The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Federal Reserve Bank of Chicago Global Inflation Matteo Ciccarelli and Benoît Mojon WP 2008-05 Global In ation Matteo Ciccarelliy European Central Bank Beno^t Mojon European Central Bank Federal Reserve Bank of Chicago First draft: March 2005; This draft: April 2008 Abstract This paper shows that in ation in industrialized countries is largely a global phenomenon. First, the in ation rates of 22 OECD countries have a common factor that alone accounts for nearly 70 percent of their variance. This large variance share that is associated with Global In ation is not only due to the trend components of in ation (up from 1960 to 1980 and down thereafter) but also to uctuations at business cycle frequencies. Second, we show that, in conformity to the prediction of New Keynesian open economy models, there is little spillover of in ationay shocks across countries. The comovement of in ation comes largely from common shocks. Global In ation is a function of real developments at short horizons and monetary developments at longer horizons. Third, there is a robust \error correction mechanism" that brings national in ation rates back to Global In ation. A simple model that accounts for this feature consistently beats the previous benchmarks used to forecast in ation 4 to 8 quarters ahead across samples and countries. Key Words: Global In ation, common factor, international business cycle, OECD countries JEL: E31, E37, F42 We would like to thank Julio Rotemberg (the editor) and an anonymous referee whose comments helped us improve the paper substantially. We are also indebted to Michel Aglietta, Filippo Altissimo, Micheal Bordo, Gabriel Fagan, Domenico Giannone, Ronald McKinnon, Je rey Frankel, Anil Kashyap, Nobu Kiyotaki, Simone Manganeli, Cyril Monnet, Philippe Moutot, Lucrezia Reichlin, Frank Smets, Mike Wickens, Micheal Woodford and seminar participants at the ECB, the OECD, the IMF, the Bank of Canada and the NBER Summer Institute 2006 for useful comments on several versions of this paper. Finally, we are very grateful to Daniel Levy for detailed and constructive comments on the draft, and to Sandrine Corvoisier for excellent research assistance. The views expressed are those of the authors and do not necessarily re ect the views of the European Central Bank, or the Federal Reserve System. Remaining errors are our own. y Corresponding author: European Central Bank, DG Research, Kaiserstrasse 29, D-60311, Frankfurt am Main. E-mail: matteo.ciccarelli@ecb.int. 1 Introduction This paper provides a formal analysis of the international comovement of in ation. We document that the international comovement of in ation has been strikingly high. We then investigate the origin of this comovement before we show one of its potential applications, i.e. that it can improve the forecasting of national in ation rates. The idea that macroeconomic developments depend on international conditions is not new. Measures of this depencence, however, were developed only recently. For instance, Kose, Otrok and Whiteman (2003) (KOW thereafter) nd that the world's common component to expenditure time series of 60 countries explains between one fourth and one half of the variance of these series in OECD countries.1 By de nition, the main risk of ignoring international developments is to overrate the importance of domestic ones. And these include domestic macroeconomic policies. As KOW put it: \ Understanding the sources of international economic uctuations is important both for developing business cycle models and making policy". Surprisingly, the studies of global macroeconomic developments had, initially, mostly focused on the real business cycle. However, the uctuations of in ation have been strikingly similar around the world. All OECD countries have experienced long-term swings in the level of in ation. In ation has progressively risen in the 1960s and 1970s before it declined in the 1980's. In ation has further declined in the early to mid-1990's and has since then remained low and stable. A more recent perspective shows that in ation rates are accelerating in most countries in 2007 and early 2008. Prominent economists have recently pointed to the common disin ation trend around the world (Rogo , 2003) or at least OECD countries (Levin and Piger, 2004, and many others). These studies may overlook two important aspects of the international comovement of in ation. First, they restrict their analyses to the post-1980 disin ation, hence disregarding the possibility that the previous phase, i.e. the acceleration of in ation between 1960 and 1980, was also very much a shared experience of most countries of the world (a point described early on by McKinnon, 1982 and Darby and Lothian, 1983, among others). Second, they focus strictly on the downward trend or on downward breaks of the in ation process, while, as we show in this paper, there is more than su cient evidence of comovements of in ation at the business cycle frequencies as well. 1 See also Forni and Reichlin (2001), Canova, Ciccarelli and Ortega (2007) and references therein. 1 We proceed in three sequential steps. We rst document the fact with a simple common factor analysis. We extract the common component to the quarterly in ation series of 22 OECD countries from 1960 to 2007,2 and quantify the extent to which this measure of \Global In ation" helps explain national in ations. Subsequently, we test alternative explanations for the international comovement of in ation. Then we study the empirical implications of our stylised fact for the dynamics and the predictability of national in ation rates and check whether it is possible to exploit Global In ation to improve in ation forecasts. Our main results can be summarized as follows. First the intuition that in ation has been a global phenomenon is decidedly con rmed by the data. We indeed show that a simple average of 22 OECD countries in ation, which we call \Global In ation," accounts for 70 percent of the variance of in ation in these countries between 1960 and 2006. The qualitative result is not only robust to di erent sample periods, but is also valid at low and at business cycle frequencies, where the variance explained by Global In ation is about 37 percent on average, and much larger in numerous countries. Second, consistently with the conclusions of several versions of New Keynesian open economy models, we reject that in ationary shocks spill over across countries have been important. Hence, the international comovement in in ation seem to come from the high correlation of in ation determinants in the OECD. At short horizons, in ation has responded to commodity prices and the international business