Friday, August 24, 2012

Steven Horwitz on Stimulus Spending and Hayek

Steve Horwitz has the following post here on the London School of Economics (LSE) blog:
Steven Horwitz, “The Work of Friedrich Hayek shows Why EU Governments cannot spend their Way out of the Eurozone Crisis,” LSE blog, 21 August, 2012.
In response, we can say the following:

(1) His analysis is based on the false Austrian business cycle theory (ABCT). Despite Horwitz, the crisis of 2008 did not occur because the US “economy’s productive structure was not sustainable.” It was about the collapse of an asset bubble, excessive private debt, and a financial crisis caused by the collapse in value of the collateralized debt obligations (CDOs) and mortgage backed securities (MBSs) that banks had loaded up on.

(2) Horwitz charges that “advocates of stimulus are arguing is that we need spending, just any old spending, to jump start struggling economies” – a manifestly false claim. Post Keynesians and, I suspect, even New Keynesians understand that spending on asset bubbles would be just another disaster, and that what is needed is thoughtful public investment and social spending, not “just any old spending.”

(3) The issue of using resources that are not idle is raised by Horwitz, but he ignores the obvious point that, even if a stimulus project takes resources already employed, eventually, as a consequence, idle resources will be drawn on by both private and public sector investment projects in need of further factor inputs. Also, Horwitz conveniently ignores that fact that expansionary fiscal policy itself involves not just public investment, but expanding the capacity of the private sector to engage in consumption spending: with more private sector demand, businesses will also hire more workers and increase output.

(4) Horwitz contends that “Politicians and bureaucrats lack the knowledge to know which pieces fit with which pieces as they cannot know the nature of the idled resources and what consumers want.” And yet Keynesian stimulus worked again and again throughout American history when tried: real output soared and unemployment fell when expansionary fiscal policy was done by the US government in 1948–1949, 1954, 1958, 1961, 1964, and 1982. The Austrians might reply that such expansionary fiscal and monetary policies caused Austrian business cycles in all these cases, but that reply is worthless, simply because the Austrian cycle theory is worthless.

(5) Horwitz writes:
“So what can we do? The answer lies in the criticism: free up competition, prices, profits, and losses so that entrepreneurs and others can finish the process of tearing down the mistakes of the boom and figure out how to reallocate those resources to their new best uses. That process takes time, but if politicians cease meddling in it and start allowing market processes to do their job, particularly by allowing failed firms to go bankrupt and sell off their assets for more valuable uses, recovery will take place more quickly.”
This all sounds like liquidationism to me. Horwitz invokes Hayek in his blog title, but fails to tell us that even Hayek eventually repudiated liquidationism, and gave qualified support to Keynesian stimulus in a depression, as we can see here:
“To return, however, to the specific problem of preventing what I have called the secondary depression caused by the deflation which a crisis is likely to induce. Although it is clear that such a deflation, which does no good and only harm, ought to be prevented, it is not easy to see how this can be done without producing further misdirections of labour. In general it is probably true to say that an equilibrium position will be most effectively approached if consumers’ demand is prevented from falling substantially by providing employment through public works at relatively low wages so that workers will wish to move as soon as they can to other and better paid occupations, and not by directly stimulating particular kinds of investment or similar kinds of public expenditure which will draw labour into jobs they will expect to be permanent but which must cease as the source of the expenditure dries up.” (Hayek 1978: 210–212).
(6) The most astonishing statement by Horwitz in the whole post comes at the end:
“Before the advent of Keynesianism, most recessions were very short lived as producers were left free to shuffle the jigsaw pieces into better combinations.”
This distorts history on so many levels it beggars belief.

Take the recession that Austrians claim proves their prescription of liquidationism: the recession of 1920–1921. This lasted 18 months, and was far longer than every post-1945 US recession on record, with the exception of the great recession of 2008–2009 (which also lasted 18 months). This can easily be verified by looking at the NBER data here.

The official estimates from the National Bureau of Economic Research (NBER) show clearly that the length of recessions fell very significantly when modern Keynesian and monetary interventions were used to ameliorate recessions after 1945:
Period | Average Length of Recessions in Months
1854–1919 (16 cycles) | 21.6
1919–1945 (6 cycles) | 18.2
1945–2009 (11 cycles) | 11.1
US Business Cycle Expansions and Contractions,
Now you could complain that the traditional data for the 19th century has been challenged by Romer and Balke and Gordon, but that does not apply to the 1919–1945 period, which shows us quite clearly that the average length of recessions fell after 1945.

