Three Octobers

I’m referring to 1929, 1937, 2008, which all saw severe stock market crashes, accompanied by falling commodity prices.  We can better understand our current crisis if we first step back and look at the two earlier October crashes, which bear some interesting resemblances to recent events.

Although in each case the problem was “monetary” broadly defined, in none of these three episodes can modern monetary economics easily identity the problem.  In contrast, the monetary model sketched out in the previous post will allow us to see the subtle forces that pushed the economy into severe recession.  For instance, in 1929 the problem was central bank hoarding of gold, in 1937 it was private hoarding of gold, and in 2008 it was banks hoarding reserves.

There were also some important differences between these three crashes.  The first was a pure demand shock, not entangled with significant financial or supply-side problems.  In 1937 the banking system was stable, but the economy experienced a severe wage shock.  And 2008 saw a significant oil shock and a severe financial crisis.

What strikes me most about these three episodes, however, is the strong sense one gets from the financial press that expectations changed dramatically around October of each year.  There is a sense that investors suddenly and dramatically downgraded their forecasts of nominal GDP growth going forward for at least 2 or 3 years.  Of course even today newspapers almost never mention “nominal GDP,” but in all three Octobers there was a palpable sense that inflation expectations and real growth expectations were falling sharply.  For instance, in mid-1929 the economy was in the midst of a powerful boom, and yet already by December there were many ominous reports that, unlike recent recessions, the slump would be extremely severe.  Industrial production fell at an extraordinary rate in the 4th quarter.  In both 1937 and 2008 the dramatic fall in the stock market was accompanied by very bearish news from commodity markets (where prices fell precipitously), the the Treasury bond market, where short term rates fell to near-zero levels. And once again, output fell at extremely rapid rates, even compared to other downturns

One might ask whether the three stock market crashes triggered the subsequent recessions.  I doubt it, as the equally big October 1987 crash (what is it about October!) was not followed by a recession, or even a mild slowdown in the economy.  Yes, each case is different, but if stock crashes exerted a powerful independent effect on growth, wouldn’t one expect to see at least a tiny blip down after the 1987 crash?

October 1929

In my manuscript on the Depression I am skeptical that we will ever be able to fully explain the 1929 crash, but I do think we can offer a partial explanation.  The key problem was monetary policy.  Under a gold standard the gold reserve ratio (gold/monetary base) is the only truly exogenous policy instrument.  Between December 1926 (when France returned to gold), and October 1929, this ratio rose at about a 2.5% annual rate worldwide–a mildly deflationary policy.  The world price level had a slight downward trend during this period.

Then, in the next 12 months, the ratio soared by 9.6%, pushing most of the developed world into severe deflation and depression.  For those who missed my course on money in the previous post, central bank hoarding of gold increases its real value.  Under a gold standard, gold is the medium of account.  So a rise in its real value (or purchasing power) means deflation.

The causes of this are too complex to cover here.  Nor is it clear whether the markets knew what was going on.  My hunch is that they understood at an intuitive level that central banks had stumbled unknowingly into a highly deflationary policy stance, albeit probably without visualizing the problem in terms of the gold ratio.  Unlike in the 1930-38 period, when it is often easy to connect up major stock market movements with policy shocks, I had trouble finding the “smoking gun,” evidence that tight money caused the crash.  Still it is at least interesting that the highly contractionary monetary policy that triggered the Depression seems to have begun around the time of the October crash.  Indeed I believe that I am the only researcher to find this link.

In the end, I found 4 factors that probably contributed to the crash.  In order of importance they are:

1.  A monetary policy stance was adopted that led to a large increase in the world’s gold ratio, especially in the US, France, and Britain.

2.  The fierce Congressional fight over Smoot-Hawley, which was the headline news story in the NYT around the time of the crash.  Jude Wanniski made this argument, and I think he is partly right, albeit for somewhat different reasons.

3.  The death of German statesman Stresemann in early October, and the deteriorating political situation in Germany in late October.  (Right wing nationalist parties gained ground, threatening a recently crafted international war debt agreement.)

4.  An October 26th administration decision to crack down on mergers (for anti-trust reasons.)  George Bittlingmayer says the news may have reached the markets a few days earlier.

