Selgin on Mellon, Harding, and Hoover

by Don Boudreaux on August 19, 2011

in Great Depression, History, Monetary Policy, Myths and Fallacies, State of Macro

Here’s George Selgin on the 1920-21 economic downturn in the U.S.  His conclusion:

Proponents of Keynesian pump-priming often berate the Hoover administration for its “liquidationist” strategy for dealing with the outbreak of the Great Depression–forgetting that it was Hoover himself who caricatured Andrew Mellon, his Secretary of the Treasury, as someone who wished to “liquidate” the stock market, farmers, real estate, and so forth, and who took pride in not having followed his advice. But Mellon was also Harding’s Secretary of the Treasury; and Harding, unlike Hoover, trusted him. It is one of the greater ironies of economic history that “liquidationist” policies, including government austerity, are blamed for prolonging a depression for which those policies were set aside, while being denied credit for perhaps helping to end one for which they really were put into practice.

Be Sociable, Share!



33 comments    Share Share    Print    Email


Paul August 19, 2011 at 2:58 pm

“… since the 1920-21 recession was not characterized by any aggregate demand deficiency, fiscal stimulus was unwarranted.”

Jacobite August 19, 2011 at 3:28 pm

Leftists lie. About everything, even events you have seen with your own eyes. You wanna go back to the Founders? These were no Libertarian freakaloons like J S Mill. They allowed freedom of speech subject to civil and criminal libel and slander laws. Being males, they also recognized the usefulness of the Code Duello in encouraging honesty.

Invisible Backhand August 19, 2011 at 4:40 pm

Have you considered using a connected series of statements intended to establish a proposition?

I can’t understand what you just wrote.

I’m a leftist. Ergo, I’m lying. I don’t understand what you just wrote. Well, I understand those are words in English with a subject/predicate structure, but I don’t understand the thesis. But since I’m a leftist, I’m lying and I do understand the thesis.

Sneaky hobbitsses, trying to steal my preciouusss….

Greg Webb August 19, 2011 at 3:34 pm

Don, excellent quote! I also appreciated your quote in today’s Wall Street Journal: “Macroeconomics was nothing more than a dismissal of the rules of economics.”

George Selgin August 19, 2011 at 3:34 pm

Paul: Not so. Check the real GNP and CPI data in the charts here:

bearing in mind that “aggregate demand” = nominal GNP = real GNP x CPI.

Mesa Econoguy August 19, 2011 at 5:23 pm

Fantastic link, thanks sir.

Amity Shlaes references Prof. Smiley extensively in her credits for The Forgotten Man.

There is a major uptick in left-wing 1920 – 1945 revisionism/revalidation attempts out there now, for obvious reasons.

juan carlos vera August 19, 2011 at 3:38 pm

The old trick of blaming the freedom what never happens in freedom…

George Selgin August 19, 2011 at 3:48 pm

Just looked at that Wikipedia article on the 20-21 episode. The statistic reported there clear show that demand shrank substantially; as must be the case whenever both output and prices are falling. I do not know how our friend Daniel Kuehn, who is quoted in the article, arrived at his conclusion that K “that Woods underemphasizes the role the monetary stimulus played in reviving the depressed economy and that, since the 1920-21 recession was not characterized by any aggregate demand deficiency, fiscal stimulus was unwarranted.” The statement is if anything backwards: monetary stimulus is itself only capable of promoting recovery by enhancing aggregate demand; indeed, were there no deficiency of aggregate demand, monetary expansion could only serve to cause inflation, while fiscal stimulus, e.g., Mellon’s reduced tax rates, might in principle “stimulate” the economy’s supply-side.

Perhaps Mr. Kuehn will explain.

Daniel Kuehn August 19, 2011 at 5:14 pm

I provided something of a chronology and interpretation to the real work that minds far more brilliant than mine already produced. I’ll quote one of them (Romer, 1988):

“The downturn of 1921 is conventionally attributed to a decline in aggregate demand. Lewis (1949, pp. 18-20) argues that private consumers and producers contributed to the decline in 1921 by overspending on all types of goods after the war. As a result, by 1921 their demand was satiated and the stock of durables was very young, so they greatly curtailed their spending. Friedman and Schwartz (1963, pp. 231-242) argue that the Federal Reserve Board caused a further fall in aggregate demand by allowing the money supply to contract sharply between 1920 and 1921. Available evidence appears to confirm the view that aggregate demand declined substantially between 1920 and 1921. For example, estimates of the money supply show that M1 fell 10 percent between 1920 and 1921. 23 In the conventional story this fall in aggregate demand is supposed to have caused a large fall in both output and prices because prices were not fully flexible.

