تورم، بیکاری، و مدت زمان انسجام سیاست پولی آمریکا

This paper verifies the performance of the Barro and Gordon (1983) model to explain the US inflation since the early 1950s. We divide the period from 1951:2 to 2010:2 according to each chairman of the Federal Reserve (FED). In addition, we consider aggregated periods, represented by pre-Volcker, Volcker–Greenspan, Greenspan–Bernanke, and whole sample. A genetic algorithm of stochastic search is applied to reduce the sensitivity of the maximum likelihood estimator to the initial parameter values. Surprisingly, our results show that the time consistency problem explains the US inflation during the Greenspan chairmanship at the FED.

Inflation and unemployment are, usually, the major concerns of policy makers worldwide. Policy regulations and new designs for Central Banks have been developed to keep inflation at a desirable target level. The ultimate goal of such policies is to keep society from experiencing the costs associated with a raising inflation rate. Regardless of successful efforts to control inflation rates, unemployment is still a problem for both developed and developing economies. A simple check worldwide reveals unemployment rates close to 20% of the labor force in some countries. Concerns regarding the ability of government to use the trade-off between inflation and unemployment led Kydland and Prescott (1977) and Barro and Gordon (1983) to question the time consistency problem of the monetary policy. The government recognizes the short run trade-off between inflation and unemployment and is tempted to make use of this trade-off, leading society to expect higher inflation. If the government does not commit to a low inflation rate, society will correctly believe that it will explore the short run trade-off. In practice, however, this would result in a higher equilibrium inflation rate without reducing the unemployment rate. Can the United States inflation rate be explained by the monetary policy time consistency problem? According to Romer (2001, pp. 491–492), time inconsistency is no longer a major source of inflation to OECD countries. In agreement with this view, Ruge-Murcia (2003) suggested that a game theoretical approach – in which policy makers are not overambitious as in the Barro–Gordon model, but have asymmetric preferences – better explains US inflation than the standard Barro–Gordon (1983) model. Ireland (1999), however, argued that “the time-consistency problem may underlie inflation's initial rise and subsequent fall over the past four decades”. This argument is corroborated by Christiano and Fitzgerald (2003) who stated that “The Barro–Gordon model is surprisingly effective at explaining key features of the inflation–unemployment relationship during the 20th century”. Recently, the time consistency problem of the monetary policy has called attention of some researchers. For instance, Berlemann (2005) used polled data from six developed countries and found significant inflationary biases in line with the time consistency problem for Denmark, Great Britain and the United States. In relation to Austria, Australia, and German, no inflationary bias was detected. Overall, his empirical results support the view of time inconsistency of monetary policy as a positive theory of inflation. An alternative explanation for the great inflation of the 1970s is offered by Surico (2008), who developed a method to measure the time inconsistency of monetary policy when preferences of the central bank are asymmetric. Inflation mean is decomposed into a target and a bias component. The results for the pre-1979 period show that the implicit target was 3.81% and the bias about 1%. For the post-1979 period, the inflation target has declined to near 2% while the average inflation bias has disappeared. That is, the policy preference on output stabilization was large and asymmetric before but not after the appointment of Paul Volcker as FED chairman. Bae (2010) investigated whether the long-run relationship between inflation and unemployment has changed since Paul Volcker's appointment to the FED. Differently from Ireland (1999), no assumption about the order of integration of the natural rate is made and the possibility of a breakdown in the long-run relationship between inflation and unemployment in the late 1970s or late 1980s is taken into account. Bae concludes that long-run correlations are significantly positive for the 1960s and 1970s, indicating that time inconsistency was at work during the pre-Volcker era, but are not statistically significant for the 1980s and 1990s, suggesting that time inconsistency is irrelevant to the disinflation of the post-Volcker period. This paper tests whether the Barro and Gordon (1983) model is able to explain the behavior of the US inflation since the early 1950s. We modify Ireland's (1999) approach and contribute with the literature in three different ways. First, we divide the period from 1951:2 to 2010:2 according to each chairman of the Federal Reserve. In addition, we consider aggregated periods, represented by pre-Volcker, Volcker–Greenspan, Greenspan–Bernanke, and whole sample. These time divisions allow us to assess whether the conduct of the US monetary policy has concurrently changed with the chairman of the FED. The intention is to capture any change of preference across the FED chairmen.2 Second, we apply the recently developed unit root tests, which do not suffer from low power and size distortions as the traditional tests applied by Ireland (1999). Third, we improve on the computational technique by introducing the genetic algorithm of stochastic search, which is less sensitive to initial values assumed by the parameters when determining the maximum likelihood estimates of the model's state space representation. Our major findings indicate that there is evidence of co-integration between inflation and unemployment during the terms of Burns and Miller, Volcker–Greenspan, Greenspan, and in the whole period. This evidence supports the conclusions by Ireland (1999), who argues that the time consistency problem is able to explain the dynamics of US inflation in the last decades. Our results, however, pointed out to the inability of the Barro and Gordon (1983) model to explain short run movements in the US inflation in all periods but the Greenspan era. Thus, time consistency problem can explain both short and long run dynamics of the US inflation during the Greenspan term. Given that Ireland (1999) did not divide his sample by FED chairman, he missed this important result. In addition, the Surico (2008) and Bae (2010) results, which claim that time inconsistency was not a problem after Volcker's term, were driven by the Volcker's chairmanship but do not hold when the Greenspan's period is analyzed separately. Our results are also in line with the argument that the great moderation experienced by the US economy in the last decades under the format of lower volatilities of both inflation and output growth was more of good luck than good policy. The time consistency problem explained long run inflation dynamics in the chairmanships of Burns–Miller, Volcker–Greenspan, and Greenspan at the FED and in the whole sample. In the short run, it also accounted for inflation dynamics during the Greenspan era. As recognized by Cogley and Sargent (2001), there is evidence of returning to an exploitable trade-off between inflation and unemployment in the US monetary policy conduction, a concern also raised by Taylor (1998). On the opposite side, claiming that the great moderation was more of good policy which is not captured by conventional econometric techniques, one can highlight Carlstrom et al. (2009), Benati and Surico (2009), Surico (2008), among others. The paper is organized as follows. Section 2 briefly describes the estimated equations. Section 3 presents the data, tests the long and short run restrictions, and discusses the results. Section 4 reports the concluding remarks.

