RDP 2012-01: Co-movement in Inflation 1. Introduction

Just as there is an extensive literature on the co-movement of real business cycles,^[1] there is growing acknowledgement of substantial co-movement of inflation across countries. For example, Ciccarelli and Mojon (2010) found that, on average, up to 70 per cent of the variation in countries' consumer price index (CPI) inflation can be explained by a common factor. Neely and Rapach (2008) found that common and regional factors together can explain around 50 per cent of the variation in national inflation rates. Monacelli and Sala (2009) find a smaller number using disaggregated data but still, on average, around 15–30 per cent of the variation in consumer prices can be explained by a common factor.

This co-movement in inflation rates can be seen in Figure 1, which plots year-ended CPI inflation for a number of economies.

Inflation rates across countries may move together for a number of reasons. Over longer periods, common changes to policy frameworks and policymakers' views about the appropriate rate of inflation can drive changes in the level of inflation, resulting in observed co-movement. In the short to medium term, if exchange rates do not adjust to offset shocks to the international prices of imported goods, these shocks can flow through to domestic consumer prices. Also, fluctuations in global output and international trade can influence domestic demand and therefore domestic inflation. Common shocks to which a number of countries are exposed (for example, an oil price shock, demand and supply shocks) could also lead to correlated movements in inflation rates.

Just how important the various channels will be, however, is likely to be influenced by structural features of different economies, such as the degree of trade integration, the flexibility of the exchange rate, and also the extent of exchange rate pass-through by domestic firms. Policy responses across countries will also be important. While an inflation-targeting central bank with an independent inflation target should determine the level of domestic inflation in the long-run, inflation rates across countries may still be observed to move together in the short to medium term due to lags in the transmission of monetary policy and the fact that certain shocks, for example, temporary shocks, may warrant only a partial or even no monetary policy response.

This paper investigates the topic of co-movement in inflation rates using a panel vector autoregression (panel VAR) model (Canova et al (2007) and Canova and Ciccarelli (2009)) for the G7 economies. The panel VAR framework is very flexible and has features in common with more familiar factor modelling techniques. We investigate the significance of international inflation in explaining domestic inflation in the G7 countries after controlling for potentially important explanations of the observed co-movement in the data. In particular, we include common and country-specific measures of real activity, and also oil and non-fuel commodity prices in the model.^[2] The panel VAR framework is also used to investigate which of these potential explanations of ‘global inflation’ are most supported by the data.

Our results suggest that common shocks to commodity prices are more important for driving global inflation dynamics than are common movements in real activity. Neither of these potentially important drivers of global inflation, however, can fully explain the observed co-movement in inflation in the G7 data. Even when controlling for these factors, the common inflation indicator, constructed as a simple average of individual country inflation rates, is found to be a significant explanator of domestic inflation, suggesting that international movements in inflation contain useful information over and above what is reflected in data on foreign real activity and, to a lesser extent, commodity prices.

Given the role for global inflation in explaining G7 countries' inflation, we then turn to examine the role that global inflation plays in explaining Australian inflation. After augmenting standard single-equation models of Australian inflation with the average of G7 countries' inflation rates, we find that coincident information on international inflation, in particular, is a statistically significant explanator of both headline and, to a lesser extent, trimmed mean inflation.

The rest of the paper is structured as follows: Section 2 reviews the relevant literature, paying particular attention to the possible determinants of global inflation; Section 3 outlines the panel VAR model and its key features (further details of the model and the estimation procedure are in Appendix A); Section 4 presents the main results and discusses the role of correlated movements in real activity and commodity prices in driving movements in global inflation; and Section 5 discusses the significance of international inflation for modelling inflation in Australia.

Footnotes

See Backus, Kehoe and Kydland (1992) for the seminal contribution. Kose, Otrok and Whiteman (2003) and Canova, Ciccarelli and Ortega (2007), amongst others, have shown that a global or common component can explain a substantial share of the variation in real variables across countries. Potential drivers of this co-movement pointed to in the literature include international trade, financial conditions and the stance of monetary policy (Canova et al 2007), along with movements in productivity (Crucini, Kose and Otrok 2011) and consumption demand (Wen 2007). [1]

Most other studies on global inflation have only worked with inflation data. An exception to this is Mumtaz, Simonelli and Surico (2011), who employ a dynamic factor model over a very long sample and incorporate data on both output and inflation in their estimation. They do not, however, control for commodity price movements or other measures of real activity such as consumption and investment. [2]