cycle. At longer horizons, changes in the level of in ation re ect either major changes in the monetary policy regime that have been coincident across countries or at least changes in the mean level of in ation tolerated by central banks. These results are important because they support the notion that in ation can be analyzed directly at the global level, and it con rms that the 70 percent of in ation variance that is global depends on both real and monetary developments. Third, Global In ation is an attractor of national in ation, i.e. national deviations from their projection on this attractor are reverted. The evidence is again uniform and robust across sample periods and countries. We also document di erences in the long run impact of Global In ation. Countries that have experienced stronger commitment to price stability (e.g. Germany) are less a ected than those with weaker in ation discipline (e.g. Italy). However, this kind of \Error Correction Mechanism" helps predict national in ation of most OECD countries at various horizons and over several samples. As a result, our forecasting model of in ation augmented with the Global In ation consistently outperforms standard AR(p) and 2 Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, Luxembourg, New Zeland, Norway, the Netherland, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States. 2 Random Walk models of in ation, as well as augmented Phillips curve models a la Gerlach (2003). This seems to be true also in the recent period where unpredictability of in ation has been documented in the recent literature.3 We argue that existing forecasting models of in ation do not appropriately account for an international perspective which can improve predictability upon a simple Random Walk benchmark. These results could lead the Global In ation model to become a new standard for forecasting in ation in OECD countries. Several papers on the international comovement of in ation have appeared since the rst circulation of this paper. Mumtaz and Surico (2008) and Monacelli and Sala (2007) use factor models to decompose sectoral national in ation rates into world and national components. Wang and Wen (2007) try to replicate the empirical fact that comovement in in ation rates is higher than the one in output growth in a variety of calibrated New Keynesian two-country models. Cecchetti et al. (2007) investigate the reasons why most G7 countries went through the Great In ation in the seventies and provide evidence in favour of similar changes in monetary regimes. The main conclusions of these papers do usually not contradict our own. Our work relates only marginally to the literature on \Globalization and in ation" that analyzes whether the integration of the world economy has changed in ation dynamics.4 We nd that in ation has been dominated by common shocks ever since the sixties and this has not changed over time. Arguably, globalization implies similar terms of trade shocks for OECD countries, an hypothesis that is fully compatible with our model of in ation but one that we do not analyze in this paper. As a nal remark, we shall note that the economic and econometric arguments we use in this paper do not claim to cover all the reasons for why in ation could be driven by global outcomes, nor does it pretend to be exhaustive on the empirical investigation of our ndings. We are con dent, however, that our results may provide a good starting point for exploring the hypothesis that in ation should {to some extent{ be modelled as a global rather than a local phenomenon. 2 In ation as a global phenomenon In this section we document the empirical fact that in ation has largely been a global phenomenon over the last 45 years. We rst describe our data and their necessary transformations. Then, we estimate the Global In ation using simple alternative measures. Finally, we provide 3 See Atkenson and Ohanian (2004), Stock and Watson (2007) and d'Agostino, Giannone and Surico (2006). See the IMF Spring 2006 World Econmic Outlook, Chen, Imbs and Scott (2004) and the debate on the global slack which opposes Borio and Filardo (2007) to Ball (2006), Rogo (2006), Woodford (2007) and Ihrig et al. (2007). 4 3 some descriptive statistics over di erent subsamples and di erent subgroups of countries. 2.1 Data Sources and transformations of all data are described in more detail in the appendix. The data used in this section are values of the CPI indices available quarterly from the OECD main economic indicators database from 1960 onward. Our analysis mainly focuses on quarterly year-on-year (y-o-y) in ation rates, which, by construction, have no seasonal pattern. To analyze the uctuations over the business cycle frequency, we consider a transformation of the data that lter out the lowest and the highest frequencies. We do this using a band-pass ltered CPI in ation rates with a pass band that removes all frequencies but the periods of 6 to 32 quarters. 2.2 Estimating Global In ation In what follows, we brie y describe and compare results for three alternative measures of Global In ation, namely: 1. A cross-country average; 2. The aggregate OECD in ation, published by the OECD; and, 3. A measure based on static factor analysis.5 Results reported in subsequent sections are mainly based on the simplest and most intuitive measure, the cross-country average. The \average" measure is the simple average of the year on year in ation rates of the 22 countries that have been members of the OECD for most of the sample period 1961:2{2006:4.6 The aggregate OECD in ation is a weighted average of all OECD countries' in ation, where the weights are proportional to GDP. Regarding the common factor analysis, we opted for a parsimonious approximate factor representation (see e.g. Forni et al., 2000; Stock and Watson, 2002) which decomposes in ation rates for the pool of countries as t n 1 = ft n 11 1 + "t (1) n 1 where the rst term captures the e ect of a common factor (ft ), to which each country responds di erently through , whereas the last term refers to the idiosyncratic dynamics which captures 5 Results are almost identical when using a dynamic factor model as introduced, e.g., by Forni et al. (2000). The 8 OECD countries that we do not include in our sample are Mexico, Korea, Turkey, the Czech Republic, Hungary, Poland, the Slovak Republic and Iceland. 