The behavior of the superior Kendrick GNP estimates suggests that this conventional explanation must be altered. Despite the substantial fail in aggregate demand, the Kendrick series indicates that total GNP fell very little between 1919 and 1921. As a result, it is impossible to argue that the decline in demand moved the economy down an upward sloping aggregate supply curve and thus drove down both output and prices substantially. This is especially true considering the magnitude of the actual fall in prices between 1920 and 1921. In this period the implicit price deflator for GNP given in table 5. falls 16 percent and the BLS wholesale price index falls 46 percent.

An obvious alternative explanation for the behavior of the economy in 1921 is that prices were very flexible. If the aggregate supply curve for the economy were very steep in the period around 1921, then one would expect movements in aggregate demand to fall almost entirely on prices and to have very little effect on output. The possibility that prices were very flexible in this period is made stronger by the fact that as discussed previously, the government spending associated with World War I also led to relatively little movement in real output and substantial movement in prices.

General price flexibility, however, probably cannot explain the entire behavior of the economy in 1921. In particular, the decline in prices is larger than one would have expected judging from the behavior of the economy during the war. Using the data in table 5 one can see that real GNP rose 5 percent between 1917 and 1918 and the GNP deflator rose 15 percent. In contrast, between 1920 and 1921 a fall in real GNP of only 2 percent was associated with a price decline of 16 percent. This seems to indicate that there may have been some type of aggregate supply shock either during the war or in 1921.

The most obvious candidate for such a supply shock are the price controls implemented during World War I. However, while price controls were in effect in some indastries in 1918, most researchers estimate that they had only a limited effect in restricting price increases. This is because many of the controls took the form of guaranteed minimum prices designed to encourage production rather than maximum prices. A more plausible explanation for the
differential behavior of prices in World War I and 1921 is the occurrence of beneficial shocks to prices in 1921.

According to a classic study of the 1920s by George Soule (1947, pp.
99-100), a surge in agricultural production drove down prices in 1921. This surge in production is due to the fact that American farmers continued to produce at wartime levels despite the recovery of European production. Soule argues that the price of primary commodities produced outside the United States such as wool also plummeted in 1921 because of a surge in supply. Soule attributes this surge to the fact that a variety of agricultural goods and raw materials had been accumulating in the producing countries for several years because the foreign shipping network customarily used to transport the goods was disrupted by the war. By 1920, the European and American shipping industries had been restored and these goods began to enter the market.

Several pieces of evidence suggest that these shocks occurred and that they were significant.”

She goes on from there, but I think you get the point of the argument. Broadberry had similar conclusions for the UK about this time, and Temin and Smith both come to these conclusions too (maybe not Temin… maybe just Smith and Temin was commenting on something else – I’d have to go back and look).

As someone who both respects Romer’s careful work in attending to pre-WWII data series and as someone who understands that there are beneficial deflations, I think you should be able to appreciate this argument.

For those interested – I currently have a shorter article under review at the Cambridge Journal of Economics on this episode, making a similar argument to the one I make in the Review of Austrian Economics.

jjoxman August 19, 2011 at 6:30 pm


Would you say that the most recent recession is characterized by a deficiency in aggregate demand?

Daniel Kuehn August 19, 2011 at 5:17 pm

On the monetary stimulus point… I suppose it depends on how you look at it. Was the reduction of rates in 1921 “stimulus” or was it simply the removal of the anti-stimulus of the high rates? The economy sucked a lot less after Volcker stopped turning the screws on it in the early 80s. Is “stimulus” the best word for that? You may be right – it may not be the best word for it. But it had a predictable impact.

Daniel Kuehn August 19, 2011 at 5:30 pm

Now I have a head-scratcher that perhaps you could explain, Dr. Selgin. You write of Harding’s austerity that it is “denied credit for perhaps helping to end one for which they really were put into practice”

The Harding tax cuts came into effect in 1922. The budget was balanced before the depression even started in 1919 (monthly federal expenditures were cut by about 75%, from $1.6 billion to $400 million by the beginning of the depression. By 1922 Harding had cut it another $100 million, where it would stay – I’m eyeballing these from my CJE pdf graph… I can track down the exact figures if anyone’s interested).

