I am currently a bit over half way through an excellent book entitled “American Default“, by Sebastian Edwards. The primary focus of the book is the abrogation of the gold clause in debt contracts, which (I believe) is the only time the US federal government actually defaulted on its debt. But the book also provides a fascinating narrative of FDR’s decision to devalue the dollar in 1933-34.  I highly recommend this book, which I also discuss in a new Econlog post. Later I’ll do a post on the famous 1935 Court case on the gold clause.

Edwards has an interesting discussion of the difference between Irving Fisher and George Warren.  While both favored a monetary regime where gold prices would be adjusted to stabilize the price level, they envisioned somewhat different mechanisms.  Warren focused on the gold market, similar to my approach in my Great Depression book.  Changes in the supply and demand for gold would influence its value.  Raising the dollar price of gold was equivalent to raising the nominal value of the gold stock.  Money played little or no role in Warren’s thinking.

Fisher took a more conventional “quantity theoretic” approach, where changes in the gold price would influence the money supply, and ultimately the price level.  Edwards seems more sympathetic to Fisher’s approach, which he calls a “general equilibrium perspective”.  Fisher emphasized that devaluation would only be effective if the Federal Reserve cooperated by boosting the money supply.

I agree that Warren’s views were a bit too simplistic, and that Fisher was the far more sophisticated economist.  Nonetheless, I do think that Warren is underrated by most economists.

To some extent, the dispute reflects the differences between the closed economy perspective championed by Friedman and Schwartz (1963), and the open economy perspective advocated by people like Deirdre McCloskey and Richard Zecher in the 1980s.  Is the domestic price level determined by the domestic money supply?  Or by the way the global supply and demand for gold shape the global price level, which then influences domestic prices via PPP?  In my view, Fisher is somewhere in between these two figures, whereas Warren is close to McCloskey/Zecher.  I’m somewhere between Fisher and Warren, but a bit closer to Warren (and McCloskey/Zecher).

There’s a fundamental tension in Fisher’s monetary theory, which combines the quantity of money approach with the price of money approach.  Why does Fisher favor adjusting the price of gold to stabilize the price level (a highly controversial move), as opposed to simply adjusting the money supply (a less controversial move)?  Presumably because he understands that under a gold standard it might not be possible to stabilize the price level merely through changes in the domestic quantity of money.  If prices are determined globally (via PPP), then an expansionary monetary policy will lead to an outflow of gold, and might fail to boost the price level.  Thus Fisher’s preference for a “Compensated Dollar Plan” rather than money supply targeting is a tacit admission that Warren’s approach is in some sense more fundamental than Friedman and Schwartz’s approach.

Warren’s approach also links up with certain trends in modern monetary theory, particularly the role of expectations.  During the 1933-34 period of currency depreciation, both wholesale prices and industrial production soared much higher, despite almost no change in the monetary base.  Even the increase in M1 and M2 was quite modest; nothing that would be expected to lead to the dramatic surge in nominal spending.  That’s consistent with Warren’s gold mechanism being more important that Fisher’s quantity of money mechanism.  In fairness, the money supply did rise with a lag, but that’s also consistent with the Warren approach, which sees gold policy as the key policy lever and the money supply as being largely endogenous.  You might argue that the policy of dollar devaluation eventually forced the Fed to expand the money supply, via the mechanism of PPP.

A modern defender of Warren (like me) would point to models by people like Krugman and Woodford, where it’s the expected future path of policy that determines the current level of aggregate demand.  Dollar devaluation was a powerful way of impacting the expected future path of the money supply, even if the current money supply was held constant.

This isn’t to say that Warren’s approach cannot be criticized. The US was such a big country that changes in the money supply had global implications.  When viewed from a gold market perspective, you could think of monetary injections (OMPs) as reducing the demand for gold (lowering the gold/currency ratio), which would reduce the value of gold, i.e. raise the price level.  A big country doing this can raise the global price level.  So Warren was too dismissive of the role of money.  Nonetheless, Warren’s approach may well have been more fruitful than a domestically focused quantity theory of money approach.

PS.  Because currency and gold were dual “media of account”, it’s not clear to me that the gold approach is less of a general equilibrium approach, at least under a gold standard.  When the price of gold is not fixed, then you could argue that currency is the only true medium of account, and hence is more fundamental.  During 1933-34, policy was all about shaping expectations of where gold would again be pegged in 1934 (it ended up being devalued from $20.67/oz. to $35/oz.)