I have my doubts about how important these factors were to the stock market in October 1929, but in retrospect the first three factors should  have been very important.  Tight money depressed the economy until March 1933.  A renewed fight over Smoot-Hawley in the spring of 1930 clearly caused a major stock market crash in May and June.  And between mid-1931 and the end of 1932 the German crisis was probably the major factor depressing the U.S. stock market.

October 1937

The 1937 October crash (which like the 1929 and 2008 crashes actually lasted from September to November) was just as interesting as 1929, and even more complex.  First we need to consider a few basic facts:

1.  In 1937 the U.S. was back in the international gold standard.  It wasn’t much of a standard (with only a few countries), and Americans could no longer own gold (officially—but they hoarded it in London banks.)  Nevertheless, our monetary policy was now at the mercy of shocks to the international gold market.  Put simply, an orgy of gold dishoarding caused the US WPI to soar about 10% between mid-1936 and mid-1937, despite the depressed economy.  Then between mid-1937 and mid-1938 it fell back my an equal amount.  By itself, this would have caused a very strong period of growth in late 1936 and early 1937, and a mild recession thereafter.  But that is not exactly what happened.  The very strong growth in industrial production in 1936 leveled off in the spring of 1937, despite the expansionary monetary policy.

2.  The second factor was the third of FDR’s five wage shocks, and the one least directly connected to government policy.  Toward the end of 1936 and throughout much of 1937 union membership soared.  The 1935 Wagner Act had made it much easier to organize unions, and FDR’s overwhelming victory in 1936 assured union organizers that Washington would support them in their battles.  The resulting wage shock looks a lot like the WPI price bubble, except shifted a half year into the future.  Wages started rising later, peaked later, and fell later than the WPI.  In addition they fell by much less that prices.  This is why output didn’t grow even faster in early 1937, and it’s also the factor that turned what would have been just a mild recession from the pullback if commodity prices, into a deep depression–comparable to 1920-21.

The real wage rate (nominal wages divided by the WPI) tracks monthly industrial production very closely throughout the 1930s, and 1936-38 is no exception.  (Actually you have to invert the real wage series—as wages were countercyclical.)  The economy boomed when real wages fell (due to the WPI rising faster than nominal wages, and fell when nominal wages rose much faster than prices.  Some growth occurred when real wages were flat.  So 1936-38 fits my overall model of the Depression.  Then it all comes down to explaining wages and the WPI.  The main exogenous factors influencing wages were the five New Deal wage shocks discussed in an earlier post.  The WPI can be explained with a model of the international gold market when the price of gold is fixed (1929-33, 1934-40) or by changes in the price of gold itself in 1933-34—as in my Warren post.

Working with this model, I found that there were three factors pushing up prices in 1936-37.  Central bank dishoarding of gold increased in late 1936, as countries such as France abandoned the gold standard.  Another factor was increasing Russian gold sales, along with the (incorrect) expectation that much more was coming from Russia.  And this also led to private dishoarding, motivated by both the collapse of the gold standard in Europe (no longer a need to hoard gold in expectation of future devaluation), and fears that the huge gold flows to the US would prove so inflationary that FDR would be forced to revalue the dollar upward.  This is what the Paul Einzig quotation in my “worse than economists expected. . . ” post was referring to.

For very complex reasons this process started to reverse in mid-1937.  As rising wages slowed the US economy, revaluation fears disappeared.  The Russian gold sales that were expected never fully materialized.  Gradually, gold dishoarding gave way to gold hoarding, as people saw that the U.S. was again sliding back into depression and investors worried that FDR might seek another monetary “shot in the arm,” just as in 1933.  By the fall, devaluation fears grew rapidly and gold hoarding increased sharply.  Gold inflows to the U.S. virtually ceased.  Under the burden of high wages and sharply falling prices the U.S. slid into a deep depression, and stock and commodity prices fell sharply.

I find several interesting parallels between 1937 and 2008.  Both years started with rapidly rising commodity prices all over the world, and ended with prices falling just as rapidly.  One difference was that the 1936-37 boom was purely monetary, as the economy was still somewhat depressed.  In contrast, the 2007-08 boom was driven by strong growth in developing countries.