So we have a huge Wilsonian austerity before the depression even starts, then we have a modest further cleaning of house by Harding after the depression ends. The connection between Harding’s fiscal austerity and recovery here seems fairly weak. The man was inaugurated the same month industrial production troughed, for God’s sake! How big of an effect could he have had? Granted, you do say “perhaps helping”, and that “perhaps” may save you. But I still don’t see what would lead someone to associate Harding’s budget cutting to this recovery.

Henri Hein August 19, 2011 at 6:52 pm

Is it important to your theory who implemented the austerity?

Daniel Kuehn August 19, 2011 at 6:56 pm

Not at all – it’s the timing. The fiscal austerity episodes occured before it started and after it ended.

The monetary austerity and subsequent loosening is a somewhat closer correspondence. I’m not sure whether we can consider it causal, but if I remember (again – this was over a year ago that I was really in the weeds of this) it has a pretty plausible six month lag I believe. Certainly plausible as a contributing factor (monetary tightening with the downturn, and easing for the recovery).

George Selgin August 19, 2011 at 7:18 pm

Whatever Wilson started, Harding continued–with a vengeance. As Randy Holcomb shows in this article, government spending as a share of real GNP fell from about 7% in 1920 to well below 4% in 1922–a very large drop.

As for Kendrick’s numbers, they don;t show that there was no substantial decline in aggregate demand between 1920 and 1922. They suggest that the decline in real output wasn’t so large, because prices bore the brunt of adjustment. This isn’t what is asserted in the Wikipedia article referred to above. What’s more, it’s hardly consistent with contemporary narratives pointing to widespread business work stoppages and a substantial increase in unemployment. Generally, when you have a lot more unemployment, you get less output, holding total factor productivity and capital input constant–which surely they were (at least approximately) over the relatively brief interval in question. So I think that it’s Kendrick’s numbers that perhaps understate the real downturn, however much they may avoid problems in the old series.

Daniel Kuehn August 20, 2011 at 7:16 am

re: “Whatever Wilson started, Harding continued–with a vengeance.”

That doesn’t seem to be what Table 1 of your article shows, George.

Anyway – that’s immaterial. I don’t care who did it “with vengence”. The monthly federal outlays data from the NBER shows a sharp drop in federal spending in 1919 and a slightly less sharp drop in 1922. Federal spending was fairly stable during the depression, as were tax cuts (that also went into effect in 1922). So this is all largely irrelevant to the depression.

George Selgin August 20, 2011 at 8:04 am

Daniel, nominal Federal spending dropped quite a lot during the depression, as the stats cited in my original article show; consequently real spending dropped even more. I don’t quite understand why you seek to deny it or what alternative measures of spending you have in mind in doing so. (Mine come straight from the fiscal accounts.) Moreover, when there’s a severe change in overall spending, it’s the government share of the total that is the most reliable indicator of the fiscal policy stance, and as I noted (and as the table in Holcombe clearly shows) that share took a nose dive after 1920. I don’t see how this nose dive can be “irrelevant” to the point in question.

Indeed, since the point in question is whether recovery from a downturn is possible in the face of fiscal “austerity,” the fact that Wilson reduced spending during the boom preceding the crash is what’s truly irrelevant.

Gene Callahan August 20, 2011 at 8:30 am

“Indeed, since the point in question is whether recovery from a downturn is possible in the face of fiscal “austerity,” the fact that Wilson reduced spending during the boom preceding the crash is what’s truly irrelevant.”

Very strange — you’d think that would be relevant to the cause of the downturn, not the recovery!

Daniel Kuehn August 20, 2011 at 8:40 am

George, I told you exactly where I got my data, but if you want I can provide the link too:

The monthly data is erratic. If you look at a twelve month moving average of it, you see exactly where the big 1919-1920 drop occurs. The budget is balanced and most of the cuts have happened by November of 1919. The depression doesn’t even start until January 1920.

re: “Indeed, since the point in question is whether recovery from a downturn is possible in the face of fiscal “austerity,” the fact that Wilson reduced spending during the boom preceding the crash is what’s truly irrelevant.”

