Friedman (1993) proposed a model of the depth of recessions and steepness of recoveries built on two empirical regularities:
- output is on average below a ceiling defined by supply capacity and tends back to this ceiling; and
- large contractions are followed by large expansions and mild contractions are followed by mild expansions.
The strength of a recovery should be positively correlated with depth of the recession but there should be no correlation between expansions and recessions (Friedman 1993; Alchian 1969).
Figure 1 illustrates Friedman’s model, which likens the time path of output to a string on the underside of an upward sloping board that is plucked downward at random intervals to various extents into busts that are followed by booms.
Source: Garrison (1996).
The upward sloping board plotted as a thick line in Figure 6 represents a ceiling on feasible output and employment in a given year that is set by resource and technology availabilities. The upward slope of this board accounts for trend real GDP growth over time due to technological progress and other factors.
Output is close to the ceiling shown in Figure 6 except for every now and then when it is plucked downwards by a monetary contraction.
There is no floor on these contractions in output to moderate the depth and violence of contractions, so some recessions are deep and sharp (Hansen and Prescott 2005; Friedman 1993; Goodwin and Sweeney 1993; Sichel 1993).
Contraction depth can vary greatly as is shown by the minor, mild and deep recessions in Figure 1. In each episode of plucking illustrated in Friedman’s model in Figure 1, the rebound mirrors the previous fall in output, but the recovery cannot go beyond the ceiling.
Friedman’s model is a bust-boom model of business cycle fluctuations. The business cycle starts with a bust caused by an adverse policy or other shock and is then followed by a boom as the market self-adjusts and the policy errors are reversed.
Without the initial adverse policy or other shock, there would neither be a bust nor a boom. The economy would track close to the ceiling on output and employment as is shown in Figure 1 by the periods between the plucks.
The correlation between busts and booms arises from the monetary contraction that caused the bust eventually inducing an offsetting correction in monetary policy.
The monetary contraction that pushed or plucked output below the upward sloping ceiling is later followed by a monetary expansion that offset the earlier contraction.
With the amplitude of monetary expansions correlated to offset the prior contractions, GDP growth will have similar plucks or falls and rebounds to the upward sloping output ceiling because of the link albeit with a lag between monetary growth and output fluctuations.
The increases and decreases in monetary growth are independent policy choices with unique causes.
The associated upward and downward movements in GDP growth are not correlated with each other but should be correlated with the prior fluctuations in monetary growth.
There would not be a bust and later boom if there is no monetary contraction to start the cycle. This is why Friedman (1993) proposed that the depths of busts are unrelated to duration and strength of prior economic booms. This upset Austrians such as Roger Garrison:
…Austrians work at a lower level of aggregation in order to allow for the outputs of the two sectors to move relative to one another and even to allow for differential movements within the investment-goods sector…
During a credit-induced boom, investment in the relatively high stages of production is excessive in that resources are drawn away (by an artificially low rate of interest) from the relatively low stages of production and from the final stage, consumption.
The decrease in the amount of resources allocated to the low and final stages is forced saving; the misallocation of resources from low to high stages is malinvestment.
Empirically, a credit-induced boom would be but weakly reflected in the conventional investment aggregate and hardly at all in the Monetarists’ output aggregate, which includes consumption.
The boom for the Austrians refers to something going on largely within the output aggregate.
It is represented in Friedman’s plucking model not by a conspicuous recovery to trend but rather by some period preceding a pluck which Friedman, operating at a higher level of aggregation, presumes to be healthy growth.
The publication of Milton Friedman’s paper in 1993, which recalled an obscure paper he wrote in 1964, lead to a large literature blossoming under the heading business cycle asymmetry.
I find it surprising that so many concentrate on rational expectations when discussing Austrian business cycle theory (ABCT). Is Austrian business cycle theory the only modern business cycle theory that must reach such a high bar?
Many modern business cycle theories build on information costs and learning and explain that people make forecasting errors because of noisy information, and repeated monetary shocks keeping up this confusion.
A good general explanation of misperceptions theories of business cycle is in Alchian and Allen (1967), which Murray Rothbard called a brilliant textbook. The business cycle is not based on money illusion or on systematic mistakes.
People take time to acquire the necessary information to interpret what has shocked the economy and what these changes mean for them. Additional shocks complicate this learning so there are more errors and confusion continues to affect market choices. Learning is neither instantaneous nor is the requisite information free to collate. People must make do with the incomplete knowledge they have and make choices about market signals that might be spurious or be meaningful signs of change.
Mises, Hayek, and Rothbard all noted in the collection edited by Garrison, for example, that a one-shot monetary shock would be soon uncovered by entrepreneurs, the malinvestments quickly reversed, and the boom would bust. Monetary shock after monetary shock require repeated entrepreneurial revisions and it will take a long time for entrepreneurs to catch up. This is also in Alchian and Allen.
Rothbard (MES pp. 1002-1005) discusses one-time versus repeated and increasingly large in size monetary shocks as the basis for booms and the reasons for the on-going deception of entrepreneurs. The shocks must increase in size to keep injecting more unanticipated noise into monetary and entrepreneurial calculations.
ABCT proposes a more complicated signal extraction problem than in say the Lucas-Phelps islands model. Dispersed and slowly unfolding information must be produced as each new monetary shock ripples its own unique way across the economy, passing through different hands each time. Only slowly does the requisite knowledge about the relative prices effects of each new monetary shock emerge as the result of market interactions and become open to entrepreneurial discovery.
What is perhaps dismissed too easily by Rothbard (but not Mises) is that under a gold standard, increases in the output of gold mining can be well forecasted by entrepreneurs. Rothbard’s best ground is when he notes that "the credit expansion tampered with all their [entrepreneurs’] moorings." A stop-go monetary policy is by definition unpredictable. Gold output fluctuations are irregular but usually small. A unique contribution of ABCT is that the longer the boom, the deeper the bust.