Saturday, January 3, 2015

The End of One Big Inflation and the Beginning of One Big Myth

I recently had a short back-and-forth with Noah Smith on the subject of Tom Sargent's article "The Ends of Four Big Inflations," in which I called it "terrible history and questionable economics." The article recounts the post-World-War-I monetary histories of Austria, Hungary, Poland, and Germany, which they all experienced hyperinflations, on the one hand, and Czechoslovakia, which did not, on the other.

That article isn't widely read by academic economists now because its historical, anecdotal approach is out of fashion, for good reasons and bad. But it is nonetheless highly influential. It and others of Sargent's publications form a sort of backdrop to debates about monetary policy that I would argue still have a substantial influence on the way economists imagine the world to work, and certainly in what they teach undergraduates.

The most fundamental problem with that article is that the titular Four Big Inflations were actually one big inflation, and that big inflation was caused by the near-state-collapse embodied in the Treaties of Versailles, St. Germain-en-Laye, and Trianon. In addition to denuding the defeated Central powers of a huge amount of territory, armaments, and industrial capability through direct transfers to the victorious Allies, they imposed reparations on the losers way beyond their economies' capacities. The classic ex-ante study of this is The Economic Consequences of the Peace, by J. M. Keynes, which holds up remarkably well (as does *almost* everything he wrote), notwithstanding revisionist historiography that has questioned Keynes' contention that Germany in particular was capable of financing reparations on such a vast scale. The revisionism is grossly teleological: it takes the eventual Nazi takeover as its starting point in an attempt to show that it could have been prevented if only the Allies hadn't been suckered by the Long Appeasement.

In economic terms, Keynes' argument is that the reparations in the Treaty of Versailles exceeded what modern economists would call the Laffer Peak: the maximum amount of government revenue that can be extracted from the productive economy using distortionary taxation. In political terms, Keynes was outraged that David Lloyd George and, in particular, Georges Clemenceau, discredited by the supremely costly victory they imposed on their own citizens, sought to use the treaty to salvage their own political lives by shifting the ruinous cost to the defeated enemy. To do so, they successfully outmaneuvered an out-of-his-league, politically weakened President Woodrow Wilson. A particularly important aspect of Keynes' argument is the tension between France and Poland's abject public finances, only made paper-solvent by ambitious reparations schedules and hence a productive postwar German economy with a large tax base, and the political impetus to simply steal Germany's entire mining, railway, shipping, and manufacturing capacity and give it to France. Keynes convincingly argued that either way, there just wasn't enough German capital or German workers willing to toil in order to save Clemenceau's reputation with his fellow Frenchmen.

In normal times, public finance had two sources of income: taxes and borrowing. And also in normal times, borrowing is a commitment to future taxation. In a crisis, on the other hand, there are two more: capital levies (basically taking private wealth for public use), and seignorage, which is printing paper money and using it to buy goods and services. For very good reasons, democratic governments that do not face existential threats to their continuity do not engage in the latter two strategies. They both victimize the citizenry to save the regime. Thus, they only become feasible in an environment of threatened or actual political collapse, where the citizenry faces a greater potential victimization from without, or in any case, from some source other than the existing government.

The defeated Central powers engaged in all four strategies for obtaining revenue, and it wasn't sufficient to satisfy their reparations obligations. A word about Sargent's other three supposed hyperinflations, apart from Germany: Austria and Hungary did lose a huge portion of their tax bases in the war, but they never really had a definite reparations schedule, and their hyperinflations ended when the vague commitment to reparations imposed on them was removed by the League of Nations. Poland of course did not exist until the Treaty of Versailles and in one of his most memorable passages from that book, Keynes asserts that it could never have existed but for the Treaty of Versailles, since its finances were premised on German reparations. He writes "Unless her neighbors are prosperous and orderly, Poland is an economic impossibility with no industry but Jew-baiting."

