Former Austrian school economist Bryan Caplan recently won a bet against Austrian school economist Bob Murphy on the path of US inflation. Caplan won by betting with the key market indicator (TIPS), Murphy lost by betting against it.
At first glance, that the ex-Austrian won by following the market while the Austrian lost by not doing so might seem strange, but it instances why I am deeply unpersuaded by Austrian Business Cycle theory–that it is an analysis from a tradition that very strongly favours taking markets seriously (particularly their information revealing qualities) yet strikingly stops doing so to get a congenial theoretical outcome.
Austrian Business Cycle Theory (ABCT) is a theory of the unsustainable boom. It notes that capital is highly varied (or, economist say, is heterogeneous)–in particular, has a range of durations until completion. Interest rates coordinate current expenditure versus future income expectations.
If the central bank, in order to foster economic expansion, sets the key interest rate “too low”–that is, below the level that will create a stable level of successful capital projects–then entrepreneurs are led to over-invest in projects because capital is cheaper than its actual long-term prospects justify. So, there is misallocation of capital–the profile of created capital does not fit actual expenditure patterns. This is what Austrian theory calls malinvestment. The result is a surge in failed business projects, consequently of failed or distressed firms, leading to income and expenditure cuts, leading to that transactions crash we call “recession” or, if sufficiently severe, depression.
My objections to the theory are twofold: it doesn’t fit the evidence and it is implausible even in theory.
Not fitting the evidence
The theory suggests that the long-term economic pattern should be one of a surge in economic output above trend (the unsustainable boom) and then a crash below it. This is not the pattern we see: on the contrary, what we see conforms much more to Milton Friedman‘s “plucking model” (pdf)–that is, there is a long term growth trend that recessions and depressions “pluck” the economy away from (pdf). A pattern which suggests the economy is pushed (temporarily) off its growth path by various shocks. Despite attempts to claim otherwise, I am unpersuaded that ABCT can be re-construed to fit the evidence.
Particularly as the theory also suggests that the crash should be correlated with the preceding boom–the further the capital overshoot, the worse the resulting crash. Again, this is not what we see (pdf). Recessions and depressions are not correlated with the preceding expansions, but are correlated with the subsequent expansions (pdf). A result which led Friedman to propose his “plucking model”. Again, this conforms far better with the economy being shocked off its growth path before returning to it.
Given that industries systematically vary by both the scale and duration of their capital creation, the theory also implies that the crash should hit in sequence and to varying degrees–the shortest capital duration industries hit first, the longest capital duration industries hit later; the lower scale capital industries hit least, the bigger scale capital industries hit most. These factors are, to a significant extent, contra-indicated–i.e. short duration capital projects also tend to be low scale capital industries while long duration capital projects tend to be high scale capital industries.
Even so, there should a capital-profile sequence to industry downturns. Again, this is not what we see: transaction crashes tend to hit all industries simultaneously. Such transaction crashes are most plausible assigned to the demand side (i.e. monetary factors) as, in a monetised economy, money is the thing which is one half of all transactions in all industries. Even when there are supply shocks, (1) monetary policy can counter-balance the effects and (2) such shocks are generally a specific shock to the economy, not rolling capital project failures.
One might counter by arguing that particular projects are engaged in a rolling fashion. But that reduces the industry sequencing issue at the cost of undermining the systematic distortion effect.
The theory also assumes that central banks are biased in one direction only–in an inflationary one. Yet the historical record shows that, while there is certainly a general inflationary trend for fiat money, there was no such trend by central banks under gold standards. And ABCT was originally devised in a gold standard world. Attempts to redefine “inflation” to mean “monetary/credit expansion” simply beg the question–an alleged cause being conflated into the presumed effect.
