A friend sent along a paper by Robert B. Lester and Jonathan S. Wolff titled “The empirical relevance of the Mises-Hayek theory of the trade cycle,” published in the Austrian Review of Economics in 2013. The paper finds, empirically, that the theory is not relevant. It can be downloaded here. As an applied Austrian business cycle theorist, I will here briefly give my comments to what I see are flaws and shortcomings of the study.
The abstract outlines the research and findings as follows:
Austrian Business Cycle Theory (ABCT), as espoused by Mises (1953, 1949) and Hayek (1935), predicts changes in the economy’s structure of production following an unexpected change in monetary policy. In particular, following a credit expansion the theory predicts that: previously idle resources are drawn into the market, previously employed resources are used more intensively, and prices and quantities of goods in the intermediate stages of production decline relative to the prices and quantities of goods in other stages. To test the theory’s implications we employ stage of process data which classify goods by their distance to final consumption. Using this data we run structural vector autoregressions and isolate each variable’s response to a monetary shock. Consistent with the theory, we find that resource use expands on the intensive and extensive margin. On the other hand, we find little evidence of the relative price and quantity effects predicted by ABCT. Since the relative price effects are the distinguishing aspect of ABCT, we conclude that evidence in favor of the theory is, at best, mixed.
In short: the empirical study performed by these economist-mathematicians shows that the intensity of economic activity does pick up following a monetary intervention by a central bank (as ABCT predicts), but that the relative price movements between lower order and higher order sectors does not always manifest as predicted by ABCT. They correctly point out that this latter price dynamic is a distinguishing feature of ABCT compared to other schools of economic thought, which is what I will address here briefly.
Shortcomings of Data Proxies
To empirically test the “relevance of ABCT”, Lester/Wolff uses US Bureau of Labor Statistics (BLS) PPI data by stage-of-process and stage-of-process industrial production data from the Federal Reserve Board, which, I agree are better proxies of economic activity and relative prices in the production structure than the aggregated output, consumption, and investment data released quarterly.
I say these are still only proxies for economic activity and actual market prices, as no statistical measure is 100% accurate. For example, the PPI selects a sample of companies in the US economy to be part of the data programme voluntarily that is representative of the overall economy. Items are selected per company using the disaggregation method (as opposed to other potential methods), whereafter all the data are thrown into various buckets, and processed. It can never be a perfect representation of what prices are doing in the general economy, but rather a proxy. You are therefore testing a theory empirically using, at best, a proxy of producer prices.
Another shortcoming of the PPI and IP data is they only capture information from a company or sector once it has been in existence for a while, which will hide the activity happening in these sectors during the early stages of a business cycle upswing. As these industries are in the process of forming, because statisticians are unable to capture the data instantly, there will be holes in the historical data.
Shortcomings of Monetary Policy Variables
Lester/Wolff set up the study to test the impact of movements in the Federal Funds Rate, the monetary base, and M1 and M2 money supply on PPI ratios that rise when those further away from the consumer sector rise relative to those nearer (PPI more remote from consumer [divided by] PPI nearer to consumer). There are four of these PPI ratios.
They find that there is a stronger link between manipulations of the FFR to PPI ratios (3/4 rise in their tests), but that the significance thereof declines after 10 periods. As you move down to broader measures of money from monetary base to M2 money supply, the link between money growth and PPI ratios decline to 1/4, with no significance after 10 periods.
It is instructive that distortions of the FFR provide the most significant response in favour of ABCT, as it is the divergence between this interest rate and the natural rate of interest that sets in motion the business cycle, according to the Mises-Hayek theory.
Looking at the next test variable, the monetary base, Murray N. Rothbard showed (The Mystery of Banking (1983) that a rise in the monetary base does not automatically equal a rise in money supply, if it is on deposit at the Federal Reserve, and hence cannot be expected to impact the economy. It is likely to provide mixed results, varying depending on the historical data of the specific case.
As we move out to broader money M1 and M2 money supply, we are now measuring the consequences of credit injected following the Federal Reserve’s fixing of the FFR below the natural market rate. Not only is M2 rejected by Austrian economists as a representative measure of actual money supply in the economy (see Salerno, 1987), but it is also a consequence of rising fiat/fractional credit growth. Artificially lowering the FFR will result in fiat/fractional credit creation in a sector at a specific temporal location in the production structure that, in turn, becomes deposits in the banking system and only then, becomes broad M1 and/or M2 money supply. This is an important shortcoming of using M1 and M2 in empirical tests of ABCT (or even TMS for that matter).
As Jesus Huerta de Soto explained in “Money, Bank Credit, and Economic Cycles (1998) (see pp. 406-408)”, we would need to revise the predicted consequences of ABCT if:
“a significant portion of credit expansion is devoted (contrary to the usual practice) to financing not durable consumer goods, but the current consumption of each financial year (in the form of goods and services which directly satisfy human needs and are exhausted in the course of the period in question)….
In this case, the new money injected into the consumer sectors immediately pushes up the prices of consumer goods and diminishes, in relative terms, the prices of the factors of production. “Thus a trend toward the flattening of the productive structure is established without a prior expansionary boom in the stages furthest from consumption (de Soto).”
Therefore, movements of PPI ratios will move counter to what standard ABCT ‘predicts’ if there is a large amount of consumer credit flowing into the economy. Not only does a distinction need to be made between credit used to finance durable versus consumption goods, but also on where credit flows into the economic structure. For example, if government is providing subsidies and other incentives to a sector nearer the consumer level, credit could first flow into that sector and push prices higher there first, before filtering through to the general economy and setting in motion a business cycle boom as ABCT would predict. Under such a scenario one could expect some of these PPI ratios to fall initially, before they begin to rise at a later stage.
The initial credit injection is specific and targeted, while its natural consequence – broad money – is broad and general. Whether rising PPI ratios or falling PPI ratios (as used in the Lester/Wolff study) should manifest during a given boom depends on the historical data of where credit is injected into the economic structure.
This is a paper that attempts to test the empirical relevance of ABCT that not only uses general proxies (PPI and IP) and very general economic data (monetary base, M1, M2), but also fails to account for the complexity and nuances of ABCT (point of credit injection and goods purchased with this credit). The study boils things down to a very simple equation of: monetary shock should result in rising activity and relative prices in sectors more remote from the consumer. This is a grossly simplified and crude way of thinking about ABCT, and I am therefore not surprised the results were “mixed”.
PS. To really understand ABCT, one should read Ludwig von Mises’ “Human Action”, Friedrich von Hayek’s “Prices and Production”, Murray Rothbard’s “Man, Economy, and State”, and/or Jesus Huerta de Soto’s “Money, Bank Credit, and Economic Cycles.” They are all available for free download here.
UPDATE: For more on the shortcomings of the Lester/Wolff study and how using empirical data can/will skew the results, see Russell Lamberti’s follow-up post to mine: “RE: On The Empirical Relevance of Austrian Business Cycle Theory.”