An even more interesting parallel is the sharp reversal of expectations of medium term growth and inflation.  In the first half of 1937 it was very obvious that both the financial markets and the press expected inflation to continue due to the expected swelling of the US gold stock.  By the end of 1937 expectations had changed radically, and there was a perception (which turned out correct) that we were headed for a prolonged period of deflation.  The WPI did decline almost continuously from mid-1937 to mid-1940.

When a similar sharp reversal of expectations occurred in October 2008, I immediately recalled the events of 1937 and began to fear that the trajectory of nominal growth could fall far below the 5% norm unless the Fed acted quickly and effectively.  They did act quickly, but focused on the wrong problem—banking instability.

But the comparisons don’t stop there.  In 1936-37 the Fed doubled reserve requirements in three steps.  Given my recent criticism of the Fed’s interest on reserves policy (which raises the demand for reserves), you might expect me to blame the 1937-38 depression on this mistake.  But that is too easy.  Investors initially paid little attention to the moves, as they lacked credibility.  The financial press indicated that the huge gold flows would overwhelm any Fed attempts at sterilization, and speculators drove commodity prices sharply higher in expectation that the Fed would be unable to restrain growth in the money supply.

The Fed’s big mistake didn’t come until late 1937, when they did not act aggressively enough after expectations turned around in the gold market.  But (unlike 1929) monetary policy was not the only culprit in 1937, the wage shock also played a major role in the depression.

Someone will probably ask me about fiscal policy.  The payroll tax instituted in 1937 was fairly small, but it did play a modest role in the wage shock, as it increased the cost of labor to employers.  Beyond that, the role of fiscal policy was minor.  To the extent that the 1937-38 depression was due to lack of demand (let’s say the depression was 50% wages and 50% deflation for the sake of argument), fiscal policy played at best a very minor role.  The sharp deflation of 1937-38 was clearly linked to a turnaround in the world gold market, just as the previous inflation of 1936-37 had been.  The timing is all wrong for fiscal policy.  Like the monetarists, Keynesian economic historians overlook the need to link up their account of policy mistakes with contemporaneous movements in the commodity, stock, and bond markets.  I believe that I am the only one to provide that sort of comprehensive analysis of the Great Depression.

October 2008

There are many factors that played a role in the monetary policy errors that led to a sharp fall in stock markets, commodities markets, industrial production, and bond yields in the fall of 2008.  The two most important were the commodity price boom and the financial crisis.

Notice I don’t say the commodity boom and financial crisis caused the fall in AD, I said they led to the monetary policy errors that caused this collapse.  In the first half of 2008 growth was modest, but positive.  Stocks held up pretty well, still within about 10% of their record highs in early June.  The world economy was even stronger, which explains the extraordinary commodity price boom that peaked in July.  It was the second half of the year when the key mistakes were made.  It is useful to distinguish between the 3rd and 4th quarters, and also two types of policy mistakes.

In the 3rd quarter policy continued to be optimal in the Lars Svensson sense; the internal Fed forecast for growth in nominal spending was close to the Fed’s target.  This is shown by the fact that the Fed still viewed the risks of inflation and recession as being roughly equally balanced in their September 16th meeting, when they decided to stand pat.  But the private sector was probably already ahead of the Fed in understanding what was going on.  We now know that nominal growth slowed sharply in the 3rd quarter.  Industrial production (which had held up for two years of housing recession), began falling sharply in August.   Basically the Fed made two mistakes in the third quarter.  First they put too much weight on internal forecasts and not enough weight on market forecasts.  And second, they focused too much on inflation, and not enough on nominal GDP growth.  (Notice how NGDP consistently provides better signals to policymakers than inflation—recall the 2004-06 housing bubble as well.)

In the 4th quarter of 2008 things immediately got much worse.  By mid-October it was obvious to me (and I think to almost everyone) that monetary policy had lost credibility even by the looser Svenssonian standards—that is, even the Fed’s internal forecast was now far below the policy target.   Even worse, the markets understood that rates were approaching zero and that the US could easily slide into a Japanese-style situation, which could persist for years.  The only way to prevent that from occurring would be through aggressive steps to enact unconventional monetary policies—it was probably too late to simply swap base money for T-bills.

