RDP 2004-04: Inflation Convergence Across Countries 5. Inflation Convergence in US Metropolitan Regions

In order to examine the role played by monetary policy in inflation convergence, it is necessary to somehow control for the impact of policy on the inflation process. One way to do this is to examine inflation convergence in a group of countries that share the same monetary policy and compare the results with countries that have sovereign monetary policy. Evidence of consistently strong inflation convergence across countries sharing the same monetary policy would suggest that similarity of policy has a role in bringing about inflation convergence. An obvious example of countries sharing the same monetary policy would be the 12 countries participating in the third stage of the EMU. However, the third stage of the EMU only commenced in 1999, which provides less than five years of data to examine. An alternative approach used in this paper is to examine the performance of US metropolitan regions for which CPI data are available.[21]

While US regions are subject to the same monetary policy, both in terms of interest rate and exchange rate policy, a complication arises from the fact that these regions share many other common factors. For example, they are subject to the same federal fiscal policy. Also, regions of the US (or any country for that matter) are likely to be much more economically integrated than the sample of OECD countries considered here, which suggests that regions are subject to more common shocks over time than even the narrow group of countries. This closer degree of integration combined with fixed exchange rates is likely to lead to stronger convergence than what was observed for OECD countries, which have had differential exchange rate regimes over the last four decades.[22] No effort is made to control for these other factors here, which limits the strength of the results. On the other hand, interest rate policy arguably has a larger impact on inflation than these other factors.

To establish whether or not inflation convergence occurs between US metropolitan regions, Equation (2) is estimated using a dataset of regional consumer prices for the US.[23] Inflation convergence is examined for the same time periods as in the earlier analysis for OECD and the global dataset. The results of Equation (2) for the US regions are presented in Table 4 and Figure 5.

Table 4: Regression Results – US Metropolitan Regions
Equation (2)
  1993–2002 1983–1992 1973–1982
α0 2.16***
β1 −0.89***
Inline Equation 0.69 0.89 0.78

Notes: Standard errors in parentheses. *** indicates significance at the 1 per cent level.

Figure 5: US Metropolitan Regions – Average Inflation and Change in Average Inflation
Figure 5: US Metropolitan Regions – Average Inflation and Change in Average Inflation

Notes: (a) Average inflation 1993–2002 less average inflation 1985–1992.
(b) Average inflation 1983–1992 less average inflation 1973–1982.
(c) Average inflation 1973–1982 less average inflation 1961–1972.
(d) Top panel – 1985–1992; middle panel – 1973–1982; bottom panel – 1961–1972.

Sources: author's calculations; US Bureau of Labor Statistics

The results show that there is evidence of strong inflation convergence occurring consistently over the past four decades. The β1 coefficients are negative and highly significant in all three periods, and notably larger in absolute terms than those obtained for the OECD, especially in the two earlier periods.[24] Also, the explanatory power of the regressions is consistently high, as seen in the adjusted Inline Equation statistics. These findings point to inflation convergence being a stable property of inflation for metropolitan regions in the United States over the past four decades.[25]

The evidence from US metropolitan regions suggests that in an environment with homogeneous policies, convergence in inflation rates occurs consistently through time. This compares with the experience of OECD countries, where the evidence points to inflation convergence not being a stable property. Interestingly, convergence in the inflation rates of OECD countries appears to occur during periods when countries share similar monetary policy goals, or at times when policy goals have changed to be more similar across countries. During the 40-year period considered, these events occurred under the Bretton Woods regime and the more recent period of inflation targeters and countries which otherwise operate monetary policy with price stability as a primary concern.

Together, the evidence points to monetary policy playing a role in bringing about inflation convergence rather than convergence being a mechanical property of inflation. It has to be acknowledged, however, that use of US metropolitan regions as a control for monetary policy is not perfect, as a number factors are likely to cause inflation to be more correlated than across OECD countries. This is possibly reflected in the strength of the inflation convergence result within US metropolitan regions. Nonetheless, the evidence appears consistent with a link between common policies and convergence in inflation.


The author would like to thank Ellis Connolly for suggesting this approach. [21]

With fixed exchange rates, in the long-run productivity differentials in the non-traded sector are likely to be responsible for most of the cross-country divergences in inflation rates (sometimes referred to as the Balassa-Samuelson effect). In the US, however, this effect is likely to be small with virtually no barriers to trade or mobility of labour across regions. In the case of floating exchange regimes, continuous appreciations or depreciations can lead to persistent inflation differentials. [22]

The series used is labelled ‘consumer prices for all urban consumers’ (CPI-U) for metropolitan areas. These data are sourced from the Bureau of Labor Statistics, available at <http://www.bls.gov/cpi/>. Here, regions that do not have data available for the years 1961–2002 are excluded, yielding a sample of 18 regions (out of 27 possible regions). [23]

In fact, there appears to be a steady decline in the absolute size of these coefficients over the past 40 years. A potential explanation for this pattern could be related to the fact that recessions in the US have become less frequent and less severe (Bernanke 2004). In this case, the larger negative coefficients in the earlier periods could be a function of greater cross-sectional volatility in inflation outcomes. [24]

Table A1 in Appendix A presents the β1 coefficients from regressions of Equation (2) on the dataset of US regional consumer prices under the alternative of five-year time periods. The results in Table A1 tell a similar story to those presented here. [25]