6 4 the components generated by shocks whose e ects remain local. We assume orthogonality between ft and "t , and normality of the error term, with "t N (0; R). An estimation of the factor is obtained using static principal component methods described in Stock and Watson (2002). Data have been previously demeaned and standardized to have unit variance before estimating ft . Figure 1 reports the three measures of Global In ation.7 Two observations are in order. First, the \average" and the factor model measures are almost identical, while the OECD aggregate deviates from the other 2 series, especially in the second half of the eighties, presumably because of the di erent sample of countries. Second, the uctuations and trends in the Global In ation re ect the major events of the last 45 years. All measures are characterized by two trends, up from 1960 until the late-seventies (associated with the two oil shocks and the decline in OECD productivity) and down thereafter (re ecting tight monetary policies and the debt crisis), and, ve or six cycles along the way. Given that both the seventies Great In ation and the subsequent tight monetary policy have been observed in most countries, the trend components of Global In ation perhaps should not come as a surprise. As a matter of fact, Corvoiser and Mojon (2005) show that breaks in the mean of in ation largely coincide through out the OECD: around 1970, around 1982 and, to a lesser extent, around 1992. Cecchetti et al (2007) show that the great in ation of the seventies coincide with prolonged periods of overly accomodative monetary policy, a point we discuss further in section 3. To gauge the extent to which the in ation in individual countries are related to Global In ation, Figure 2 reports the in ation series of the G7 and of the euro area with their projections on the common factor. Visual inspection reveals not only that the trend is captured accurately, but also that the most relevant cyclical movements are indeed common. 2.3 Descriptive statistics Table 1 reports the share of the variance of national in ation series that is explained Global In ation for each of the three measures introduced in the previous section: the simple crosscountry average, the OECD aggregate in ation, and the rst static common factor.8 In each case, the national idiosyncratic variance is the complement to one of the gures reported in the table. The last column also shows the share of the variance explained by the second static factor. The table also reports the variance decomposition exercise for the euro area in ation rate. 7 The OECD aggregate and the \average" have been de-meaned and standardized for the gure. This share is de ned as 2i var(ft )=var( it ). It is equivalent to the R-square of a regression of the national in ation rate on Global In ation and a constant. 8 5 First, all measures of Global in ation explain more than two thirds of national in ation rates uctuations on average. The comovement of in ation is decidedly large. By way of comparison, we nd that the global business cycle accounts on average \only" for about one third of the variance of industrial production growth in OECD countries.9 It is also clear that the second common factor of the in ation series explains only a very limited share of the variance of national in ation series, on average. We consider this fraction small enough that we can model national in ation rates with one common factor only. We also note that the OECD aggregate in ation under performs the other three measures. We conjecture that this is because this aggregate includes countries that are not in our sample. Moreover, within our sample of countries, we also found that averages that are weighted by country size under perform the factors and the simple unweighted average (not reported). Table 1 ranks (the column `average' being the reference) the countries by increasing share of the in ation variance that is explained by the common factor. Only ve countries have less than 60 percent of this variance explained by Global In ation. Four of these ve countries, Greece being the exception, are usually seen as low in ation economies. We also note that the ranking of the countries has little to do with geography nor the nature of the exchange rate regime. The fact that non-European countries are spread through out the distribution casts doubt on the argument that Global In ation among OECD countries is just a re ection that a majority of these countries are located in Europe. We actually estimated another measure of Global In ation using a sample of six countries evenly split across time zones: Canada, US, UK, the euro area, Japan and Australia. We obtain a even higher median (0.75 instead of 0.73) and mean (0.74 instead of 0.71) share of in ation variance that is explaned by Global In ation (see the top panel of Table 2). This result reinforces our conjecture that the comovement of national in ation rates does not necessarily re ect only European economic developments. Moreover, the high degree of comovement in in ation may be seen as trivial because (European) countries in our sample have participated to a monetary union since 1999 after they had pegged their currency to the Deutsche Mark in one way or another since the late seventies. For these countries, most of our sample period, from 1960 to 1973 and then from 1979 to 2007, would be closer to a xed exchange rate regime than to a one of oating exchange rate. For the other countries in the sample, the high degree of comovement could also come from the long periods of the last 45 years when exchange rate were xed, mainly up to the 9 A similar proportion has been found by KOW and used to document the importance of a common world real factor. 6 mid seventies or under some form of pegs.10 However, the degree of comovement of in ation remains strikingly high if one looks at countries that did not pursue any sort of xed exchange rate policies. This can be seen from the second and third panels of Table 2 where we consider the same sub-set of six countries as in the previous paragraph, though, this time, on the post 1974, as well as on the post 1983 sample. We obtain again a high degree of comovement of in ation among countries whose exchange rates were not formally tied together.11 Another somewhat easy explanation of the magnitude of in ation comovement is that it simply re ects common trends in the in ation series. This is why we now explore how much of the business cycle uctuations in in ation are correlated across countries. In Table 3 we report (again ranked taking the column `average' as reference) estimates of the share of de-trended in ation that is associated to a common factor. The national in ation series were detrended using Baxter and King (1999) band pass lter, which extracts cycles of length comprised between 6 and 32 quarters long with a truncation of 12 lags. These cyclical components of in ation are then used for extracting the common factor at business cycle frequencies (Figure 3). Again, the share of national in ation variance that is common is very large by any standard