The Classical Dichotomy

In particular, Meltzer claims that theoretical knowledge was, in principle, available, but largely forgotten, which would have allowed the Fed to perform much better during this period. This is the purpose of the opening Chapter 2, which launches into an otherwise somewhat anomalous discourse on nineteenth century UK monetary debates, banking and currency schools, etc. [though, surprisingly, with no reference to Fetter (1965)]. Henry Thornton, hero of Chapter 2, appreciated that there would be wage/price stickiness in the short run, so nominal monetary changes could have real effects, and argued for a steady expansion of the money stock. Yes, indeed, but how long is the short run? How quickly does the long run, in which wage/price flexibility returns, money is neutral, and the real economy returns to equilibrium, come about? If you believe that wage/price stickiness is relatively transient, then the classical dichotomy, whereby real outcomes are (primarily) determined by real factors and monetary variables only influence nominal magnitudes, is the obvious paradigm.

It is not an outlandish position; after all, real-business-cycle theory embodies that today. Indeed, some economists claim that recent economic problems in Japan, the closest present analogue to the 1930's in the United States, are owing to adverse productivity shocks, not to mishandling of monetary policies [Hayashi and Prescott (2002), Andolfatto (2003)]. It was an even more beguiling intellectual position prior to the Great Depression. The violent immediate postwar cycle had been accompanied by extremely sharp price adjustments, yet equilibrium had been rapidly restored. Again, nineteenth century (and 1907) crisis episodes, (many of which were provoked, or at least accompanied, by major banking failures), had been generally short, if sharp, and terminated without (countercyclical) discretionary monetary policy.

So, I am more sympathetic than Meltzer to the Fed's passive approach in the early years of the Depression, say 1929–1931. However, even if the classical dichotomy was the natural starting point (and shared, as Meltzer records, by most economists and bankers), the awful experiences of 1929–1931 should have made participants willing to change their minds and theories. Some did, notably Adolph Miller and Randolph Burgess, for whom this book increased my respect. Most did not, including Harrison, worsening with age.

In my view the real, but largely unacknowledged, hero here is Keynes. That should not, perhaps, be surprising. The centerpiece of any book on the first half of the Fed's history must be the calamity of the 1930's. Friedman has argued that Keynes's (1936) General Theory was not “general” at all, but a specific response to particular U.S. conditions of the 1930's. Insofar as the Great Depression was the launching pad for Keynesian theory, it would, almost by definition, suit it best then.

Or the Real Bills Fallacy?

Amongst the many virtues of Thornton's (1802/1962) analysis was that he saw through the “real bills” doctrine, as Meltzer notes (pp. 29, 729). In contrast, the Federal Reserve System was virtually founded on a “real bills” basis, and developed and extended that position. Meltzer puts considerable weight on what he describes as the Riefler–Burgess doctrine (pp. 161–165), based on their books [Riefler (1930) and Burgess (1936)]. This goes roughly as follows: For several reasons, commercial banks were reluctant to borrow from the discount window. Open-market operations would change bank indebtedness/free reserves. To return to their desired position, banks would adjust portfolios, and market interest rates would change. It would, therefore, be possible to exert pressure on banks by open-market policy, independently of changes in official interest rates, and the thrust of policy could be assessed by examining banks' indebtedness/free reserves, as well as by the level of interest rates. No attention was paid to the monetary aggregates.

Both the “real bills” doctrine and measuring the impact of monetary policy by bank borrowing/free reserves are indeed mistaken, as Meltzer emphasizes. However, I wonder whether this was a major cause of the Fed's mistaken policies. As long as one believes in the classical dichotomy, that real variables determine real output (over the relevant horizon) and that monetary expansion will just end in more inflation, does it matter much if one also assesses likely real growth in terms of the value of “real” commercial paper presented? Perhaps it does, but I am not sure. Moreover, those questioning the “real bills” theory (e.g., Benjamin Strong) were primarily pragmatic problem solvers. If it were not for the economic theories prevalent (of which I blame the classical dichotomy more than “real bills”), the pragmatists might have had more chance to counter deflation, instead of facing so many others worrying about hypothetical problems of an inflationary impact of bond purchases; N.B. Bank of Japan has recently argued against “unconventional” measures to expand monetary growth on broadly similar worries about their potential inflationary impact, without reliance on the “real bills” doctrine.