Moreover, the historical record also shows that, in the right circumstances, central banks can be biased in a contractionary direction. This was most dramatically true in 1928-32 but also true from 2008 onwards: on both occasions, the contractionary bias was because central banks prioritised policy credibility (commitment to the gold standard; commitment to low inflation) over economic activity. In doing so, contractionary central banks created the most severe economic downturns of the
C20th[last 100 years]. A business cycle theory that is so dramatically wrong about the two worst economic downturns of the C20th[last 100 years]–central bank policy in the opposite direction as predicted and economies consequently being shocked off their growth path–is not much of a business cycle theory.
Implausible in theory
So, there are severe evidentiary problems with the theory as any sort of general business cycle explanation. Even saying “but it is just a theory of the unsustainable boom” suffers from the lack of instances it accurately describes.
There are also some serious theoretical problem with the theory. The first is, ironically, not taking heterogeneity of capital seriously enough. Heterogeneity of labour and of capital leads to heterogeneity of debt and debt/equity profiles. How can there be a key single, natural or otherwise, rate which can distort the entire structure of investment? Including across its varying time frames, across which interest rates also vary.
What we are looking at is a schedule of interest rates varying by time and asset. It can be argued that the central bank policy rate (the interest rate used to signal policy) effectively anchors the entire schedule, as the central bank is the monopoly supplier of the monetary base. Its policy rate is really an indicator about the future path of monetary policy, and an indicator which is a function of it being said monopoly supplier and its policy credibility. But an indicator which has far more direct effects on nominal interest rates rather than real interest rates.
But to put so much emphasis on interest rates in investment decisions looks perilously like reasoning from a price change. The central bank has signalled, by cutting its policy interest rate, a more expansive path in monetary policy. But that is, for the economy, a general tendency: entrepreneurs still have to make assessments about particular assets and particular production decisions. As the localised nature of the housing market booms and busts in the US have demonstrated, housing markets experiencing the very same monetary policy can have very different dynamics.
The claim that entrepreneurs will be sufficiently homogeneous in their responses, across very heterogeneous asset and production markets, to create the bust looks suspiciously like only embracing complexity when it is convenient. (Noting that to claim more decisions to invest will be made is not the same as claiming that the structure of production will be distorted.)
More seriously, the claim runs into an information problem–as others have noted, Austrian theory apparently has access to information than none of the market participants do. The central bank knows enough to inflate the economy but none of the market participants have the knowledge to work out what the central bank is doing and the consequences thereof. There is a serious consistent expectations (i.e. rational expectations, but consistent expectations is a more accurate term) problem here.
In his (losing) bet with Bryan Caplan, Bob Murphy was being very “Austrian” in assuming his theory gave him information hidden from market participants–Austrian theory really, really believing in markets until it suddenly really, really doesn’t. Bryan Caplan was being much more consistent (dare one say rationally consistent) in his expectations by going with the market indicator.
What is more plausible–that there is enormously-important-for-future-income information lying around being ignored by everyone except by clever Austrian school folk or that economies are shocked off their growth path: an economic shock being an unanticipated change?
Austrian school, meet the Australian economy
These theoretical and empirical problems come together in the record expansion of the Australian economy since 1991. That is, Australia has not had an economic recession (in the sense of two quarters of [negative] economic growth) since 1991. It still has a business cycle, just a very flat one.
What is more plausible–that the Reserve Bank of Australia (RBA) got its policy interest rate essentially correct for 23 years straight, so that Australian entrepreneurs got their capital projects (on balance) continually right? Or that the RBA sufficiently anchored inflation and income expectations that the Australian economy has not been shocked enough off its growth path since RBA introduced its policy of aiming for a 2-3%pa inflation rate on average over the business cycle ?
Surely, the second option is much more plausible.
So, I do not agree with the Austrian School business cycle theory. In particular, I am very unsurprised that an Austrian economist lost by betting against the key market indicator. The Austrian Business Cycle theory presumes special knowledge against market agents and the indicators they generate: a presumption which is not a strength. Still less a reason to accept the theory.
[Cross-posted from Thinking Out Aloud.]