with mean and median on the order of 37 percent.12;13 The comovement of in ation is not only due to the trend component associated with the seventies great in ation and the coincidence of the countries's in ations gradual acceleration up to 1980 and the gradual disin ation that followed. Global In ation actually explains a large share of the in ation variance also in countries like Switzerland and Germany, that is countries where the seventies in ation have been much smaller than in the average of OECD countries. A comparison of the ranking of countries in Tables 1 and 3 indicates that, in relative terms, Global In ation seems to matter more at business cycle frequencies for low in ation countries, where the share of variance explained by Global In ation is among the lowest when we don't remove the trend (Table 1), and just below the average when we do remove it (Table 3). This should be contrasted with the experience of countries such as Sweden and Portugal where the common factor of detrended in ation has less explanatory power for local in ation developments than the non-detrended measure. These results are a rst indication that the 10 For instance, the US and Japan were de facto pegging their currencies between February 1973 and February 1978; Australia had its exchange rate to the US Dollar uctuate within a narrow horizontal band form October 1974 until November 1982, and the UK were shadowing the ECU in the late 1980's (Reinhart and Rogo , 2002). 11 One notable exception is Germany for the post 1983 sample. The divergence of German in ation from the world evolution around the reuni cation explain this low degree of co-movement. 12 These results hold for other detrending methods such as the HP tler or the rst di erence lter of in ation. 13 An alternative approach, which consists of comparing the coherence of the cross spectra of Global In ation and national in ation rates at each frequencies, provides very similar results. For most countries, this coherence is positive and typically superior to 0.5 at both low and business cycle frequencies. The results (not reported) are available upon request to the authors. 7 low frequency comovement of in ation is likely due to monetary policy, a point we come back to in section 3. Finally, we have computed the cross-correlation of Global In ation with national in ation series at several leads and lags. This exercise is useful to guring out whether in ation tends to lag or lead Global In ation in some of the countries. Results (not reported, but available upon request) show that almost no country is markedly leading or lagging global developments. This is a rst indication against the possibility that in ation in a particular country (e.g. the U.S.) has been systematically spilling over to the rest of the OECD countries and that our focus on Global In ation mistakenly picks up that one country sets the OECD in ation trend. 3 Why has in ation been a global phenomenon? Given the nding that Global In ation explains a substantial proportion of national in ations' variance, this section analyzes the likely causes of such commonality, provides quantitative estimates of the sources of the commonality, and investigates the determinants of the global factor. We make two important steps in the analysis of international in ation comovement. First, we reject that this comovement is due to spillovers of country speci c shocks, thereby bringing support to the conclusions of New Keynesian models of monetary policy in open economies. Second, we describe the nature of the common shocks that explain the comovement of in ation across countries. At a short horizon, the variance of global in ation is largely explained by commodity price shocks and the world business cycle. At a longer horizon, global in ation echoes major changes in the monetary policy regime that have taken place simultaneously across OECD countries. 3.1 Some theoretical considerations The comovement of OECD in ation rates can stem from two general sources: common shocks that spread evenly or at least simultaneously across countries; country speci c in ationary shocks that spill over from one or a subset of countries. The rst possibility is somewhat uncontroversial. Commodity price shocks, internationally correlated productivity shocks, cost push shocks or changes in the stance of monetary policy could have analogous impacts on OECD in ation rates and therefore induce comovement in in ation rates. The common shocks should induce more comovement in in ation if they have a strong permanent component, or account for a larger share of the variance of in ation 8 determinants, and if the responses of national in ation rates to these shocks are similar across countries. The possibility of synchronization stemming from spillovers is more challenging. For instance, in the context of a stylized two country New Keynesian model, such as the one derived by Clarida et al. (2002), there is no spillover of in ation across countries when the central banks implement optimal policy independently from one another { a situation that is more likely to characterize the oating exchange rate regime that prevailed since 1975. In fact, in absence of cooperation, domestic in ation and output depend only on domestic cost push shocks.14 In the case of policy cooperation, cost push shocks in one country can spill over abroad. E ectively, the domestic central bank would set its output gap as a function of both home and foreign in ation. But its policy would depend on the sign of the e ects of foreign output on domectic marginal costs. Clarida et al. (2002) stress that this sign is indeterminate.15 Hence, cross-country in ationary spillovers need not be positive. And negative spillovers would hardly explain the international comovement of in ation.16 The case against positive spillovers is reinforced by Wang and Wei (2007) and by Woodford (2007). Wang and Wei (2007) explore calibrated New Keynesian models with either sticky information or sticky prices in order to assess whether monetary shocks could generate the high degree of international in ation comovement that we see in the data. They show that, unless the monetary shocks are themselves correlated, there is little international comovement in the in ation. Woodford (2007) shows that in the context of sticky price two country models, central banks keep the control of domestic in ation. For instance, stimulative foreign monetary policy does not a ect domestic in ation directly, and its indirect e ects through the foreign output gap occurs through depressing the domestic output gap. These modeling exercises have two important limitations. First, they tend to focus on one speci c shocks (cost push for Clarida et al. and monetary for Wang and Wei). We cannot rule out though that, in the data, a particular combination of shocks could boost the comovement in in ation rates that would be due to spillover.17 One such spillover channels put forth by 14 Note, however, that this solution is compatible with a high correlation between di erent country in ations also if similar and independent cost-push shocks hit the two economies at the same time. Such coincidence would be observationally equivalent to common shocks. 