The issue of the length and severity of pre-1914 recessions is highly controversial, of course. We will only ever have estimates, whose validity and accuracy is open to question. The old Kuznets-Kendrick GNP series showed beyond any doubt that the pre-1914 era saw far more severe and longer recessions than those after 1945. While that data was challenged by Romer (1989), whose findings could be used to support Horwitz, it is now clear that there are problems with Romer’s estimates and method.

At best, one might appeal to the work of J. Davis (2006; 2004), but he only found no difference between the frequency and average length of recessions in the period 1880-1914 (the National Banking era) and the post-1945 era. If you really think Davis is correct, then that would still refute Horwitz.

In contrast, the work of Balke and Gordon essentially vindicate the Kuznets-Kendrick GNP series in terms of the improvement in output volatility after 1945 as compared with the pre-1914 era (see Appendix below).

Now the measure of how serious a recession is consists of (1) length of output contract, and (2) how long unemployment persists after the recession.

To look more seriously at the 19th century data, let us take the real GDP estimates of Balke and Gordon and the unemployment estimates by J. R. Vernon for the US from 1870–1900.

First, real GDP:
Year | GNP* | Growth Rate
1869 | 78.2 |
1870 | $84.2 | 7.67%
1871 | $88.1 | 4.63%
1872 | $91.7 | 4.08%
1873 | $96.3 | 5.01%
1874 | $95.7 | -0.62%
1875 | $100.7 | 5.22%
1876 | $101.9 | 1.19%
1877 | $105.2 | 3.23%
1878 | $109.6 | 4.18%
1879 | $123.1 | 12.31%
1880 | $137.6 | 11.77%
1881 | $142.5 | 3.56%
1882 | $151.6 | 6.38%
1883 | $155.3 | 2.44%
1884 | $158.1 | 1.80%
1885 | $159.3 | 0.75%
1886 | $164.1 | 3.01%
1887 | $171.5 | 4.50%
1888 | $170.7 | -0.46%
1889 | $181.3 | 6.20%
1890 | $183.9 | 1.43%
1891 | $189.9 | 3.26%
1892 | $198.8 | 4.68%
1893 | $198.7 | -0.05%
1894 | $192.9 | -2.91%

1895 | $215.5 | 11.7%
1896 | $210.6 | -2.27
1897 | $227.8 | 8.16%
1898 | $233.2 | 2.37%
1899 | $260.3 | 11.6%
1900 | $265.4 | 1.95%
* Billions of 1982 dollars
(Balke and Gordon 1989: 84).
It should be noted that these estimates are annualised, and no doubt conceal a number of recessions which lasted for 2 or 3 quarters in one year, where overall annual GDP growth was higher in the year of recession on the previous year. Likewise, where recessions lasted 2 or 3 quarters and spanned two years, but where overall annual GDP growth was higher in the second year on the previous year.

What is notable here is the double dip recession of 1893–1896: not a short recession by any means.

Next let us look at US unemployment in the late 19th century:
Year | Unemployment Rate
1869 | 3.97%
1870 | 3.52%
1871 | 3.66%
1872 | 4.00%
1873 | 3.99%
1874 | 5.53%
1875 | 5.83%
1876 | 7.00%
1877 | 7.77%
1878 | 8.25%
1879 | 6.59%

1880 | 4.48%
1881 | 4.12%
1882 | 3.29%
1883 | 3.48%
1884 | 4.01%
1885 | 4.62%
1886 | 4.72%
1887 | 4.30%
1888 | 5.08%
1889 | 4.27%
1890 | 3.97%
1891 | 4.34%
1892 | 4.33%
1893 | 5.51%
1894 | 7.73%
1895 | 6.46%
1896 | 8.19%
1897 | 7.54%
1898 | 8.01%
1899 | 6.20%

(Vernon 1994: 710).
I have highlighted in yellow those years where unemployment was over 5% and the years where unemployment showed a tendency to rise when it was above 5%.

By these figures, there was a marked rise in unemployment in the 1875–1878 and 1894–1898 periods. The double dip recession of the 1890s explains the rising unemployment from 1893–1896, and there was stubbornly high unemployment until 1898.

The high unemployment in the late 1870s also suggests serious problems in these years. The GDP growth after the recession of 1874 was a “jobless recovery.”

Although Balke and Gordon show positive GNP growth rates from 1875–1878, earlier estimates of GNP showed that the US economy experienced a recession in these years, with the NBER data showing the longest recession in US history from October 1873 to March 1879 (a 65 month recession). In view of the unemployment estimates for these years, at the very least there appears to have been contraction in certain important sectors. It is notable that, most recently, Davis (2006: 106) has found a recession from 1873 to 1875 lasting about two years, a contraction of around 24 months, or about 6 months longer than the longest post-1945 contraction on record (that of 2008-2009, which lasted 18 months).

By these figures alone, it is obvious that the late 19th-century US economy was sick in both the 1870s and 1890s. Recovery from recession was not rapid or smooth in either decade. Above all, the double dip recession of the early 1890s was quite long in terms of actual real output contraction.

For the 19th century, when we consider the impact of high unemployment following a recession, this makes nonsense of Horwitz’s statement that before “the advent of Keynesianism, most recessions were very short lived as producers were left free to shuffle the jigsaw pieces into better combinations.”

Appendix: Output Volatility pre-1914 and post-1945

What about output volatility before 1914 as compared with the post-1945 period?

I will quote Selgin et al. (2010) on this, since Selgin and his co-authors are libertarians:
“According to Romer‘s own pre-1929 GNP series, which relies on statistical estimates of the relationship between total and commodity output movements (instead of Kuznets‘ naïve one-to-one assumption), the cyclical volatility of output prior to the Fed‘s establishment was actually lower than it has been throughout the full (1915-2009) Fed era ... More surprisingly, pre-Fed (1869-1914) volatility (as measured by the standard deviations of output from its H-P trend) was also lower than post-World War II volatility, though the difference is slight.” (Selgin et al. 2010: 10)

“Romer‘s revisions have themselves been challenged by others, however, including Zarnowitz (1992, pp. 77-79) and Balke and Gordon (1989). The last-named authors used direct measures of construction, transportation, and communication sector output during the pre-Fed era, along with improved consumer price estimates, to construct their own historic GNP series. According to this series, the standard deviation of real GNP from its H-P trend for 1869 to 1914 is 4.27%, which differs little from the standard–series value of 5.10%. Balke and Gordon‘s findings thus appear to vindicate the traditional (pre-Romer) view.” (Selgin et al. 2010: 11).
While Selgin et al. cite other evidence that they believe vindicates Romer (see their paper), my point here is to make clear that Balke and Gordon confirm that modern macroeconomic management of the US economy has improved output volatility.


Balke, N. S., and R. J. Gordon, 1989. “The Estimation of Prewar Gross National Product: Methodology and New Evidence,” Journal of Political Economy 97.1: 38–92.

Davis, J. H. 2004. “An Annual Index of U.S. Industrial Production, 1790-1915,” Quarterly Journal of Economics 119.4: 1177-1215.

Davis, J. H. 2006. “An Improved Annual Chronology of U.S. Business Cycles since the 1790s,” Journal of Economic History 66.1: 103-121.

Hayek, F. A. von. 1978. New Studies in Philosophy, Politics, Economics, and the History of Ideas. Routledge & Kegan Paul, London.

Romer, C. D. 1989. “The Prewar Business Cycle Reconsidered: New Estimates of Gross National Product, 1869–1908,” Journal of Political Economy 97.1: 1–37.

Selgin, G. A., Lastrapes, W. D. and L. H. White, 2010. “Has the Fed Been a Failure?” Cato Working Paper no. 2 (November 9).

Vernon, J. R. 1994. “Unemployment Rates in Post-Bellum America: 1869–1899,” Journal of Macroeconomics 16: 701–714.

Zarnowitz, V. 1992. Business Cycles: Theory, History, Indicators, and Forecasting, University of Chicago Press, Chicago.


  1. Recessions should be getting shorter, as Steffan Karlsson has outlined here:

    I also invoke the Hayekian view of the use of knowledge in society. With more knowledge becoming available, this should also cause an improvement in the functioning of economies. You will likely contend with a theory of incentives to hide information, but I contend that is due to perverse incentives caused by government through intervention. Political incentive, regulation, regulatory arbitrage, supplantation of spontaneous order.

    You could argue in the line of Minsky's Financial Instability hypothesis that the changes in the size and power of the financial sector can make things worst without keynesian regulation, but this is severely flawed. The FIH has been critiqued numerous times at your much hated site. Brace yourself for the URl, Lord Keynes, I know you are delicate:

    1. (1) "Recessions should be getting shorter,"...

      And the data cited above show they did after 1945, certainly compared with the 1919–1945 period. And this is only the data for the US. No doubt one could cite evidence from European countries, Canada, Australia, etc.

      (2) First, I find the criticisms you cite from unconvincing, and secondly not all of them are dismissive at all:

      "So while Minsky's story accurately describes the present financial-market turmoil, it does not provide any satisfactory explanation based on previously established and identified phenomena. It arbitrarily puts the blame for instability on the capitalistic economy without even making the slightest attempt to establish a logical verification for this claim. We have seen, however, that it is the existence of the central bank that lays the foundation for financial instability. This means that, contrary to Minsky and the other post-Keynesians, the source for current financial turmoil is likely to be the central bank, i.e., the Federal Reserve's monetary policies."

      Shostak's only real "criticism" as far as I can see is that he thinks the central bank that is the main cause of the instability. That ignores the role of effective financial regulation, which can prevent Minsky cycles from happening.

  2. "With more knowledge becoming available, this should also cause an improvement in the functioning of economies."

    Maybe this is true in a world of no change or quasi change . But in a world of constant change of data, more knowledge does not necessarily mean that we have reduced uncertainty.

  3. You create a rather misleading impression in quoting from my, Lastrapes, and White's paper, in that the passages immediately following those you quote supply further evidence that appears to favor Romer's general conclusion over that of Balke and Gordon.

    1. I have noted this above after the quotation, to make clear to readers what Selgin et al. 2010 argues.

    2. Thanks for that. You might also consider linking to your later post concerning Joseph Davis' work, since that's also relevant here.

  4. More Austrian distortions of history. They tend to take the view that the 19th century was an idyllic capitalist era untainted by what they call "Keynesianism". However, it is quite clear that governments during the 19th century often intervened during recessions.

    Example: In 1874 Congress shot down the "Inflation bill" which sought to inject newly printed dollars into the system -- something like a 19th century version of QE. While this was rejected it led the way to the silver purchases of 1878 which were basically the same thing.

    Not to mention the fact that outside of the US central banks had been in existence for some time. Intervention was rife across the 19th century, but the Austrians -- being a cult -- prefer mythology to history.

    1. Actually, central banks were in fact not all that common at the time to which you refer. Switzerland didn't have one; neither did Sweden, or any of the British dominions, or Italy, or...well, it's a long list. That Canada didn't have a central bank until 1935, and yet (despite heavy dependence on trade with the U.S.) altogether avoided the crises that regularly afflicted the U.S. from 1873 through 1914, and avoided both bank failures and much of the monetary contraction that slammed the U.S. in the early 1930s, is particularly instructive, or will be to anyone not wedded to the naive view that central banks prevent more crises than they help to bring about.

      In any event, although I don't call myself an Austrian, Larry White, who is one of the coauthors of "Has the Fed Been a Failure" and the author of a darn good book on Scottish banking, as well as plenty of other historical stuff, hardly fits your sweeping generalization about Austrian economics. There are after all Austrians and "bastard" Austrians, just as there are Keynesians and bastard Keynesians.

    2. Some quick points on Canada:

      (1) In the 1800s, Canada had a large private bank called the "Bank of Montreal" that in some ways acted like a de facto central bank.

      For government interventions to stabilize Canada's financial system pre-1930, see Charles Goodhart and Gerhard Illing (eds.), Financial Crises, Contagion, and the Lender of Last Resort: A Reader, p. 121:

      "However, the Bank of Montreal (founded in 1817) very early became the government's bank and performed many central bank functions. Because Canadian banks kept most of their reserves on 'call' in the New York money market, they were able in this way to satisfy the public's demand for liquidity, again precluding the need for a central bank. On two occasions, 1907 and 1914, however, these reserves proved inadequate to prevent a liquidity crisis and the Government of Canada had to step in to supplement the reserves."

      (2) Canada did not see a financial sector meltdown from 1929-1933 because of implicit government guarantees and interventions:

      Kryzanowski, Lawrence, and Gordon S. Roberts. 1993. "Canadian Banking Solvency, 1922-1940," Journal of Money, Credit, and Banking 25: 361-376.

    3. The Canadian troubles in 1907 were due to a Canadian regulation which, quite inadvisedly, limited banknote issues (but not deposits) to bank capitalization, exposing the banks to problems when, as in the year in question, sharp increases in the public's desired C/D ratio put them up against the regulatory constraint. The regulation was relaxed, and that was the end of that.

      1914 of course is much more complicated. Suffice to say that the problem here was WWI and the consequences of British gold export restrictions, and consequent mass liquidation of foreign securities, which after all the canadian system handled rather well.

      Finally, as for the Bank of Montreal, look: of course in the absence of central banks, which let us not forget) are privileged statutory monopolies, banks in financial distress will look to other banks for loans, and to big banks especially. But to claim this as evidence that Canada "really" had a central bank before 1935 is, first of all, poppycock and, second of all, more an argument against the need for statutory currency monopolies than one favoring such.

      Indeed, there were many occasions during the 19th c. when genuine central banks turned to commercial ones for short-term loans.

      Finally, regarding the 1930s, I've read the article you refer to and there were no government guarantees of deposits. All that the government can be said to have done was to practice forbearance by not shutting banks down while they were temporarily insolvent, which other authorities claim was the case only for a couple of them and only for a year or so around 1932. This forbearance was, furthermore, only necessary the the extent that a strict (non-market based) "mark to market" rule was in play; under traditional bank accounting principles those banks would almost certainly have been kept open anyway.

    4. (1) " as for the Bank of Montreal, look ... in the absence of central banks ... banks in financial distress will look to other banks for loans ... . But to claim this as evidence that Canada "really" had a central bank before 1935 is, first of all, poppycock"

      You are correct. But I have not asserted that Canada "really" had a central bank.

      I said only that "the Bank of Montreal ... in some ways acted like a de facto central bank," a proposition that seems entirely correct to me.

      (3) "Finally, regarding the 1930s, I've read the article you refer to and there were no government guarantees of deposits."

      The article says there was no "explicit" government guarantees of deposits, only "implicit" ones, which the private sector took notice of.

      E.g., in 1923 the Government of Quebec provided $15 for the merger of the Bank Nationale with the Banque
      d'Hochelaga to avoid a bank failure (Kryzanowski and Roberts 1993: 365).

      When in 1923 the Home Bank of Canada failed, "depositors in the Home Bank petitioned the Canadian Government for compensation and received payment up to 35 percent of the value of their deposits" (Kryzanowski and Roberts 1993: 365).

      "Archival evidence suggests that after 1923 the Canadian government provided an implicit guarantee that no chartered bank would be allowed to fail and cause depositor losses. Although never formally embodied in law, this implicit guarantee was
      equivalent to 100 percent deposit insurance."
      (Kryzanowski and Roberts 1993: 365).

      "In summary the archival evidence is consistent with our hypothesis that beginning in 1923, an implicit guarantee from the Canadian government (amounting to 100 percent implicit insurance) stood behind all domestic bank deposits. The government actively facilitated mergers during the 1920s to avoid firesale insolvency and successfully created public confidence that no bank would be allowed to fail. This confidence persisted during the 1930s" (Kryzanowski and Roberts 1993: 365).

    5. Corrections:

      "E.g., in 1923 the Government of Quebec provided $15 million"

      "... This confidence persisted during the 1930s" (Kryzanowski and Roberts 1993: 366)."

  5. As someone who generally takes a Post-Keynesian approach to macroeconomics, but Austrian approach to political economy, I think you correctly point out several issues in Horwitz's post. However, I think one of your critiques fails to get at the main issue highlighted by Horwitz.

    You note, "Post Keynesians and, I suspect, even New Keynesians understand that spending on asset bubbles would be just another disaster, and that what is needed is thoughtful public investment and social spending, not “just any old spending.”

    I agree, but this doesn't answer the question of how government can be persuaded to not spend money on unproductive uses (ie. another asset bubble) or what represents "thoughtful public investment and social spending." In theory those efforts would be grand, but I remain unconvinced that political incentives will, in reality, lead to that result.

    IMO, that is an important takeaway from Horwitz on Hayekian limits of knowledge (

    Lastly, these comments offer some great thanks.

  6. LK, I think we are not really disagreeing substantively w.r.t. Canada, since I referred only to the 1930s, and there were neither failures nor mergers during that time, the 1923 HB failure having been the last of its sort prior to the Great depression. I will however say that, if there were "implicit" guarantees the evidence for which consists of "archival evidence," it is hard to see how those guarantees could have had any bearing on the Canadian public's confidence in the banks concerned.

  7. Just a quick point: Kryzanowski and Roberts's reference to "archival evidence" is a tad misleading, for there were quite public statements from finance ministers reported in the press promising government intervention to protect depositors.

    E.g., the Minister of Finance to the McKeown
    Commission, 1924:

    "Under no circumstances would I have allowed a bank to fail during the period in question . . . If it had appeared to me that the bank was not able to meet its public obligations, I should have taken steps to have it taken over by some other bank or banks, or failing that, would have given it necessary assistance under the Finance Act, 1914.

    The full evidence is given Kryzanowski and Roberts 1993: 365ff.

  8. LK,

    Why do you use GDP data from the 19th century when you stated on your blog dated 8/25/2012 Davis “All this alert us to how questionable is the whole project of real GNP/GDP estimates for the 19th century”. Would if not be more factual to state that reliable date in regard to GDP is not available.