A History of the Federal Reserve, Volume 2 (32 page)

Tufte (1978) offered a political interpretation. Based on work such as Kramer (1971) and many later studies, he showed that election outcomes depend positively on unemployment, real disposable income, and similar variables and negatively on inflation. Quoting Nordhaus (1975, 185), Tufte argued that “politically determined policy choice will have lower unemployment and higher inflation than is optimal.” Barro and Gordon (1983) reached a similar conclusion in a different model by assuming that the desired unemployment rate is below the so-called natural rate.

One problem with these models is that they explain policy outcomes for a period restricted to the Great Inflation. They explain neither the period before nor the one after the Great Inflation. To explain observed changes in the inflation rate, the models require improbably large changes in the so-called natural rate of unemployment. They suggest why it can be politically costly to reduce an inflation that has started, but they do not adequately explain either why inflation ended or, once ended, why it did not return. Second, the political models explain what politicians prefer, but they avoid an explanation of why an ostensibly independent Federal Reserve cooperated.
¹¹

Economists’ explanations fall into three groups. The first cites theoretical errors; policymakers used the wrong model to choose actions or interpret data. A second is misinformation; policymakers believed that their actions would reduce or prevent inflation, but the data misled them. Third, officials neglected or dismissed money growth as important for inflation. This is a special case of the first explanation that merits separate consideration. I discuss each in turn.

Theoretical
Errors

There is little reason to doubt, and abundant evidence to support, the conclusion that in the late 1960s the Council of Economic Advisers under Gardner Ackley and the Board’s staff under Daniel Brill relied heavily on a simple Keynesian model with a non-vertical, long-run Phillips curve.
Romer and Romer (2002b) develop this reasoning.
¹²
Combining this model with a belief that in James Tobin’s familiar phrase—it takes many Harberger triangles to fill an Okun gap—we get a rationalization or defense of inflationary policies.
¹³

10. The summary is based on Meltzer (2005).

11. Many economists invoke the inflation tax as a cause of inflation. The inflation tax—the cost to money holders from a fall in the real value of their balances—transferred wealth to the government, so the government gained from inflation. It gained also as taxpayers moved to higher tax brackets. I have found no evidence that suggests that this encouraged inflation or that policymakers chose this tax. Concern about unemployment was far more important for monetary expansion and Federal Reserve actions.

Another explanation of this kind points to the misinterpretation of interest rates or neglect of the distinction between real and nominal interest rates. This was a long-standing Federal Reserve problem (Meltzer, 2003). According to Taylor (1999), Clarida, Gali, and Gertler (2000), and others, until 1981, the Federal Reserve did not increase the market interest rate enough in response to inflation to offset the negative effect of inflation on (ex post) real interest rates and on expected future interest rates. Orphanides (2003a, 2003b) shows that, at the margin, the Federal Reserve’s response to its (flawed) inflation forecasts would have been sufficient to compensate for inflation. It remains true, however, that ex post real shortterm interest rates remained negative during much of the 1970s.
¹⁴

If we accept Taylor’s interpretation and conclude that the Federal Reserve did not raise nominal interest rates enough, we are left with two questions. First, didn’t the market recognize the error and raise (the more relevant) long-term interest rates and other asset prices? Second, then as now, the Federal Open Market Committee looked at many different series. They knew that inflation continued and rose at times to new levels. How could they fail to see (or learn) that their actions were inadequate to slow or stop inflation? The data in Chart 7.1 above, or similar data for the period, were available at every meeting.

There is little doubt that the simple Keynesian model as used at the time, such as is found in Ackley (1961), with a non-vertical long-run Phil
lips curve, misled policymakers in the 1960s by overstating the role of fiscal policy, especially temporary changes, understating the role of money growth, failing to distinguish between anticipated and unanticipated inflation and between the effects of temporary and permanent tax rate changes, and neglecting the role of inflationary anticipations on interest rates, wages, and prices. However, in the 1970s the Nixon administration economists did not share many of these beliefs. They accepted that the long-run Phillips curve was vertical, and they emphasized the importance of money growth for inflation. Nevertheless, under their guidance, inflation increased before the oil price shock of 1973 and continued through their term in office. Despite their beliefs about money and inflation, they urged faster money growth in 1970–72 and at other times.

12. In Hargrove and Morley (1984) Council chairmen state their interpretations. Okun (1970) explained that he regarded Friedman’s (1968b) explanation of the vertical long-run Phillips curve as of little practical relevance.

13. The argument is flawed. Tobin compares the one-time loss from unemployment (Okun gap) to the loss from non-indexed inflation (Harberger triangle). Losses from inflation continue as long as inflation continues. Fischer (1981) shows many ways in which inflation is costly that are not captured in the Harberger or Bailey triangle. See also Feldstein (1982) for effects on capital.

14. Recent papers compare two explanations of negative real short-term rates. Collard and Dellas (2004) attributes the result to chance, principally unfavorable shocks (oil); Velde (2004) cites policy errors. These are not alternatives. Both could be and probably were relevant. One problem is that the bad luck mainly affected the price level, not the maintained inflation rate. A market that recognized temporary and permanent changes would have different responses of short- and long-term interest rates to such changes, hence different responses of economic activity. Between the end of December 1972 and December 1973, three-month Treasury bill rates rose from 5.13 to 7.50 percent; ten-year constant maturity Treasury bonds rose only from 6.40 to 6.87 percent. This is one illustration of the difference between the two definitions of inflation. The Federal Reserve did not recognize the distincti
on at the time.