PPS.  There is a related post (with bonus coverage of Trump!) over at Econlog.

My book on the Great Depression is officially being released on December 1st.  At that time I plan to do a few posts discussing the book.  But since some have already received copies, I thought it might help to provide a quick overview for what is a fairly complicated hypothesis.  The book is certainly not “easy reading”, and I’m hoping the following two charts will make it easier to follow the thread of the argument.  (In other words, unlike Garett Jones I didn’t take Bryan Caplan’s advice.)

Here is a flow chart in the first chapter:

The final chapter (13) includes a more detailed model, intended for serious scholars. In that chapter I also have a more detailed flow chart, which explains how all the pieces fit together:

If you get confused while reading the book, consult these flow charts.  They explain how each piece of the puzzle fits together.  When I wrote the book (it was done by about 2005) I had no idea that more than 50 people would ever read it, so I wrote it at a level aimed at specialists.  Now it seems that many more people than I expected will read the book. Obviously I’m pleased by the interest, but quite honestly if you only have time to read one economics book this year, make it Garett Jones’s Hive Mind, not my book.

The following isn’t Brad DeLong’s entire summary of Greenwald-Stiglitz, but it gets at the central assumption:

However, even though I do not fully buy it I do think I understand the argument. And I do not think it is incoherent.

As I understand the Greenwald-Stiglitz hypothesis–about the Great Depression as applied to agriculture and about today as applied to manufacturing–it goes like this:

1. Rapid technological progress in a very large economic sector (agriculture then, manufacturing now) leads to oversupply and steep declines in the sector’s prices. Poorer producers have less income. They come under pressure to cut back their spending. Others–consumers–are now richer because they are paying less for their food (or their manufactures), but their propensity to spend is lower than that of the stressed farmers or ex-manufacturing workers.
2. Moreover, the oversupply of agricultural commodities (or manufactured goods) means that only an idiot would invest at their normal pace in those sectors. To the shortfall in consumption spending is added a shortfall in investment spending as well.
3. Thus we have systematic pressures pushing spending down below economy-wide income. These aren’t going to go away until the declining sector (agriculture then, manufacturing now) is no longer large enough to be macroeconomically significant.
4. Macroeconomic balance requires that the economy generate offsetting pressures pushing spending up. What might they be?

First let’s translate this into a monetarist framework, and then we can examine what’s wrong.

Even the Keynesian model requires a big drop in NGDP (below trend) to get a demand-side recession.  So how does the G-S hypothesis do that?  The Keynesians would argue that if the central bank held the money supply fixed, these shocks would cause a fall in velocity.  That’s not an unreasonable assumption, as DeLong is talking about a situation where desire to save rises relative to desire to invest.  That does reduce interest rates.  Lower interest rates mean a lower opportunity cost of holding base money, and thus lower velocity.  So far so good.

My complaints lie elsewhere.  On empirical grounds almost every single step in this argument is wildly implausible.  And that’s not hyperbole; I don’t mean one step, I mean every single step:

1.  I see no evidence that technological change in the farm sector that pushed farmers toward the city would boost aggregate saving rates or reduce the propensity to invest.  This process was going on continually from the late 1800s, until it began slowing in recent decades because there were so few farmer left.  Market economies are really good at adapting to these slow and predictable types of creative destruction.  Perhaps there was less investment in the farm sector, but there was more investment in the urban sector.  And was there really less investment in the farm sector?  Farmers were moving to the cities in the 1920s precisely because farming was becoming more mechanized, as machines were replacing workers.  So I’m not sure the 100 year trend of farmers moving to the cities depressed real interest rates at all.

2.  But let’s suppose I’m wrong.  Business cycles aren’t caused by 100 year trends, they are caused by “shocks.”  Where is the shock?  If each year 2% of farmers move to the city, how does that cause a sudden demand shock?  The economy was booming in the 1920s despite the gradual decline in farming.  Now you could argue that something else was propping up the economy, like a thriving manufacturing sector, and when the bottom fell out of manufacturing then the economy slumped.  But in that case why not blame the collapse of manufacturing?

3.  Stiglitz claims things got much worse in farming in the early 1930s, citing evidence that incomes fell between 1/3 and 2/3 (which doesn’t seem very precise data to build a theory around.)  But total national income fell by roughly 1/2, right smack dab in the middle of the Stiglitz estimates.  So what’s so special about farming?