Is that the question? Well of course it can recover in the face of fiscal austerity. Anyway, yes I agree that Wilson’s spending reductions are irrelevant to the crash.

George Selgin August 20, 2011 at 9:39 am

Daniel, I’m not contesting your data or claims for budget-cuts in 1919-20; I’m disputing your suggesting that there was no further, deep trimming in 20-22.

You write, “Well of course it can recover in the face of fiscal austerity.” But what you say is possible “of course” is exactly Skidelsky claimed was impossible, which was the reason for my original blog. I’m glad you agree with me that he’s wrong; but it seems that somewhere you missed the context of my claims.

Daniel Kuehn August 20, 2011 at 10:21 am

George I can’t and don’t pretend to speak for Skidelsky. The whole point of my recent discussion of 1920-1921 was to say that it is possible to recover from a downturn in the face of fiscal austerity (or in the absence of stimulus), and if your blog post agrees with that that’s great. I’m not denying there was trimming in 1919, 1920, 1921, and 1922 either. I am saying the 1919-1920 decline is sharper – that’s in the data I linked to, and a lot of the Harding cuts (to spending and taxes) were too late to make all that much of a difference in the depression.

Daniel Kuehn August 20, 2011 at 10:30 am

My concern was over your claim that “perhaps” the austerity revived the economy. That doesn’t seem credible for me for reasons I go into a lot more detail in elsewhere. That shouldn’t be confused with making Skidelsky’s argument (if he made it – I’m taking your word on that… I could imagine him saying something like that) that the economy can’t recover without fiscal stimulus. It’s always been my position that 1920-1921 is a perfect example of where you don’t need fiscal stimulus.

BonnieBlueFlag August 19, 2011 at 6:43 pm

Andrew Mellon is the unsung hero of the Captains of Industry. The way FDR desperately tried to lynch him for tax evasion was atrocious.

George Selgin August 19, 2011 at 7:21 pm

P.S.: Even Romer’s numbers show a sharp rise in unemployment–though not as sharp as other unemployment estimates for the period. How, I wonder, does she suppose that firms kept production up while laying all those workers off?

And yes, I do generally think very highly of Romer’;s economic history–muchy better than her work since becoming a champion of Obama’s policies!

Daniel Kuehn August 20, 2011 at 7:01 am

If you continue reading she provides evidence of stocks of goods being build up which predictably leads to a supply shock, deflation, and quick contraction. Benjamin strong talked about this too in his 1921 Congressional testimony.

George Selgin August 20, 2011 at 8:09 am

“Stocks being built up” doesn’t answer the question concerning how industrial production can fail to decline when 5 million workers are let go. On the contrary: the existence of such stocks of unsold goods would only give firms that much more reason to halt production pending a decline in costs and restoration of profit margins.

None of this is to deny that prices proved remarkably flexible in the episode in question, allowing recovery to be swift. But the point is that this was a swift recovery from something more than than a statistical mistake!

G. Lammert August 19, 2011 at 11:16 pm

Keynesian stimulation will soon be become evident.

The Null Hypothesis and the Science of Saturation Macroeconomics

On Monday and Tuesday, 22 August and 23 August there will be a global nonlinear equity crash of historical proportions.

And on these two days one of the greatest hypothesis of the 21st century, ‘quantitative saturation macroeconomics’ will be validated.

The debt/money/asset macroeconomic system has its own intrinsic very quantitative operating laws that represent ideal fractal time dependent self assembly of asset saturation curves and define the counterbalancing limits of the macroeconomic saturation system. An understanding of saturation macroeconomics as a science with self assembly and self organization laws equal to physics and chemistry and biology can potentially guide global economic policy, monetary policy, and banking money-creation policy, and change rules regarding speculation on leverage assets.

These very simple laws were empirically observed in repetitive time based fractal patterns of varying time dimensions throughout the time based evolution of asset valuation curves and were defined in 2005 in the Main Page of the Economic Fractalist.

x/2.5x/2x/1.5-1.6x . The first three fractal phases are asset saturation growth fractals and the final or fourth fractal is an asset valuation decay fractal. X is a unit of trading time whose dimension can be minutes, hours, days, weeks, months, years, or decades. The 1.5-1.6x fourth fractal can itself be composed of a decaying x/2.5x/2x/1.5-1.6x 4 phase fractal series or a y/2-2.5y/2-2.5y 3 phase decay fractal series. The third fractal is ideally 2x in length but can be extended to 2.5x in length if growth is favored by underlying money supply growth or if the preceding valuation decay is significantly great.