The German hyperinflation was a consequence of the postwar settlement, and most importantly, the huge reparations it faced with a much-depleted tax base.Throughout the period 1919-1924, the Weimar Republic was continuously de-stabilized by turbulent internal politics. But the hyperinflation was a creature of Allied insistence on reparations. It peaked disastrously in 1923 after France invaded the Ruhr industrial region to foreclose on reparations, and it ended with the publication of the Dawes Plan providing the Weimar Republic an international loan and a delayed reparations payment schedule.

Sargent tells a very different story. Each of his four hyperinflations ends when a central bank is reorganized to be politically independent of a national treasury. Thereafter, although the circulation of national currency continued to increase, inflation was kept under control because newly-independent central banks adhered to reserve requirements and only purchased securities on the open market, rather than accepting worthless government bonds from the Treasury. At the same time, national treasuries were disciplined by their lack of access to the money-printing presses to enforce fiscal austerity. That combination of institutional changes constitutes what Sargent calls a "regime shift." The whole point of his article is that specific economic policies (e.g., printing money) can have different effects (hyperinflation vs. not) under different regimes because private agents have different expectations about future government behavior. In the former case, government bonds forced on the central bank will only be repaid with more money printing, while in the latter case, governments will raise revenue or reduce spending in order to finance their debt over the long run.

The notion of a regime change might be helpful, but Sargent seriously mis-deploys it here (which, after all, is the paper that invented the concept to my knowledge). There was a regime shift: the Allies decided that if the Weimar Republic were allowed to fail, the result would be catastrophic. More specifically, the US and Britain (whose policy changed by the replacement of Lloyd George in 1922 and the decline of the Liberal Party) prevailed over France and abandoned the most onerous aspects of the postwar settlement.

Before turning to the later intellectual history following Sargent, let's consider his "control" case of Czechoslovakia. Czechoslovakia was also created at the Treaty of Versailles Saint Germain-en-Laye, but unlike Poland, it was well-endowed with the most productive territory of the former Austria-Hungary. As a perceived victim of the war, it never had to pay reparations, and unlike Poland its financial health was neither dependent on those reparations being paid by anyone else, nor did it have to fight wars with its neighbors to survive (which Poland did with the Soviet Union from 1919-1921). On the contrary, since Czechoslovakia was essentially created to be a French ally, the only conceivable threat would be from Germany, and Czechoslovakia therefore was not threatened precisely because Germany was hobbled. Talk about omitted variables bias!

 From Sargent's perspective, the "sound" monetary and fiscal policies undertaken by the new state of Czechoslovakia is the fifth observation confirming his hypothesis, but in reality there was never an existential threat to its politics or economy once it had been carved out of Austria-Hungary and set up as an anti-German bulwark.

So what's the harm in Sargent's paper, besides its bad history? As I see it, two things:

1. The conflation of inflation and hyperinflation. Sargent seeks economic lessons in the cost to output and employment from ending the sort of stagflation that characterized the 1970s in developed economies. For that, he looks to this episode(s) of hyperinflation and sees that ending them was not costly. To the contrary.

In fact, hyperinflation happens because of state collapse or near-collapse. That's fundamentally different from why there was much lower but persistent inflation in the 1970s. The postwar hyperinflation ended because the Allies decided that Weimar and other successors to the Central powers should be allowed to exist.

2. More importantly, "The Ends of Four Big Inflations" propagates the myth that monetary policy is easy to get right, once it's in the hands of a superman central banker whose force of discipline cows disorderly workers into accepting at least a restraint in wage increases, if not wage cuts. That economic view was popular in the 1980s, even though, as Paul Krugman notes here, it basically failed: the Volcker disinflation was very costly. Much more harmfully, thanks to the Maastricht Treaty, it is enshrined in the mandate of the European Central Bank, to devastating effect during the past several years. That crisis has only mitigated when the officials with discretion over Eurozone macroeconomic policy were able to wriggle free of the constraints the treaty put on the ECB's facility to buy distressed government debt.