Unfortunately, the Fed made two key mistakes.  First, after going 5 months without a rate cute, the Fed adopted the highly deflationary policy of paying interest on reserves, which negated any expansionary effects their base expansion might have had.  But even worse, they got distracted from monetary policy under the erroneous assumption that the financial crisis was the “real problem,” and had to be cured before AD could be boosted.  This is certainly an understandable mistake.  I recall in October 2008 being one of the very few economists who worried that the real problem was monetary, not financial.  And the financial crisis certainly played a big role in the recession in two ways.

1.  It distracted policymakers.

2.  It reduced the Wicksellian equilibrium rate required for macro stability.

The crisis depressed the effective demand for credit so sharply that fears of a liquidity trap developed.  It didn’t cause the sharp fall in AD, but it made the Fed’s job much harder.  But it is important to remember that the Fed itself claims their job is to stabilize prices and output, they themselves say their duty is too offset shocks to money demand and velocity, they themselves acknowledge that their responsibility to be proactive, especially when there is a threat of deflation, they are the ones who have said that market indicators also provide useful signals.  Bernanke has written that monetary policy can still be effective at zero interest rates.  They failed, even according to their own criteria.

This quick summary glossed over many important details.  My chapter on 1929-30 is roughly 40 pages.  The two chapters on 1937-38 are even longer.  The manuscript is 14 chapters.  You still need to buy my book (if it ever finds a publisher.)  🙂

I will continue to touch on other aspects of the Depression in this blog.



12 Responses to “Three Octobers”

  1. Gravatar of Phil P Phil P
    8. March 2009 at 13:05

    Scott, thanks for another fascinating post. I do want to comment on one thing. I do believe that the 1929 stock market crash did worsen the ongoing contraction by affecting business confidence (and maybe for other reasons as well). You dismiss it by comparing it to 1987. Peter Temin made the same point in one of his books and it struck me then as a superficial comparison. First, stocks were much more overvalued in 1929 than in 1987. I know the idea that there was a bubble in 1929 is still controversial in some quarters, but it doesn’t really matter, because there was widespread controversy at the time about speculation and the level of stock prices. The optimists, like Irving Fisher painted rosy pictures of future growth. When the crash came, it must have struck many businessmen that the skeptics were right, and that they’d better adjust their forecasts. Very important in this connection is that the economy was already in recession. This was not as obvious at the time as it is to us today, but there were indications of declining business activity that were reported in the press (e.g. steel, autos and commodity prices). By the way, I don’t know why you ignore the economic indicators as a factor in the crash.

    By contrast, in 1987 there was no concern about a possible stock market bubble and no ongoing recession. The crash was mostly a one day affair and it was recognized as being partly due to technical factors (i.e. portfolio insurance). If the crash created concerns at all it was only because of memories of 1929. But there weren’t enough objective reasons for concern to cause businesses to overreact.

    There are doubtless good arguments for questioning the importance of the 1929 crash in the subsequent Depression, but I don’t think the 1987 comparison is one of them. Both the psychological and objective factors were too different.

