Evan Soltas
May 13, 2012

Schumpeter Was Wrong

Nick Rowe and I had an interesting discussion in the comments section of my layman's guide to NGDP targeting post in which I more or less accidentally came up with an argument for why the famous economist Joseph Schumpeter was wrong about recessions.

Schumpeter famously said that recessions were purifying events, in which the "perennial gale of creative destruction" frees up resources -- capital and labor -- for new and more efficient application:

Our model presents features that seem to differ from widely accepted...opinion. It does not give to prosperity and recession...the welfare connotations which public opinion [typically] attaches to them. Commonly, prosperity is associated with social well-being, and recession with a falling standard of life. In our picture, they are not, and there is even an implication to the contrary.
The problem with Schumpeter's theory of the purifying recession is that it begs the question. Schumpeter's logic is that recessions are punctuated moments of creative destruction, and creative destruction is good, and therefore recessions are good. This statement is only true if one presumes that recessions are such creative-destruction episodes.

Yet the evidence for that contention is rather weak. The rate of formation of startup firms and hiring by startups is empirically independent of the business cycle, and such dependence, if real, would have been arguably the essence of Schumpeter's argument. That we don't see any relationship suggests that creative destruction happens during both the macroeconomic boom and recession.

Nevertheless, Schumpeter's view is still seriously entertained in public opinion, for example in this article in the Boston Globe:

[T]o see recessions in such exclusively negative terms suggests a misunderstanding of what a recession is. According to a classic economic view, recessions can play a painful but healthy role in the cycles of the economy. They wring inefficiencies, imbalances, and dangerous levels of risk out of the market.

Today, few mainstream economists share Schumpeter's belief in the unerring regularity of business cycles. And some point out that there's plenty of room for creative destruction when the economy is strong - witness the turmoil over the past few years in the music industry, or the airlines, or, for that matter, the newspaper industry.

Economists do agree, however, that recessions help to right economies that have lost touch with reality. Recessions not only cull unhealthy companies, they expose financial gimmickry. They punish groundless optimism and the rampant speculation it feeds - in fanciful Internet ventures in the 1990s, for example, or in housing over the past few years.

Recessions do not look like the creative-destruction episodes to me. Rather, they appear to wash away both good and bad firms -- they are not selective, or at least no more selective than during macroeconomic stability. In fact, there is reason to think that macroeconomic volatility would make the creative-destruction process substantially less efficient. Here's why -- this is the refined version of the argument I posed to Rowe.

Imagine that there are two pools of firms, the first of which are inactive firms who have not been assigned an industry, the second are active firms operating in a single industry. Firms move from the first to the second, or the second to the first, by means of formation and destruction respectively.

When a firm is formed, it is assigned a level of revenue which is determined by a random number and its industry. On top of this "structural revenue," there is a cyclical component. Costs are constant. When costs exceed revenues for one period, that firm folds, reentering the inactive first pool and "waiting" for a reassignment to a new industry.

Firms decide to form in particular industries based on which industries they think are growing or contracting. They gather this information through a signal extraction problem similar to Lucas (1972), in which

[i]nformation on the current state of these real and monetary disturbances is transmitted to agents only through prices in the market where each agent happens to be. In the particular framework presented below, prices convey this information only imperfectly, forcing agents to hedge on whether a particular price movement results from a relative demand shift or a nominal (monetary) one.
The signals come from the formation or destruction of other firms, which is public information analogous to prices in Lucas' model. When firms move in between these two pools, they send signals, with firm creation in industry A sending a signal +1 for industry A, and firm failure in industry A sending a signal +1 for industry A. These signals are public data, which firms in the first inactive pool must interpret as either firm creation/failure induced by macroeconomic conditions or by industry-level growth or contraction.

Macroeconomic volatility makes the information firm formation/destruction imperfect by introducing "bad signal" -- serially correlated error, with additional random noise. Signal extraction becomes increasingly challenging with a higher amount of macroeconomic volatility relative to the valuable information of secular growth or contraction in any particular industry. Signal extraction is not impossible under such conditions, just more difficult and likely to result in incorrect decisions.

If you don't believe the idea that firms in pool A have imperfect information, try it yourself. Here's a simple case: the level of employment and the number of firms in the manufacturing industry have been in secular contraction in the United States for decades. However, since 2008, there has been unusual employment growth in manufacturing. To what extent does this reflect cyclical conditions, given that the 2008 recession was very deep and eliminated a large fraction of manufacturing jobs in the United States? To what extent, conversely, does it represent a secular recovery in that industry? The fact that any answer would contain a significant amount of uncertainty means that firms in the first pool face a signal extraction problem under conditions of imperfect information.

By this logic, which embraces as did Schumpeter the central importance of creative destruction, recessions do not assist the process. In fact, by corrupting the signal by which firms decide to enter or exit industries, recessions make the creative destruction process slower, less efficient, and more costly.