15 In the more likely case of intertemporal elasticities of substitution (1/ in Clarida et al's notations) inferior to one, the elasticity of domestic marginal cost with respect to foreign output is positive and the spillover of a foreign cost push shock on domestic in ation is negative. 16 Even in the case of a monetary union, the ultimate degree of monetary cooperation, it is di cult to rule out that negative in ation spill-overs could be important. In particular, as long as the central bank loss function depends on the average in ation in the two countries, a shock to in ation in one country would lead to a higher interest rate for both countries and, potentially, a negative response of in ation in the other country. 17 The simulations reported in Wang and Wei (2007) also show that the degree of in ation comovement obtained in calibrated two-country models can be very sensitive to modeling assumptions. Some of these, such as opting 9 McKinnon (1983) result from the resistance of central banks to see the dollar either depreciate too much or appreciate too much. He argues in particular that large scale interventions on Forex market could have led to over issuance of money when the dollar depreciated in the seventies and under issuance in the early eighties. For these consideration, in section 3.2 we assess systematically whether spillovers have been important in the data. Second, these models describe in ation uctuations around constant steady state in ation rates. Hence, they may not be the best tool to analyze the trend evolution of in ation and the observed international comovements at low frequencies. Both theoretical and narrative evidence strongly point to monetary policy as the single determinant of trend in ation. Ball (2006) argues that even in small open economies, central banks retain the ability to stabilize in ation at the level of their choice. Moreover, the narrative evidence on shifts in the practise of monetary policies is compelling. We know for instance that major changes in monetary policies clustered in two periods of a few years in a majority of OECD countries. First, the early eighties saw both the US disin ation and the European Monetary System based disin ation in Europe. Ten years later, most OECD countries were either embarking on the low in ation single currency planned by the Maastrischt treaty or adopting in ation targeting at lower in ation rates than the one that prevailed in the eighties. Turning to the United States, Goodfriend (2007) argues that a common understanding that core in ation should be kept near 2 percent arose in the Federal Open Market Committee in 1995. Hence, the nineties have witnessed all countries of our sample setting up monetary policy regimes with an in ation objective graviting around 2 percent, and the central banks adopting an explicit or implicit commitment to keep it there. Ex post, notwithstanding the acceleration of world in ation in 2007 and 2008, in ation has remained remarkably close to the quanti ed in ation objective or target of the central banks.18 Hence, the common trend in in ation since the late seventies is easily related to the evolution of monetary policies. What about then in ation acceleration in the sixties and seventies? Cecchetti et al (2007) show that the Great In ation of the seventies relates to continued periods of loose monetary policies in several G7 countries. The main exception is Germany, where the low tolerance of the independent Deutsche Bundesbank for in ation in that period is a widely acknowledged fact. It is precisely the anti-in ation credibility of the German central bank that led other European countries to anchor domestic in ation expectations through a peg of their currency to the Deutsche mark from 1979 on within the European Monetary System. for \cash in advance" rather than \money in the utility function", are however di cult to relate to the evolution of OECD countries since 1960. 18 Diron and Mojon (2008) show for instance that at one and two-year horizons, in ation targets have been predictors of in ation. 10 Altogether, theory and narrative evidence suggest that trend in ation have been dominated by changes in the mean level of in ation chosen by monetary authorities. This conclusion shifts the problem to understanding why major changes in monetary policies with comparable impacts on trend in ation have coincided across OECD countries. Potential explanations include \peer pressure" between central banks,19 changes in the dominant paradigm in monetary economics,20 and common changes in preferences due to common demographic trends such as the baby boom. It however goes beyond the scope of this paper to test these alternative explanations. Fluctations of in ation around its trend also have a fair amount of commonality as we showed in section 2. These uctuations might reasonably have less to do with shifts in monetary regimes and more with responses to non permanent common shocks.21 The most likely suspects are shocks to the price of commodities as well as common shocks to output gaps that re ect the international business cycle (KOW, 2003). In section 3.3, we test these hypotheses by reporting the ability of various common in ation drivers, be they monetary or real, to forecast global in ation either in the short run or over a longer horizon. 3.2 Common shocks versus international spillovers Here we check whether spillovers have been important in the data. Following Stock and Watson (2005) we extend the speci cation in (1) with an autoregressive component and decompose the shocks to in ation of G8 countries (G7 and Australia)22 into three sources: common, crosscountry spillovers and domestic shocks. Using the same notation as before and denoting by t the vector of year-on-year in ation rates, the reduced form VAR is t = A (L) t 1 + t where the error terms have a factor structure t = ft + 19 t See for instance Besley and Case (1995) for a model of endogenous in uences of public policies across geogrpahic areas. It is also stricking that breaks in the mean of in ation in the OECD have clustered around a few years, a result that reinforce the view that central banks have implemented good and bad monetary policies together (Corvoisier and Mojon 2005). 