On 11 August 2011 in Alpha’s The Economic Fractalist Instablog a final decaying growth sequence was predicted as a 27 July 2011 3/8/6-8/5 day 4 phase fractal with a potential of a 2.5x 8 day third fractal extension.

The 20 August 2011 actual fractal progression is currently 3/8/7/3 of 5 days with a dropping of the Wilshire composite equity valuation near the 9 August 2011 interday low.

The NASDAQ final fractal sequence leading to today’s near low begins its final fractal crash journey with its first base fractal including the key reversal day 3 year high on 2 May 2011. The first base fractal of the ensuing pristinely perfect x/2.5x/2x/1.5-1.6x sequence is a 13 day fractal starting 18 April 2011 and ending on 5 May 2011 intraday low to intraday low. The preceding end of one fractal is the beginning of the next fractal; incipient growth begins in final decay. There is an elegant integration process in quantitative saturation macroeconomics where larger initiating growth time unit, for instance, an incipient growth fractal denominated in the unit of a year incorporates a decay period of the last few of the final lower order time units, for instance, one or two terminal decay months of the preceding monthly low to low fractal series may be incorporated lasting into the follow fractal that last for a 100 months.

32 trading days later, after the first 13 day fractal end on 5 May 2011, an averaged NASDAQ low was made on 20 June 2011. The intraday low fell two trading days before but the averaged low of June 20 was equivalent to the averaged low of that day and was decidedly lower than the 20 June preceding day’s trading average. Third fractal growth concluded 26 trading days later on 26 July 2011 with lower low gap between the 26 July and 27 July delineating the 4th decay fractal of an expected 26 July 2011 1.5-1.6x :: 20 to 21 trading days. 19 August was day 19 of the expected 20 to 21 day :: 1.5:1.6x fourth fractal.

Interpolated terminal fractal sequences are everywhere:

Starting on 14 July 2011 an interpolated fractal series : 3/8/6/5 days:: x/2.5x/2x/1.6x ending on 5 August followed by and starting on 5 August a 2/5/3 of 5 days :: y/2.5y/2.5y decay fractal…..

Starting on 18 July 8/18 of 20 days x/2.5x …..

Starting on 27 July for the 18 day second fractal of the 8/18 of 20 day fractal series: 3/8/7/3 of 5 days

and finally starting on 18 April the reflexic fractal series proportionally identical to the 20/50/40 days x/2.5x/2x series that prospectively was predicted by Saturation Macroeconomics in the Huffington Post to be the Wilshire’s nonminal final high on 11 October 2007, a 13/32/26 day :: x/2.5x/2x reflexic growth fractal with a (decaying) 26th day lower high followed by a delineating gap and a 26 July 2011 19 of 20-21 day 1.5-1.6x 4th fractal which on 20 August 2011 is sitting on the edge of 154 year US Composite Equity second fractal nonlinearity that began in 1858..

The Null Hypothesis for the Science of Saturation Macroeconomics

Technically via the Chartist and in the qualitative perspective of a collapsing Euro Union and Euro currency and a polarized, nonnationalistic, corporate owned US government now hawking austerity, the 22 and 23 August 2011 asset collapse in weeks months years retrospectively will be perceived as obvious and expected. But the null hypothesis will remain: Why did the collapse occur on the 20th and 21st day of an ideal 4 phase 13/32/26/20-21 day :: x/2.5x/2x/1.5-1.6x Lammert fractal.

Why did it end in precisely this time course. It is the implosion of the system’s supporting money supply that is causing the collapsing fractal patterns. This is how the macroeconomic system works.

Null hypothesis: the collapse on 22 and 22 August 2011 is not related at all to the simple fractal quantum laws of saturation macroeconomics and the easily observed 18 April 2011 Lammert x/2.5x/2x/1.5-1.6x pattern of 13/32/26/20-21 days with the crash coming on days 20 and 21 of the defined fourth decay fractal is occurring by chance and chance alone.