One reason why the “real bills” doctrine was so seductive was that the United States had been, till 1914, a largely agricultural economy, and the Fed was established partly to dampen resulting seasonality in interest rates and financial fragility. At a seasonal frequency, the real bills hypothesis, and the Fed's policy based on that, seemed to work fine. Each year you could see seasonal fluctuations in bills related to agriculture (and other seasonal activities); the Fed adjusted its operations (discounts, etc.) in line, and pre-1914 seasonal fluctuations disappeared, as if by magic; see Miron (1986). Why should not a policy, that patently worked at the seasonal frequency, also work at the cyclical frequency? It was a nicely unified theory. Perhaps because eliminating seasonal fluctuations is now taken for granted, Meltzer pays it little attention, even though I believe it played a role in bolstering the “real bills” theory.

A Price Stability Objective?

Meltzer gives much weight to congressional discussions in 1926–1928, to include price stability as a formal objective for the Fed (pp. 181–192). Meltzer claims that, “If the mandate for price stability had been passed and followed, the Federal Reserve could not have permitted deflation during the Great Depression of 1929–33 or inflation during the Great Inflation of 1965–80” (p. 192), a point Alan Greenspan takes up in his Foreword. The key words, however, are “and followed.” Including these makes the sentence tautological. Without them, one can query whether a system imbued with pervasive, erroneous theories would have done better even with price stability as a formal objective. They thought that accommodating the volume of “real bills” was actually the best way to maintain medium and longer-term price stability.

Real and Nominal Interest Rates

Meltzer's second main criticism of the Fed is that it failed to distinguish between real and nominal interest rates. He is certainly correct that no such distinction was made, but would it have made much difference if they had? What is important to current borrowers and lenders is the prospective future rate of inflation, and real interest rates, over their planning horizon. The history of prices, prior to 1939, was that these normally remained stable during peacetime. If you forecast by extrapolating this, then nominal and real rates are identical. Perhaps there were occasions, for example, in the 1919 boom and severely depressed 1930's,when expecting nonzero future inflation might have seemed rational, but otherwise an approximate equality of real and nominal interest rates seems plausible. Indeed, Meltzer's estimates of predicted inflation (Charts 4.6 and 6.16) strike me as being consistent with that.

There remains debate between those suggesting that price declines in the Great Depression were largely unanticipated, and those arguing for partial prior expectation. In their (survey) paper on deflation, Borio and Filardo (2004, p. 5/6) write that

However, when he assesses real rates of interest for analytical purposes, Meltzer deflates by the ex post annual change in the GNP deflator (p. 118, n.101). Whereas price stability was the normal expectation, actual prices could move sharply, influenced by various shocks. Amongst Meltzer's many theoretical contributions is his distinction between transitory and permanent. Is it proper to measure real interest rates by deflating current nominal rates by an ex post, and often transitory, annual change in the deflator?

It matters, especially when ex post inflation was so variable each year (e.g., during 1918–1923). There was probably much unanticipated inflation in 1919 and unanticipated deflation in 1921. Using the ex post deflator will then bias real interest rates downward in 1920, and upward in 1921–1922. The question whether to use forward-looking or backward-looking inflation to deflate monetary aggregates onto a “real” basis seems less clear. Nevertheless, the sharp price fluctuations in the postwar cycle will have lowered ex-post real money growth in 1920, and raised it in 1921. Meltzer uses this case to argue that real base money growth has much greater effect on the real economy than real interest rates. I wonder whether using a short-term ex-post deflator, rather than a forward-looking deflator, might influence that argument.

Growth of Real Monetary Base

That said, there is much research showing that in many cases adding a variable for growth of (real) money significantly improves, or even dominates, a reduced-form equation including just interest rates, but not always. Chart 8.1 (p. 743) shows that periods of rapid growth in real monetary base were clearly 1931–1936 and 1939–1946. Of course, there is an explanation, and this point, always made by nonmonetarists [e.g., Kaldor (1982)] is somewhat unfair. Demand for both currency and bank reserves surged in 1930's because of bank failures, and for currency during 1939–1946 because of war. No doubt, but with central banks, such as the Fed, setting interest rates, the monetary base is generally demand determined. Its (real) movements may often be a useful leading indicator of real expenditures (for various reasons). It does not, however, follow that, if the regime shifts and the central bank manages the growth rate of base money directly, previous “normal” relationships would continue. Note the Japanese example, where base growth has been remarkably high for several years now, with little apparent benefit [Kimura et al. (2002)].

Which Monetary Aggregate?