20 One could consider (and eventually check) that the Cogley and Sargent (2005) explanation of the Great In ation is also valid in other countries. 21 The existence of comonalities in the evolution of monetary policy help explain the commonalities in the trend or long run dynamics of in ation, but could in fact reduce the importance of the global component in in ation at the business cycle frequencies if, for instance, all central banks following the same reaction function were perfectly o seting in ation movements due to global forces. We thank the referee for pointing out this important issue to us. 22 Australia helps rebalance this subset of countries across time zones. 11 As before, ft is the vector of (possibly k) common international factors and, as in Stock and Watson (2005), we assume the following E ft ft0 = diag 2 f1 ; :::; 2 fk 0 t t = diag 2 2 E 1 ; :::; n The common factors are identi ed as those factors that a ect domestic in ation in all countries contemporaneously. This is the same speci cation as in (1) with a VAR structure of the endogenous variable which now, given the assumptions, allows us to decompose the h-step ahead forecast error variance for in ation into the sum of three sources: common shocks, idiosyncratic shocks, and spillovers of domestic shocks from the other countries. In Table 4 we report the variance decomposition at 1, 4 and 16 quarters ahead for three sub-periods.23 Note rst that given the identi cation assumption, at one-quarter horizon spillovers do not account for any explained variance. At longer horizon they might account for a variance between 4 percent and 24 percent for particular countries, and between 9 percent and 15 percent on average across countries. As expected, most of the variance is explained by the international factors. Summing up column (1) and (2) the percentage of the variance explained by the common factors is roughly the one that we have found in Section 2, both at short and long horizons. Overall, then, these ndings qualify those that have been discussed before, and show that among the common sources, spillovers { though not entirely ruled out { account only for a small portion of the common in ation variability.24 3.3 The determinants of global in ation To determine whether, when and by how much Global In ation may be linked to commodity prices, real or monetary shock or a combination of these and perhaps other shocks, we evaluate the predictive power of a set of standard in ation determinants. We proceed with a Bayesian model selection analysis which is particularly suited to select relevant regressors among a wide pool of candidate explanatory variables. A detailed description of the methodology is available in Ciccareli and Mojon (2005). 23 Estimation is done using Maximum Likelihood methods and a factor model approach a la Forni et al (2000) where the k common factors are generated by q shocks, where q < k. Note that the system is overidentied. Likelihood tests for overidenti cation restrictions reject the null of up to 2 international factors and one shock against the unrestricted alternative of having full rank. Results of this part are therefore based on a speci cation with three international factors generated by one common shock. 24 Because spillovers could in principle depend on the degree of monetary cooperation between countries (Clarida et al 2002) we tested in cross section whether the bilateral spill overs were correlatated with the variance of bilateral exchange rates. This correlation is based on 42 observations of bilateral spill overs and exchange rate variability among the G7 countries for either the full sample, 1961 to 1984 and 1985 to 2007. These correlations are very close to zero. 12 We limit our analysis to a number of variables widely argued to either a ect or help forecast in ation. Among these, we include two indices of commodity prices, variables that should be correlated with marginal costs, i.e. real GDP growth, unit labor costs and wages and asset prices. We also consider the possibility that the U.S. macroeconomic policies trigger in ationary pressures both at home and abroad. We measure the U.S. scal stance by the scal de cit. This allow us to test in particular the view that the increase in public deci t associated to the Vietnam war is correlated with the take o of Global In ation in the rst part of the sample. The potential spillovers of U.S. monetary policies on Global In ation, i.e. the McKinnon (1982) hypothesis, is also tested through testing by including the e ective exchange rate among the explanatory variables. Finally, we consider the stance of monetary policy. Because there is no consensus on how one should measure this concept, we investigate two possibilities. For the rst one, we compute a Taylor rule residual as follows: Taylor residualt = 1:5 where it is a short-term interest rate, t t 0:5(yt is in ation and yt yt ) it yt is the HP lter based output gap. For the second measure, we use the growth rate of a monetary aggregate (M3). For each variable, we extract a common factor in a similar way as we had done for in ation, i.e. taking unweighted averages of the variables of interest. However, because some of the variables are not available at quarterly frequencies early in the sample, indicators are obtained by averaging across national variables from the G7 and Australia. These averages explain usually between 1/3 (e.g. for real GDP growth) and 1/2 (e.g. for monetary aggregates) of the variance of national time series on average across countries.25 This extraction is not performed for U.S. scal de cit, the dollar exchange rate and the in ation of commoditiy prices for obvious reasons. We focus the analysis on the 1970-2006 sample because many of our variables were not available beforehand or since 2007. Within this 35 years sample we further check the stability of the results across the 1970-1990 sample and 1991-2006 sample. This latter sample should be particularly interesting with respect to the causality between monetary policies and Global In ation. In that period, central banks have aimed at stabilizing in ation around a constant explicit or implicit in ation target. As the common approach to monetary policy has become to anchor in ation expectations at a constant level, and constants cannot be correlated, indicators 25 Results using the dynamic factor or the existing OECD aggregates to compute the \Global" explanatory variables of in ation are quite similar to the ones reported here. The exact gures are available from the authors upon request. 