When I TAed undergraduate macroeconomics at the University of Chicago, there was a problem set with the following scenario, presented as a factual historical statement: in order to win the 1980 general election, the government of Brazil decided to fool voters into thinking their wages had gone up by printing money at a greater rate. What is the time path of nominal and real wages? Then, in 1985, a new government decided to tame inflation by appointing a central banker from the University of Chicago. What happens to inflation then?

In 1980 (and until the late 80s-early 90s), Brazil was governed by a US-backed military junta. Needless to say, there was no general election in 1980, and the story about its monetary policy dynamics is false from beginning to end. If it suffered from high inflation, that wasn't because its democratically-elected regime was too spineless to give the voters a dose of UChicago patent medicine.

Nonetheless, this idea of the craven politicians and the savior economists/central bankers has been a persistent but useful myth. The actual treatment can start with a dose of historical reality.


  1. as i understand your argument, 1920s hyperinflation was political, due to WWI treaties, and the inflation was stopped when the treaties were amended, while Sargent mistakenly attributes the end of hyperinflation to central bankers

    my taste in blogs, you should have put the lede a little higher in the story
    I guess there is nothing wrong with a few hundred words of history on 1920s europe, but the reader shouldn't have to dig to find your message; we really don't need all that stuff about Keynes

  2. The Sargents of the world are unstoppable if people of good will insist on tying one hand behind their back.

    Why endorse bullshit like "borrowing is a commitment to future taxation."?

    Really? Or an expectation of a growing economy, also known as what has always happened. The "future taxes" idea only makes sense if your baseline is zero taxes. Moral philosophy gets smuggled into econ in this manner. And it's the philosophy of a sociopath.

  3. I have collected 50 different explanations of how hyperinflation works. Most hyperinflations have nothing to do with reparations. There is always high government debt and high deficit but governments have lots of ways to get these. Hyperinflation is a bunch of positive feedback loops that can be explained in many different ways.


  4. My impression (from Schacht's memoir, among other places) was that the German hyperinflation was a direct, and more or less deliberate, response to the French occupation of the Ruhr.

  5. I agree with the argument here that Sargent's 4 inflations should not be treated as independent observations, given their common origin in the post-1918 settlements. But (as I remember his paper) a key point he makes is that a prerequisite for the end of inflation is a mechanism by which the government can raise taxes other than from seignorage revenue. This was not a new argument - Keynes's contemporary writings on the German inflation made the same point. The French occupation of the Ruhr in 1923 both reduced German tax revenue, and increased Government spending: in that sense it contributed powerfully to the hyperinflation.

  6. Good post. Correlation is not causation. Monetarists can find weak correlations between printing money and hyperinflation but it does not mean cause and effect. If a currency hyperinflates, more money must be printed to accommodate the increased amount of currency required for trade. All hyperinflation is caused by situations where there is fear of a currency becoming unacceptable for commerce. Printing money is typically not enough to cause hyperinflation, nor is monetary tightening enough to stop hyperinflation.

    Similar arguments have been made about hyperinflation of Confederate Currency during the 1860s. Confederate Currency became worthless in April of 1865 when the government fled Richmond. The Federal Government did not accept the currency and with no government to back the currency, it hyper inflated.

    Revisionist historians make a weak case that changes in the money supply were responsible for stabilizing the currency in 1864. The correlation is not very close. Inflation increased rapidly after Gettysburg and Vicksburg. The losses set in motion a panic that led to rapid inflation. A currency swap and monetary tightening were followed by lower inflation. This currency tightening is given by some as a reason for a pause in inflation. The reality is that the war stalemated after the Federal victories at Vicksburg and Gettysburg and the panic ended. For much of 1864, inflation remained low as Grant and Lee were in a stalemate, Inflation returned after Sherman's victory at Atlanta ensured Lincoln's reelection. As Sherman marched across Georgia and through the Carolinas, the currency was unacceptable in Savannah and elsewhere in his wake. The hyperinflation was due to unacceptability of the currency, not printing money.

    German currency during the period of hyperinflation was unacceptable for payment of reparations and other transactions.
    -jonny bakho