  2. Gravatar of ssumner ssumner
    8. March 2009 at 13:51

    Phil P, I partly agree and partly disagree. First a few factual issues. The 1987 crash was not a one day affair. If you overlay the graphs from 1929 and 1987 they almost perfectly match. Stocks start falling in early September, the decline gradually picks up speed, culminating in a 23% decline in 2 days (1929) or a 22% decline in one day (1987). The the patterns were almost identical. You are right that the economic indicators were declining even before the crash, and indeed in my manuscript I argue that the crash of 1929 was partly a reaction to those declines. But I interpret those indicators differently from most people. The Fed had been gradually raising interest rates in an attempt to restrain the bull market. Finally in September the policy (along with tightening in Britain), started to reduce both prices and output significantly. When both prices and output fall in an otherwise booming economy with no supply disturbances, there is really no other explantion–falling AD caused by tight money. So I think the markets looked at the indicators you cite, and realized “uh oh, looks the the Fed’s tight money policy has finally got traction.”
    Then when the stock market crashed things even got worse. A stock crash causes a fall in the Wicksellian natural rate required for macroeconomic stability. So this fits in with your theory–the stock crash helped cause the Depression. In a sense I agree but in a sense I disagree. The reason why it helped cause the Depression, is that it made monetary policy effectively tighter–the Fed’s target rate was now far about the Wicksellian natural rate. That’s tight money by many people’s definition.
    Here’s another way to put my point. Sometimes when I make the 1929–1987 comparison people will say “but in 1987 the Fed cut rates to prop up the economy.” OK, but if this is what prevented a recession in 1987, then it suggests monetary policy is pretty important. The sort of Fed rate cuts in 1929 that would have been required for it to adhere to the “rules of the game,” a stable gold ratio, would have been very dramatic. The Fed would have had to cut rates sharply. But it should have done so. It should have adhered to the rules of the game and kept the gold ratio stable. Had it done so, there would have been no Great Depression in my view. So yes you could say I am back to my “errors of omission” argument I use against the current Fed. But I think it is fair, I didn’t come up with the rules of the game, I’m not even in favor of the gold standard. It was the conservatives of that era that caame up with that benchmark. Well, by their own benchmark they were highly contractionary, and if they had not been it might have been merely another 1987.

    To summarize, much of your criticism is exactly right, but there is another way of framing my argument that makes it stand up, even if your specific points are correct.

  3. Gravatar of Winton Bates Winton Bates
    8. March 2009 at 20:08

    I wonder whether you agree with John Cochrane’s view that the financial crisis is analogous to several major oil refineries being blown up, with tankers full of oil sitting in the harbor, low oil prices and high gas prices (i.e. large risk premiums). He suggests that the right policy response would be to run whatever government refineries could be cobbled together on short notice to run at full speed i.e. to focus on making sure funds flow to business and consumers. See:

    This reasoning seems to suggest that it would be possible to get a larger expansion in AD per dollar spent by buying corporate bonds rather than government bonds. Does that make sense?

  4. Gravatar of Alex Golubev Alex Golubev
    9. March 2009 at 07:25

    Winston – right on. man, i gotta catch up on my readings today! my analogy is a clogged … let’s say SINK and we keep pouring more water into it to get it unclogged. and the … funniest/saddest thing is that instead of using draino, we’d rather keep the nasty crap that’s plugging the drain. Of course this is too simplistic. LEVERAGE is the huge difference. it’s like sleeping with your partners’ partners. i think both analogies don’t quite capture it and i can’t work sinks into this one. ok i gotta catch up on teh rest of the new posts.

  5. Gravatar of Phil P Phil P
    9. March 2009 at 09:19

    Scott, thank you for taking so much trouble in your reply to my comment. I was aware that in 1987 stocks had declined for several days before Black Monday, but I still think the crash was viewed more as a stock market event than a harbinger of possible depression – at least that’s how I remember it. I pretty much agree with what you say about 1929 now that you’ve clarified it for me. I do think that your “acts of omission” causality lends itself to misunderstanding – a rhetorical rather than an analytical failure. Consider the 2008 crisis. You reject the idea that the commodity boom and financial crisis “caused” the fall in AD, only that it led to monetary policy errors. But then what caused the decline in industrial production in August, which the Fed failed to anticipate? The damage that Hurricane Katrina did to New Orleans might have been prevented if they had built the levees higher, but it was still “caused” by the hurricane. Perhaps Bernanke failed to rise to the occasion last year, but he wasn’t the hurricane.

    Why not just say the commodity boom and financial crisis started a recession and the Fed then misdiagnosed the symptoms and adopted the wrong remedies? It would be a lot clearer, unless I’m misunderstanding the substance of what you are saying.

  6. Gravatar of DanC DanC
    9. March 2009 at 11:11

    I would like some clarity.

    Assume that output will, in the long run, move to potential output. Assume that regulations and taxes can shift that AD curve to the right at a point below potential output. A simple version of supply side economics that assumes that government interventions and taxes can negatively impact the economy.