13 of the stance of monetary policies should not help forecast Global In ation in that period. The results of the Bayesian selection algorithm are shown in Table 5. The prob column gives the probability that the variable is signi cant, i.e., the probabilty that the variable is included in the searched model, b gives the elasticity of global in ation vis-a-vis the variable and the last column gives the standard error of b. Several ndings are worth emphasizing. Looking rst at the 1970-2006 sample, only a few variables contain forecasting power with regards to Global In ation. Cost variables, including commodity prices and real GDP have a positive impact on Global in ation within 4 quarters. At 8 quarters horizon, stock prices, the dollar and indicators of the world stance of monetary help forecast in ation. In particluar, fast growth of M3 and smaller interest rates than the Taylor rule norm tend to be followed by higher in ation. This result strongly support the narrative evidence reported above and the conclusions of Cecchetti et al (2007) that the trend evolution of in ation in the OECD has been dominated by commonality in the stance of monetary policy and, possibly, by changes in the preferences of monetary authorities. We should also stress that the predictive power of the Dollar, though only over the full sample, comforms to the prediction of McKinnon (1983) that changes in the stance of monetary policies across countries might have been partially induced by spillovers of low frequency changes in stance of U.S. monetary policy. A weaker dollar has been followed, 8 quarters later, by a rise in Global In ation. The sub-sample results are also quite interesting. In particular, the information contents of M3 growth and the Taylor rule residuals are much less relevant for Global In ation eight quarters ahead, as one would expect given the success of central banks in stabilizing in ation. We even notice that a weaker stance of monetary policies, as de ned by more positive or less negative Taylor rule residual help forecast a decline in in ation at horizon 4 and 8 quarters. This may indicate that such deviations from the Taylor rule are much less persistent and more likely to be reverted in the recent subsample. The relevance of wages and house prices since 1990 also is worth underlining. A major shock that could have a ected wages similarly across countries in the last two decades is the emergence of China. However assessing whether this major labor supply shock has a ected the impact of OECD wages on in ation would require further research that goes beyond the scope of this paper. Turning to house prices, we know that there has been a worldwide acceleration of house prices since 1995 in most OECD countries (Germany and Japan being the notable exceptions). Our results suggest that central banks should monitor house prices, not only because they may relate to credit fed asset bubble cycles that put banking systems in 14 danger, but also because they carry relevant information for future in ation.26 Finally, both wages and house prices might surely play a valuable role to overcome the recently noticed unpredictability inpredictibility (Atkenson and Ohanian, 2001; Stock and Watson, 2007 and d'Agostino, Giannone and Surico, 2006). Overall, the ndings reported in this section reveal a robust sensitiveness of Global Ination to real and monetary determinants when measured at the global level. This reinforces the view that, possibly, economists working on in ation may need to reconsider the relevance of closed economy models of in ation.27 As a matter of fact, in a majority of OECD countries, reduced form models of the type we estimated for Global In ation are unable to obtain signi cant coe cients for any variables beyond the own lags of in ation itself (Corvoisier and Mojon, 2005). From this perspective, our results for Global In ation are good news because they show that there exist one level of aggregation at which leading indicators of in ation indeed contain exploitable information about future in ation. Finally, the response of Global In ation to both real determinants {at short horizons{ and monetary determinants {at longer horizon{ invite central banks to monitor both categories of in ation determinants. This surveillance, however, should be done not only at the level of countries, but also more globally to account for informational content of common international evolutions of in ation. 4 Predictive implications of Global In ation In this section, we describe the impact of Global In ation on national in ation rates. We show that Global In ation behaves as an attractor of the national in ation rates. This mechanism is important both to guide our understanding of the in ation process and to pursue practical policy purposes such as forecasting. 4.1 Global In ation is persistent and \attractive" Using the simple framework of the factor representation of section 2, it is easy to show that domestic in ation reverts to the global component { which acts as an \attractor" { and is characterised by stationary uctuations around the latter. In the factor representation, an estimate of national in ation is simply given by 26 27 ^ i f^t "^i;t = i;t "^i;t = ^i;t 1 i" + i;t See the discussion in Borio and Lowe (2002). This point is also made by Borio and Filardo (2007) though for di erent reasons. 15 where ^ i is an estimate of the country-speci c loading and f^t is our preferred measure of global in ation. To check whether domestic in ation reverts to the global component it su ces to check the stationarity of "^i;t . Table 6 reports the estimates of the rst autoregressive coe cients in an AR(1) representation for "^i;t , while Table 7 reports the ^ i .28 Estimates of i are on average i not higher than 0.5, which implies that, in the available sample of countries, a temporary shock to in ation is on average absorbed in ve to seven quarters at most. However, the same estimate for the global in ation is on average much higher than those of countries' in ations, both on the whole sample and on single subsamples. On the whole sample, therefore, the global component captures the most persistent and possibly non-stationary part of in ation.29 This result is in line with the nding that estimated global factor would capture the non-stationarity of the data used to estimate the factor (see e.g. Bai and Ng 2002). In this case, then, the global in ation behaves as an attractor and domestic in ation uctuates around its projection on this attractor. Incidentally, the importance of the global component of in ation leads us to reconsider the debate on in ation persistence. Two main conclusions emerge from the recent studies on in ation persistence. First, empirical estimates of in ation persistence fall when statistically signi cant shifts or breaks in the mean of in ation are accounted for.30 Second, the question of what drives