There were three recessions in this period: 1920–1921, the Great Depression, and 1937–1938. For the Great Depression, nearly everyone now agrees that monetary contraction and bank failures played a major contributory role in extending a downturn that may (or may not) have had nonmonetary causes. Meltzer does not attribute much blame to the stock market (pp. 252–257), and believes real growth in 1928–1929 was punctured by tight monetary policy, as measured by the growth of real base. Nevertheless, there remain different views about the most crucial features of the subsequent monetary contraction: For example, was it credit growth, broad money, narrow money, bank failures, rising risk premia, or what?

As noted, Meltzer cannot use his preferred metric—real monetary base—during this period. So, he switches to real M1. In contrast, Friedman and Schwartz used M2 as their main metric, though this choice was imposed partly by data limitations in earlier years. I did not find any detailed discussion here of the pros and cons of alternative monetary measures, M0, M1, M2. Meltzer just takes M0 as best, except in the Great Depression and WWII.

The Cycle of 1919–1922

The immediate postwar cycle, 1919–1922, was severe, involving sharp fluctuations in inflation, and so distinguishing between forward- and backward-looking estimates of inflation matters most here. Official interest rates were raised in January 1920. On a backward-looking basis, real rates still remained negative but, depending on estimation methods, rose sharply forward-looking. Even so, I doubt whether monetary forces, either interest rates or base, were responsible for the downturn. A more credible candidate is the postwar cutback in fiscal deficit, though it is difficult to assess this from Meltzer's account, since no time series for the fiscal position, nor indeed for any national income components, are given. Meltzer's case for emphasizing the growth of real monetary base in ending that contraction is stronger. Real interest rates remained high, however measured. This was, perhaps, an example wherein downward price flexibility did help restore equilibrium, but several questions remain. Did such flexibility work via increasing real balances, as Meltzer argues, or through other routes, for example, improving the trade balance? If such flexibility worked in 1921–1922, why not also in 1931–1932?

The 1937–1938 Contraction

A key question about the final sharp contraction (1937–1938) is the effect of the Board's increases in reserve requirements (in July 1936 and March–May 1937). There are two possible positions. First, as long as the Fed makes additional reserves available by buying assets (e.g., government securities), from commercial banks at unchanged interest rates, then “[t]he true estimate of the effect is approximately zero,” (pp. 518, 658). If so, perhaps the extent of pressure could be measured by the rise in market interest rates, as commercial banks unloaded securities, and the Fed refrained from buying them. Meltzer notes (p. 517) that in April 1937, “Bond yields reached a local peak of 2.80 percent at the end of the week. In all, rates increased 0.34 (14 percent) from the January low.” This seems hardly sufficient to get top billing as proximate cause of the 1937–1938 contraction, though undoubtedly a step in the wrong direction.

The second position is that the liquidity of bank portfolios is what they desire, since otherwise they would be changed. There is no such thing as “excess” reserves. Liquidity includes government securities as well as excess reserves. Required reserves cannot be liquid since they are “required” and hence unusable. A sizeable call for additional required reserves thereby reduces commercial bank liquidity and represents a tightening, even if the Fed facilitates the exercise by buying government securities at unchanged interest rates. This might be particularly so if bankers' confidence is low, the financial system is fragile, and future policy of the central bank is uncertain, all true in 1936–1937.

So, what does one conclude? Meltzer notes that the question “remains controversial.” I interpret him as allocating less blame for the contraction to the reserve requirement hikes than Friedman and Schwartz (or even me).

SUMMARY

This excellent history throws much new light on the development and travails of the Fed in its early years. It is a major contribution. It has also been carefully produced and proofread, with barely a typo. It is certainly a book of which the author can be proud.

Nonetheless, Meltzer's own main focus is on testing theories against historical data. Here I have, perhaps excessively, emphasized those aspects of his analysis where I remain to be fully persuaded.

The lessons I take from the sorry story of the Fed's errors in the 1930's is that the worst judgments and decisions came from those who stuck to their prior theories, even when developments were not supportive. Better decisions came from those who were naturally pragmatic and atheoretical (such as Strong and Morgenthau), or who had the mental flexibility to adjust from their priors (Adolph Miller, Randolph Burgess, and above all, Keynes). The theories then espoused were, of course, erroneous, notably the (extreme form of the) classical dichotomy and the “real bills” doctrine. No doubt, we have somewhat better theories now, but there are few theories in social sciences that are not capable of improvement. Perhaps the moral is that we should always stand ready to adjust our theories in face of new data, and not try so hard to interpret facts to fit in with our preconceptions.