4.  Let’s say everything I said was wrong.  Let’s say DeLong is right that the decline in farming during the 1920 gradually led to a saving/investment imbalance, which eventually got so bad it triggered the Great Depression.  How would this show up in the data?  We would see falling real interest rates.  When they got close to zero the real demand for base money would soar, triggering a sharp fall in NGDP (unless offset by lots of money printing.)  And something like that did happen in the early 1930s.  But the problem is that it didn’t happen in the 1920s.  After the 1920-21 deflation, prices were pretty stable for the rest of the 1920s.  So nominal rates should give us a rough estimate for real rates.  (I’d add that expected inflation rates were usually near zero when the dollar was pegged to gold.)  During the 1920s short term nominal interest rates fluctuated in the 3.5% to 6.0% range.  Those are strikingly high risk free real rates by modern standards.  Even worse for Stiglitz, they were trending upward in the latter part of the 1920s.  Thus there’s not a shred of evidence that the migration away from farms during the 1920s had the sort of macro implications that are necessary for the Stiglitz model to be plausible.

I suppose some would try to resurrect the model by pointing to all sorts of “ripple effects.”  How problems in agriculture spilled over into other sectors.  The problem with this approach is that it proves too much.  There has only been one Great Depression in US history.  (In real terms the 1870s and 1890s weren’t even close.)  If capitalism is so unstable that a problem in one area causes ripples which eventually culminate in a Great Depression, then one might as well argue the Depression was caused by my grandfather sneezing.   His sneeze passed a cold to several other people, and voila, via the “butterfly effect” we eventually get the collapse of the world economy and the rise of the Nazis.  I happen to believe that any useful model has to be more than “coherent” in a logical sense, it also has to be empirically plausible.

Suppose someone walked up to you in mid-1929 and said a depression was on the way for reasons outlined by Stiglitz.  What would you think?  What data would support that conclusion?  Did the economy seem to be having trouble accommodating farmers gradually moving to the city?  No.  Was there a savings glut?  No.  Was the real interest rate trending downward?  No.  Were there a host if exciting new technological developments that would lead one to be very excited about the future of American manufacturing?  Yes.  You’d ask Stiglitz why we should believe his model.  What pre-1929 facts was it able to explain?  As of 1929 I don’t see it explaining anything.  Of course we did have a Depression, which is exactly what you’d expect if:

1.  The Fed, BOE, and BOF all tightened in late 1929, sharply raising the world gold reserve ratio over the next 12 months.

2.  Then falling interest rates and bank failures increased the demand for base money after October 1930.

3.  Then international monetary collapse and more bank failures led to more demand for both cash and gold after mid-1931.

4.  Then FDR raised nominal wages by 20% overnight in July 1933, aborting a promising recovery in industrial production.

5.  Then in 1937 the Fed doubled reserve requirements and sterilized gold, slowing the economy.

6.  Then when the economy slowed a dollar panic (fear of devaluation) led to lots of gold hoarding, sharply depressing commodity prices world-wide in late 1937.

That would be a theory with explanatory power.  Something Stiglitz’s theory lacks.

Of course I haven’t even discussed the much deeper problems with the Stiglitz worldview.  Because he insists on a “real” theory of AD, he has no explanation for movements in NGDP.  I’d guess this is where DeLong would part company with Stiglitz.  In Delong’s worldview there is a trend rate of inflation high enough to prevent liquidity traps, and hence (unless the Fed crashes the monetary base) high enough to prevent collapses of NGDP.  Not in Stiglitz’s world.  Although he often uses the language of Keynesianism (aggregate demand, etc) his model is in some ways even more primitive, like the early progressive models that Keynes pushed aside during the 1930s.  Recall that Keynes saw the problem as the failure of our monetary system.  As Nick Rowe likes to say, the problem isn’t saving, it’s money hoarding.

PS.  It seems to me that DeLong mildly scolds Nick Rowe for roughly the same reason that Nick Rowe scolds Bryan Caplan.  I agree with Caplan and Nick Rowe.  That is Nick Rowe the victim, not Nick Rowe the villain.

🙂

Joe Stiglitz has a new article, where he continues to develop his rather unconventional view of business cycles.  Is his theory a real theory or a demand side theory?  I can’t quite tell, maybe one of my commenters can help me out.

