleveraging. There are basically three categories I see:

1) People who think that deleveraging is real, and no amount of monetary stimulus will help. 

These are people like Richard Koo. Their argument goes that the presence of excess debt is the key constraint holding back economic growth. No amount of monetary stimulus will fundamentally change the asset position of households, and so there’s no way it will alter consumption or output (or, at least, not to the degree that is necessary). Raghuram Rajan may believe something like this, as best as I can tell. The MMT folks are probably best placed here as well. 

As Beckworth and Rognile point out above, this view doesn’t make sense given the conventional understanding of how monetary policy ought to operate. If we desire greater spending from households or creditors; we can always make that happen by flooding the system with money. 

2) People who think that deleveraging is real, monetary stimulus could help, but the Fed won’t deliver enough. 

These are people like Paul Krugman. As Rognile points out — Krugman is careful to note how deleverage is only an issue if you’re in a liquidity trap, but that nuance tends to be lost among many other commentators. Elsewhere, he has argued that fiscal stimulus is only worthwhile as long as interest rates are zero — at other times, he often takes for granted that monetary policy ought to handle the brunt of aggregate demand management (or at least he did in the '90s). 

In that sense, Krugman actually agrees with Scott Sumner on more issues of intellectual substance than, say, with Keynes. It’s just that Krugman believes that in this particular instance, we happen to be in some kind of liquidity trap in which monetary policy won’t be sufficient to tackle the headwinds of a deleverage cycle. 

3) Then; there are people who believe that deleveraging may be a concern; but monetary policy (even with a zero-rate bound) ought to handle everything.

Here are the market monetarists like Scott Sumner and David Beckworth, as well as Matt Rognile. The belief is not only that monetary policy can fix any conceivable deleverage shock; but that the Fed could do so tomorrow given the set of tools they have; involving perhaps the adoption of a price level, getting more QE, imposing interest on reserves, or offering guidance on the future path of interest rates. 

Many people on sides 1 and 2 agree on issues; but there’s a fundamental conceptual difference there. Suppose, as Mike Konczal likes to imagine, that we wake up tomorrow and find that interest rates are actually 2%, rather than at 0% for the short-term Treasury rate. What should we do? Some people (like perhaps Koo?) would argue that changing that rate wouldn’t do very much. But people like Krugman would argue that, if we were in an environment in which conventional monetary policy could operate, then that’s basically the only policy channel we should use to get output back.

The only difference between sides 2 and 3 is whether or not the liquidity trap proves binding. This seems like a fairly trivial issue; but it determines entirely whether or not you think that we should adopt fiscal stimulus, or simply Ben Bernanke with a more aggressive Fed Chair. 

Where’s the Evidence for Deleveraging?

Some of the best evidence in favor of a deleverage model comes from cross-sectional cuts comparing debt to economic outcomes. For instance, Mian and Sufi find that high household debt areas have lower employment than low household debt areas:

One way to think about this is in some kind of Bernanke-Gertler approach in which households use their homes as collateral. When home prices were increasing, households used their houses like credit cards and extracted some of the equity. Now, when debt levels are high, the same households are cutting back; and employment is suffering. (Philippon and Midrigan have a paper highlighting this channel, though they also emphasize the role of monetary policy to counteract that)

The issue with this type of finding is ensuring identification. It seems intuitive that the parts of the country which participated in the housing boom most heavily would have some of the worst outcomes right now. But what’s the channel by which that operates? If it’s the debt, than we have some possible fixes — do some mass refinancing or mass principal writedowns (which may be good ideas themselves for other reasons).

But I don’t think it’s obvious that demand-side issues are at work in explaining the poor economic outcomes of post-real estate bust areas. Erik Hurst instead points to supply-side effects coming from the structural challenges in re-orienting a local economy away from real estate investment. He points to the following graph:

Here, he shows that the change in unemployment rate mirrors closely the change in the composition of output away from real estate-financial sectors. You have laid off construction workers, for instance, who find it difficult to retrain and find new jobs. Lowering outstanding mortgage principal won’t necessarily help retrain a laid-off-construction worker as a nurse. In fact, lowering the principal on a mortgage might induce that laid-off-worker to remain in a housing bust area instead of moving to North Dakota, where the unemployment rate might as well be zero; raising unemployment (this is basically what Lee Ohanian and Kyle Herkenhoff argue). The economy may just be in for a sustained period of slowdown as individual agents attempt to find a new sustainable equilibrium. 

I’m not wed to either the demand/debt or construction/supply approach — I just want to point out it’s not obvious to think about the role of debt, or even the relative ratio of supply side and demand side issues in explaining a weak economy. No one has credible causal estimates of how lowering debt burdens would help household or economy-wide welfare. Household and bank-centered approaches are very appealing in trying to explain why the recovery has been as weak as it is, but I don’t think all of the stories hang together.