    Then assume that monetary policy can shift AD to the right but fiscal policy is pushing the curve to the left. How do you know which push has the stronger impact?

    Simply, you seem to quickly discount the negative impact of fiscal policies on output. I’m not sure why.

  7. Gravatar of DanC DanC
    9. March 2009 at 12:03

    sorry, regulations and taxes shift AD curve to left

  8. Gravatar of DanC DanC
    9. March 2009 at 14:52

    I’m very sorry. I also meant LRAS = Y* = Vertical line at full employment.

    Sorry for such a poor post

  9. Gravatar of ssumner ssumner
    9. March 2009 at 16:39

    Winton, I have read the paper and I see John’s point. Here is how I would put it. I think we can solve the problem without getting into buying corporate debt. But I also believe in moving fast, not waiting around. If the markets didn’t find my plan credible I would absolutely do what he says if necessary. So I agree with his technical analysis. BTW, in an email he said some very nice things about my futures targeting idea–although he also said he’d have to study it more–so I don’t want people assuming he endorsed it.

    PhilP, If you ever read my Great Depression book (which I now feel will eventually find a publisher) you’ll get a sense of how complex I view the whole causality question as being. One way of thinking about causality is as policy counterfactuals. If the optimal monetary policy could have prevented X, then can monetary policy be said to have caused X? I realize that may seem too high a bar, and I want to stake out a more moderate position here. In principle, I favor 5% nominal GDP growth. But although the Fed only got 3.3% in 2007.4-2008.2, I am on record saying their response to the sub-prime crisis was reasonable. So I hope that shows I’m not nitpicking too much. But after three quarters of slow nominal growth my feeling is they should have been a bit more aggressive after midyear, right at the time commodity prices started falling fast. But I have also given them a bit of a break on 2008.3, arguing that it was only in 2008.4 that we can clearly, unambiguously, say policy lost credibility. So I’d say money was a bit too tight in 2008.3, but if someone wants to say the slowdown was initially supply-shock/financial driven, I’d say that’s also a reasonable conjecture given the ambiguous data. And perhaps it might have been some of each. Indeed in probably was. As an aside, the recession technically began in December 2007, and I have written that although their quarter point cut in December was too little, they effectively caught up in January, and the recession was still theoretically avoidable up to mid-2008. Now saying it was theoretically avoidable doesn’t mean it was the Fed’s fault. They can’t be perfect. Can you buy this position:

    1. The initial mild recession was caused by financial/supply-side factors.
    2. The sharp intensification after September was caused by monetary policy being too contractionary relative to what was needed.

    That’s not exactly how I put it, but its actually not far off either. It’s the fourth quarter that I really feel passionate about.

    DanC I agree that fiscal policy and regulation can reduce AS. Right now I am more focused on the 6-7% fall in NGDP. But your point will be more important in the long run. In addition, and I have generally steered away from this complexity, adverse supply shocks can indirectly reduce AD, by lowering the Wicksellian natural rate of interest (assuming the policy rate is unchanged.)

  10. Gravatar of Jon Jon
    9. March 2009 at 20:20

    Nice review.

  11. Gravatar of 三つの10月 by Scott Sumner – 道草 三つの10月 by Scott Sumner – 道草
    11. February 2011 at 04:21

    […] 三つの10月 by Scott Sumner // スコット・サムナーのブログの初期エントリをぼちぼち翻訳紹介するシリーズ。今回は、”Three Octobers“(8. March 2009)です。この前日のエントリが「金融政策の短期講習」だったのですが、その準備をした上で書き上げたこちらこそが渾身の本命エントリと位置づけられるでしょう。 […]

  12. Gravatar of スコット・サムナーのマネー観、マクロ観 by Scott Sumner – 道草 スコット・サムナーのマネー観、マクロ観 by Scott Sumner – 道草
    7. May 2011 at 23:59

    […] a.  1929年(邦訳)は主要な中央銀行が実施した金融引き締めがNGDP急落期待を引き起こし、それが株式市場クラッシュの原因となった。NGDP期待の下落を伴わない同規模の市場クラッシュ(例えば1987年)は不況に至っていない。まして恐慌にも。   […]

Leave a Reply