This null hypothesis can be applied to larger order fractals including the 34/84 of 85 Wilshire x/2-2.5x quarter (3 month) fractal beginning in 1982 and the 70-71/153 year US equity fractal beginning near the ratification of the constitution in 1788-89.

Adam Betts August 20, 2011 at 4:40 am

Rothbard has an interesting take on the period following this sharp recession, 1921 -1929, a period often seen as one of ‘price stability’, but where the money supply increased dramatically (familiar?).

jjoxman August 20, 2011 at 8:04 am

Since Daniel hasn’t seen my comment or deemed it worthy of response, I’ll spell it out here.

The 1920-1921 change in real GDP was somewhere around -2.6% (Romer, 1988) or -3.5% (Balke & Gordon, 1989). According to Balke & Gordon, the 1919-1920 change was -2.05%. Daniel says this recession wasn’t characterized by aggregate demand deficiency. Rather it was induced by supply shocks. Okay.

The 2008-2009 change in real GDP was -3.83% (my calculations from BEA data). The 2007-2008 change was -0.3%. Excluding the stimulus changes these values by decreasing the 2008-2009 change to -4.13%. The cumulative 2007-2009 change then is around -4.5% (rounding to nearest nice number).

The cumulative 1919-1921 change was -5.5%. If the 1919-1921 depression was not characterized by aggregate demand failures, why assume that the 2007-2009 depression was, when it exhibits a smaller drop in production?

Romer attributes the shock to agriculture production and a dropping price for agricultural goods. And, it’s true, farm wages went down, but farm employment didn’t change very much (I’m getting this from Smiley, that the excellent Prof. Selgin linked to.) I don’t see curtailment of production. So I don’t see exactly agriculture’s role here. I guess I’m dense.

Let’s accept for the sake of argument it was supply shocks. Isn’t the sharp increase and then collapse of the house building sector a supply shock? All of a sudden all these framers, masons, electricians, etc. are not needed anymore. Their ability to demand is now constrained by their inability to supply valuable labor.

Both depressions were induced by the Fed’s pursuit of below-market interest rates. In the most recent case to inflate out of the tech-boom crash, in the the 1919-1921 case to help the Treasury market it’s V-bonds. So when the Treasury raises its rates, the inflation stopped and a reordering of the production distribution took place. Makes sense to me.

The Fed steadily increased the fed funds rate from 2003 to 2006, and then dropped it back as the economy faltered because it was trying to reorder itself.

The only difference I see is that there was an (apparent) fiscal crisis in the recent depression. I can’t see if there was one or not for the 1919-1921 depression. This calls for increased money supply, not fiscal stimulus.

Another big difference is in the nature of the malinvestment. In the most recent action, there was deliberate government intervention to channel investment into housing, and so given that, one could surmise it would naturally take longer to work out the malinvestment. Of course, all the interventions by the government, aside from the stimulus, hinder progress in purging the rot.

Right, so my question: if the 1919-1921 depression is not an aggregate demand depression, why say that the 2007-2009 depression is?

George Selgin August 20, 2011 at 8:19 am

Let’s please not concede the “no aggregate demand decline” thesis. The usual way of thinking about aggregate demand is simply as the total volume of nominal spending, Py. By that ordinary measure, the recession of 1920-22, in which both P and y dropped, the former more sharply than the latter, certainly did involve a sharp drop in aggregate demand! In contrast, a pure supply-side decline in real output (y) will be accompanied by rising prices.

If Daniel or anyone else can come up with a set of AS-AD diagrams in which a pure AS shock is shown to be capable of accounting for the conjunction of P and y movements between 1920 and 1922, I will concede the claim that this wasn’t an AD-driven crisis. Otherwise I’m standing pat.

George Selgin August 20, 2011 at 8:32 am

For “an AD-driven crisis” in my last comment kindly read “a partly AD-driven crisis.” I did not intend to imply either that there was no supply-side component, and especially didn’t wish to deny a role for Austrian-style post-boom readjustments.

jjoxman August 20, 2011 at 8:39 am

Prof. Selgin,

I’m not trying to concede the thesis. I’m merely pointing out to Daniel, and anyone else who is listening, that if one says the 1919-1921 depression was not, in part, an AD decline, then neither was the 2007-2009 depression.

George Selgin August 20, 2011 at 9:40 am

Fair ’nuff, jjoxman.

Previous post:

Next post: