Asset Prices, the Business Cycle, and Unemployment

A few months ago, I blogged about research done by Roger Farmer and Naryana Kocherlakota on understanding unemployment; and in particular relating trends in unemployment to asset markets and investor confidence.

The key issue is that it's difficult to understand why unemployment remains so high so far after the financial crisis. A real business cycle approach would look for real shocks to production; in particular idiosyncratic shocks to technology. Yet it appears that this recession involves a number of nominal and financial-sector related frictions difficult to rationalize using that model.

Another approach relies on New Keynesian thinking. In this view, prices are “sticky” as it is costly to adjust prices in the short-run in response to a moderate shock. This induces a friction in in economic activity, especially in the market for labor, that can be fixed through macroeconomic stabilization in the form of monetary or fiscal stimulus. However, if you look at scanner data, retail prices are actually fairly flexible. On top of that, the economy faced such a substantial shock, and we are sufficiently far out in the future, that surely price-setters have had the chance to adjust prices by now.

In short: the particular frictions and shocks underlying traditional macroeconomic models seem to be of limited relevance in explaining this recession. As a result, I’ve been trying to read up on alternate models — relying, for instance, on financial market frictions, household balance sheets, etc. Roger Farmer, in a new paper, offers up another such model that relies on old-style Keynesian thinking.

The basic logic comes out in this graph:

This shows asset prices and unemployment being closely linked throughout the business cycle; which isn’t a trivial fact. The logic is that asset prices follow bubbles and crashes due to self-fulfilling optimism and pessimism from investors. These booms and crashes result in larger social consequences in the form of higher unemployment through search frictions in the labor market.

In this model — as in some sense as in Keynes original work — investors do have rational expectations. They expect a bubble; and a bubble happens. They expect an economy that performs poorly over an extended period of time; and that too materializes.

Nevertheless, the paper still relies heavily on psychological assumptions about investor behavior. Investor confidence drives both asset prices, as well as willingness to hire. This response is, as I suggested in my last post, asymmetrical, as it is harder to hire than it is to fire.

I find this paper interesting as it brings the problem of unemployment into the domain of asset pricing. While Farmer emphasizes the psychological basis of bubbles and busts, someone like John Cochrane would emphasize how ultimately discount rates — the rates at which we value future income relative to current income — determine the values of current financial assets (which are just claims to future cash flows).

In Cochrane’s world — asset prices fluctuate in conjunction with macroeconomic outcomes. Is the economy looking bad? Do you anticipate losing your job, your business, or other such negative shock? If so, you wish to hold less risky assets. Yet we can’t all rebalance away from risk, as there are only so many stocks and bonds and so forth. Instead, the price of stocks and bonds fall in order to compensate us for bearing this sort of risk in a time of economic uncertainty.

Yet Farmer would point out that the process also works in reverse. High discount rates — equivalently, “pessimism” — felt by the owners of capital manifest themselves in an unwillingness to hire, particularly if there are frictions in the labor search process.

I suspect that tying Farmer’s model into a more "rational" model of asset prices that emphasizes the links between financial markets and real markets would amplify the multiple equilibrium nature of unemployment. This was the key feature of Keynes of course, and it’s interesting to see Farmer resuscitate this idea. And with unemployment at 9% or what have you it doesn’t seem implausible to think that numerous economic possibilities are open to us, depending on the nature of economic equilibria we end up at.

Bubbles and Unemployment

There’s a lot of commentary going around on why unemployment has proved to be persistently high during the recovery. As Yglesias notes, this boils down to the question “Are recessions caused by asset price busts fundamentally different from recessions caused by central bank efforts to curb inflation?” Paul Krugman has a strong take on this:

Brad DeLong has recently written up a clearer version of a story I’ve been telling for a while (actually since before the 2008 crisis) — namely, that there’s a big difference between inflation-fighting recessions, in which the Fed squeezes to bring inflation down, then relaxes — and recessions brought on by overstretch in debt and investment. The former tend to be V-shaped, with a rapid recovery once the Fed relents; the latter tend to be slow, because it’s much harder to push private spending higher than to stop holding it down.

The idea that the precise conditions of this recession are different has implications for the favored policies of both the right and left. On the left, some folks believe that the notion of balance sheet recessions calls for more measures to tackle household negative equity, optimism for fiscal policy, and skepticism regarding monetary policy (say, Krugman). On the right, other people emphasize mismatch problems in the labor market and the role of structural forces behind unemployment. They are frequently skeptical of how fiscal policy can fix these problems (sometimes, also monetary policy). In general, there are widespread beliefs that some particular features of the crash have limited the scope for traditional macroeconomic stabilization policies.

Via Stephen Williamson, Minneapolis Fed President Narayana Kocherlakota has a new paper that goes into this issue. He draws on the Keynesian work of Roger Farmer, who shows this graph:



Farmer’s idea is that the rate of unemployment at any time is indeterminate due to problems in the labor search market. In the absence of markets for the search time of workers, price signals are not necessarily sent to match workers with the right jobs. Instead, the level of unemployment is determined by expectations of the strength of economic activity, which is proxied by stock market performance. One problem for this idea lies in explaining why unemployment has been slow to recover even as the stock market has recovered. Farmer writes,

This paradigm provides us with a new way to think about large recessions like the Great Depression and the Great Recession of 2007—2009. Using the model from this paper I would argue that the world economy in 2008 was headed rapidly towards a high unemployment, low wealth, equilibrium. The move to this bad equilibrium was triggered by a loss of confidence in the value of assets, backed by mortgages in the US subprime mortgage market. The inability to value these assets led to an amplification of the crisis as panic hit the global financial markets.

In the winter of 2011, the US labor market had still not recovered. I believe that much of the problem is connected with a lack of confidence bylobal investors who are concerned with the possibility of a further collapse. Even though the US stock market may be appropriately valued based on historical price earnings ratios — market participants are concerned that the value of stocks could fall further. Variations in the level of confidence are manifested in changing risk premia that are fully rational given the unpredictable behavior of future traders in the asset markets.

I find this argument more persuasive in explaining the employment dowturn than the failure of employment to recover; but it is easy to imagine alternate models in which employment growth is asymmetric with respect to the business cycle.

Kocherlakota's innovation is to bring this unemployment picture into a broader model involving bubbles and monetary policy. While his model is fairly complex, the end result is simple — as with Farmer’s model, the level of unemployment is ultimately determined not by prevailing wages, but rather by the amount of aggregate demand. The collapse of an asset bubble results in a substantial drop in demand, and will result in a hike in unemployment unless the central bank proves sufficiently accommodative in lowering nominal interest rates.

One way to think about this is to compare the stock market bust in 2000 with the housing bust in 2008. In both cases, you have an asset that drops dramatically in value (tech companies, housing) that results in large drops in comparable financial securities (tech stocks, mortgage-backed securities). The total wealth loss in the economy was roughly comparable between the two cases. Yet for the 2000 crash, the Fed was able to lean against the drop by moving conventional monetary policy enough. In the second case, the Fed quickly hit the short-term nominal interest rate barrier of 0, and was unable to ease further though conventional channels. It did pursue unconventional policies like QE2, but was hesitant to do so and faced unprecedented levels of backlash for the easing that they did happen.

So, one way to read Kocherlakota is the following: given that monetary policy-induced demand fixes the rate of unemployment, recovery difficulties now reduce to the fact that the Fed has been insufficiently accommodating as the some interest rates hit the zero-rate bound. If the Fed instead proved more willing to consider unorthodox policies like quantitative easing or currency depreciation, we could have whatever degree of unemployment we liked.

The ultimate origin of a recession, in this model, is completely irrelevant to the possibility for the recovery. Issues with debt overhangs, structural unemployment, etc. are all second order effects relative to the fact that Fed-induced nominal spending has lagged; and the optimal recipe is not fiscal stimulus, but further Fed-based easing. Recessions caused by asset-bubble bursts do not differ from the garden variety recession, as long as the Fed is in fact appropriately accommodating.

So, the real issue is not that bubble-induced recessions are diffrerent in some way, but rather that policymakers respond to them differently. Rather than saying “the recovery in the 80s was quick because it was a Fed induced recession”; the issue instead is that the Fed had more scope to tackle that recession than this one.