Of course the most important stylized fact of the Great Depression was the horrific collapse in industrial production between 1929 and 1933.  And what caused this collapse?  Apparently productivity improvements in the farm sector.  Productivity gains that were also occurring in the booming 1920s.

And how does Stiglitz know this?

At the beginning of the Depression, more than a fifth of all Americans worked on farms. Between 1929 and 1932, these people saw their incomes cut by somewhere between one-third and two-thirds, compounding problems that farmers had faced for years. Agriculture had been a victim of its own success. In 1900, it took a large portion of the U.S. population to produce enough food for the country as a whole. Then came a revolution in agriculture that would gain pace throughout the century””better seeds, better fertilizer, better farming practices, along with widespread mechanization. Today, 2 percent of Americans produce more food than we can consume.

Interesting, but this has been going on for 120 years.  So why was the economy booming in 1929, and flat on its back in 1932?  And why doesn’t Stiglitz discuss what happened to the incomes for the other 80%, the people not in farming?  It turns out that their incomes also fell “between one-third and two-thirds,” indeed nearly 50% by the fourth quarter of 1932.   So nothing particularly interesting was going on in farming, and yet this somehow explains the far greater collapse in output in manufacturing, mining, and construction.  There may be a model there, but I don’t see it.

What about the conventional explanation, that it was an adverse demand shock?  The view that errors of omission or commission by the Fed explain the 50% fall in NGDP:

Many have argued that the Depression was caused primarily by excessive tightening of the money supply on the part of the Federal Reserve Board. Ben Bernanke, a scholar of the Depression, has stated publicly that this was the lesson he took away, and the reason he opened the monetary spigots. He opened them very wide. Beginning in 2008, the balance sheet of the Fed doubled and then rose to three times its earlier level. Today it is $2.8 trillion. While the Fed, by doing this, may have succeeded in saving the banks, it didn’t succeed in saving the economy. Reality has not only discredited the Fed but also raised questions about one of the conventional interpretations of the origins of the Depression. The argument has been made that the Fed caused the Depression by tightening money, and if only the Fed back then had increased the money supply””in other words, had done what the Fed has done today””a full-blown Depression would likely have been averted. In economics, it’s difficult to test hypotheses with controlled experiments of the kind the hard sciences can conduct. But the inability of the monetary expansion to counteract this current recession should forever lay to rest the idea that monetary policy was the prime culprit in the 1930s.

.   .   .

Monetary policy is not going to help us out of this mess. Ben Bernanke has, belatedly, admitted as much.

Readers of Vanity Fair are being led to believe that Bernanke went into this downturn thinking that monetary stimulus was the answer to depressions, and then had a sort of epiphany that monetary stimulus doesn’t work.  Is this really true?  Has Bernanke suddenly become an adherent to the view that the Fed is out of ammunition?  Or that they have ammo, but that nominal growth can’t solve real problems (as RBC adherents believe.)  I’ll pay $100 to the first person who can convince me that Bernanke believes monetary policy can’t boost AD, or that he believes AD can’t boost output.  (I should offer $10,000, but proportional to wealth my offer is just as generous as Mitt Romney’s.)

Stiglitz is a distinguished Nobel Prize winner.  But he didn’t win for business cycle theory.  I doubt even his fellow progressive Paul Krugman could make heads or tails out of Stiglitz’s essay.  That doesn’t make Stiglitz wrong (I’m also a contrarian.)  But if you are going to make that sort of argument, you need more evidence than farm incomes falling during the Great Depression.

Paul Krugman likes to show a graph indicating that each country began recovering from the Depression right after it adopted expansionary monetary policies (i.e. currency devaluation.)  If Stiglitz has an explanation for that, I’d love to hear it.

Back in the 1930s many people thought the Great Depression was caused by too much output.  This led FDR to adopt programs aimed at reducing output (such as the AAA and the NIRA.)  They “worked.”  Today economists tend to scoff at such ideas, as falling output is essentially the definition of a depression.  But not Stiglitz:

Throughout the 1930s, in spite of the massive drop in farm income, there was little overall out-migration. Meanwhile, the farmers continued to produce, sometimes working even harder to make up for lower prices. Individually, that made sense; collectively, it didn’t, as any increased output kept forcing prices down.

HT:  Gregory Barr, Larry

Update: I see Ryan Avent is equally puzzled by Stiglitz’s model.  And so is Nick Rowe.  Will Krugman give Stiglitz the Fama/Lucas/Cochrane treatment?