At most, reliance on the simple Keynesian model is part of an explanation of the start of the inflation. There has to be more to the story because it is the Federal Reserve, not the Council of Economic Advisers, that makes monetary policy. William McChesney Martin, Jr., was chairman of the Board of Governors at the start of the inflation and until 1970. Martin did not rely on explicit economic models, Keynesian or other.
¹⁵
He said many times that he did not find economic models useful and he gave most attention to market data and market participants, not economists. This reinforced the short-term focus of Federal Reserve actions and the neglect of longer-term consequences. Arthur Burns was an empirical economist who disdained deductive models. Martin and Burns made many speeches opposing inflation and pointing out its costs.

Gordon (1977, 276) concluded that his model based on a Phillips curve failed “to explain the increased variance of inflation during 1971–76 as compared to the pre-1971 period.” The model did better at explaining the cumulative change. Gordon concluded that the short-run Phillips curve became steeper after 1971, but he offered no explanation of the change. The change in the estimated coefficients of his equations from estimates for earlier periods suggests that the underlying structure had changed. The likely reason was that the public had learned to expect inflation and
increased expected inflation in response to higher actual inflation.
¹⁶
A common finding at the time was that the tradeoff between inflation and unemployment became steeper (a more inflationary cost of reducing unemployment) as time passed.

15. Of course, anyone who makes repeated decisions and does not act haphazardly can be described as having a framework in mind. This is far different from saying that Martin had an economic model relating interest rates or free reserves to output and prices. As he often said, he thought of policy as a river that had to be controlled enough to irrigate the fields without flooding them. After reading Martin’s statements in Board and open market meetings, in White House conferences, and in the question-and-answer sessions in Congress (as opposed to statements that his staff wrote for him) I cannot find an economic model. In 1963–64 as a temporary member of the House Banking Committee staff, I interviewed Chairman Martin and asked him to explain how he thought monetary policy worked. He explained about rivers irrigating fields.

Subsequent work showed that the short-run Phillips curve did not give reliable forecasts of inflation. That should have been clear from the large errors in forecasts of inflation during the 1970s. (See below and Orphanides (2003a).) As the charts above suggest, inflation and unemployment tended to increase together in the 1970s and, subsequently, to decline together. A summary of some recent research concludes, “The short-run Phillips curve should be viewed as a limited tool for forecasting purposes” (Lansing, 2002, 3).

Misinformation

In a series of papers, Orphanides showed that the information available to policymakers from 1977 to 1992 differed, at times substantially, from the data published subsequently for output and inflation. One paper (Orphanides, 2001, Figure 2) shows that the output gap, as measured at the time, was generally larger than the output gap based on data recorded in the revised national accounts. The difference was often sufficient to mislead policymakers adjusting policy in response to the output gap and inflation. In Orphanides (2003a) he showed that the principal sources of error were two misperceptions: (1) through much of the 1970s, policymakers assumed that full employment meant an unemployment rate of about 4 percent; they were slow to recognize that the so-called natural rate of unemployment had increased; (2) productivity growth slowed in the late 1960s or early 1970s (see Chart 7.6 above), but policymakers continued to expect a return to the higher productivity growth of earlier postwar years.

Orphanides’s explanation has considerable verisimilitude, as he shows. I would add that policymakers erred in treating the output loss following the 1973 and 1979 oil shocks wholly as evidence of recession, instead of partly a one-time transfer to the oil producers that permanently reduced the level of output. This contributed to the mismeasurement of the output gap and the desire to raise output by monetary expansion. This is an example of the pervasive problem created by failing to distinguish between one-time changes and maintained rates of change. The problem remains currently in discussions of inflation targeting. At the time, West Germany,
Switzerland, and Japan did not make that error and experienced less inflation despite greater dependence on imported oil. This shows that alternatives were known.

16. Sargent (1999) developed an explanation that depends on the belief that there was a permanent (or long-run) tradeoff between inflation and unemployment. Sargent (2002, 80–85) supplemented that explanation by pointing to several additional errors.

The more general point based on Orphanides’s work is that the Federal Reserve underestimated inflation throughout the Great Inflation. The persistence of the error raises a question: Why did the FOMC members not recognize the error after a few years and adjust their procedures and policy actions?

The
Role
of
Money
Growth

A noticeable change occurred in the 1960s. By 1960–61, policy had driven the CPI inflation rate from an annual rate of 3.5 percent in 1958 to 1 percent or less in 1959–61. Under the influence of Winfield W. Riefler, Chairman Martin at times testified about keeping the average rate of monetary growth close to the average rate of output growth.

After Riefler retired at the end of 1958, this model of inflation disappeared from the Board and its staff. Malcolm Bryan of the Richmond reserve bank and D. C. Johns and Darryl Francis of the St. Louis bank brought this analysis to the Federal Open Market Committee in the 1960s and 1970s without much impact on decisions. Governor Sherman Maisel, at the Board from 1965 to 1972, is an exception. He often urged a policy of controlling money growth. He was not, however, willing to control inflation if it required more than a modest increase in the unemployment rate.

Friedman and Schwartz (1963) emphasized the importance of money growth for inflation. Their work was well known but largely ignored by most members of FOMC. Economists in the Nixon administration understood the importance of money growth for inflation but yielded to political pressures.

Chart 7.5 above suggests that, in addition to its error in measuring growth of potential real output, neglect of money growth—here growth of the monetary base—contributed to the policy error. Growth of the base in excess of output growth leads the inflation rate throughout the period. Excess growth of the base would have been a useful statistic for future inflation. The Federal Reserve Board staff gave it little weight.

Economists in the Nixon administration did not neglect money growth. They watched reported money growth closely and overemphasized the effect of short-term changes. Their larger error was that most often they wanted to increase money growth to reduce the unemployment rate, and they encouraged President Nixon to talk to Arthur Burns about money growth frequently, as shown in chapter 6.