This paper addressed the ability of the time consistency problem, as defined by Barro and Gordon (1983), to explain the behavior of the US inflation during the period from 1952:2 to 2010:2. We modified Ireland's (1999) approach in three different ways. First, we considered sub-samples corresponding to each chairman of the FED, and aggregate periods for pre-Volcker, Volcker–Greenspan, Greenspan–Bernanke, and whole sample. Second, we applied recently modified time series unit root tests, which have better size and power properties than the traditional ones used by Ireland (1999). Third, we introduced a genetic algorithm of stochastic search to find the maximum likelihood estimates of the model's state space representation. In the long run, the results show evidence of co-integration between inflation and unemployment during the terms of Burns and Miller, Volcker–Greenspan, Greenspan, and in the whole sample. This finding can be taken as primary evidence that the time consistency problem explains the dynamic of the US inflation during those periods. In fact, this result suggests that there is an important long run co-movement between inflation and unemployment in the US economy, as also suggested by King and Watson (1994) using a different approach. The short run restrictions shed new light on the time consistency problem. The results showed that the Barro and Gordon's (1983) short run restrictions can be rejected in all but one sub-period. Surprisingly, only for the Greenspan chairmanship at the FED the model's short run restrictions were not rejected. This means that the simple version of the Barro and Gordon (1983) model here considered was able to account for the short run US inflation dynamics during that period. Given that inflation and unemployment were found to be co-integrated, it is possible to conclude that the time consistency problem fully explained US inflation dynamics during the Greenspan era. This finding contradicts the view that the Greenspan period was characterized by a strictly anti-inflation monetary policy. In fact, movements in the unemployment rate accounted for the inflation dynamics during that period. Thus, the decline in the volatility of the US inflation observed in the last decades, which was attributed by Stock and Watson (2002), Ahmed et al. (2004), Surico (2008), and Bae (2010) to good conduction of the monetary policy, should be attributed to exogenous causes, or to the “good luck” hypothesis, according to the testable restrictions imposed by Barro and Gordon (1983) model. This was the case at least during the Greenspan's chairmanship at the FED.