the break in the mean has not received a clear answer yet.31 Both evidence on the importance of the mean of in ation and of common patterns in possible breaks in the mean are consistent with the view that in ation is a global phenomenon. Therefore, consistently with our ndings and considering our measure of global in ation as a common long run mean, we can conclude also that in ation of 22 OECD countries exhibit lower persistence once we control for the dependence of the national in ation processes on Global In ation. In a previous version of this work, we have also shown that in ation persistence might have not been stable over time.32 The question of stability is relevant from an econometric point of view, as any measure of persistence of a time-varying structure is biased if time variation is not accounted for. Results here broadly con rm the time-varying ones, where the global factor captures the 28 For this exercise we use annualised quarter-on-quarter trasformations of seasonally adjusted in ation series, 1). i;t = 400 (Pi;t =Pi;t 1 29 With a year-on-year transformation, the AR(1) coe cient of the global component is not di erent from one, whereas the average coe cient of country in ations is not higher than 0.85. 30 Robalo Marques (2004), among others, has recently argued that the mean of in ation plays a crucial role in the de nition of persistence and that any estimate of persistence should be seen conditional on a given assumption for the mean of in ation. 31 See for instance the discussion by Rogo (2003). 32 See Ciccarelli and Mojon (2005) and Mumtaz and Surico (2008). i.e. 16 persistent component of in ation on the whole sample, and its persistence declines ove the last 10-15 years.33 It is also worth noting that the estimates of the loadings of in ation rates on Global In ation are evenly distributed, across countries, around one (see Table 7). These loadings summarizes the \echo" of Global In ation changes on national in ation rates, on average, since 1960. Countries that have historically been considered as high in ation countries, have, not surpringly a loading higher than one. Germany, on the contrary, has the lowest loading among G7 countries. The sub-sample results, with notable exceptions, would also indicate some sort of convergence to more similar values. 4.2 A new benchmark for forecasting in ation? A well documented result in the forecasting literature is that reliable leading indicators of in ation are scarce. For example, Stock and Watson (1999, 2003), Banerjee et al (2003) and Banerjee and Marcellino (2002) all conclude that, while some leading indicators of in ation outperform the forecasts based on simple AR(p) models of in ation in some countries and for some sample periods, none has yet emerged that systematically beat the AR(p) (typically AR(1) or AR(2) of level in ation), or even the Random walk (RW). It has been also argued therefore that { especially over the last 10-15 years { in ation has become harder to forecast, in the sense that it has become much more di cult for an in ation forecaster to provide value added beyond a univariate model (Stock and Watson, 2005). This nding had already been documented by Atkeson and Ohanian (2001) { who found that backwards-looking Phillips curve forecasts were inferior to a RW forecast{ and more recently by D'Agostino et al.(2005) { who show that the ability to predict several measures of in ation and real activity declined remarkably, relative to na•ve forecasts, since the mid-1980s. All this literature, however, only focuses on the U.S. economy. To the best of our knowledge, a systematic comparison of similar features for the other industrialised economies has not been carried out yet. The issue is also partially related to the debate on the Great Moderation, which has also not attained an international momentum. Both topics are on our current research agenda. For our purposes here, however, the previous discussion can help shed new light on the issue of predictability of in ation, particularly by taking into account the international commonalities of in ation. The question is: Can the international environment help predict national in ation? In this section, we sketch an answer by considering a parsimonious speci cation simply 33 For a similar result with disaggregate data see also Angeloni et al. (2006). 17 augmented with the global component of in ation. Consistently with previous sections, we start from the usual common h-step ahead speci cation (Stock and Watson, 2002) h i;t+h = h i;0 + h i;1 (L) + i;t h ^ i;2 (L) ft + ui;t+h (2) where the factor f^t ; instead of summarising hundreds of series, is simply the common component of the 22 national in ation series: = i;t i ft + "i;t estimated with the average or the static principal component approach. Notation and strategy are similar to the ones employed by Stock and Watson (2002). In particular, the multistep forecasts is linear in f^t and t (and lags), and an h-step-ahead projection is used to construct the forecasts directly. Therefore, after estimating all unknown up to time T , we use (2) and forecast h from ^ h + ^ h (L) i;T + ^ h (L) f^T for each unit i and step h. The dependent variable is de ned i;0 h as i;t annualized in ation in the price level Pt { whereas i;t i;T +h i;1 i;2 = (400=h) ln (Pt =Pt 1 i;t = h) { the h-period is the quarter-on-quarter quarterly in ation rate. As said, this speci cation is a well known factor-augmented econometric relationship. Here we simply argue that a very parsimonious search of the factor ft , only based on an average of in ation series, can outperform or be as competitive as the usual benchmarks, without the need to choose an optimal number of factors from hundreds of variables. The important issue, however, is the appropriate consideration of an international ingredient summarised in ft , which, as noted previously, works as an attractor for national in ations. Because we want to keep the discussion limited to the scope of the paper, we check the forecasting performance of our model only against three natural competitors. The rst one is an AR(p) of the form h i;t+h = i;0 + i;1 (L) i;t + "i;t+h (3) where the lag length p is imposed equal to 1 or optimally chosen with a standard BIC. The second model is a Random Walk (RW) h i;t+h = i;t + "it+h (4) We pay particular attention to this naive speci cation { especially on recent samples { to check the issue of the unpredictability raised e.g. by D'Agostino, Giannone and Surico (2005) and Stock and Watson (2007). 18 A third benchmark can be considered along the lines of Stock and Watson (1999), Nicoletti-Altimari (2000) and Gerlach (2003) by setting an augmented Phillips curve model where the rst di erence of in ation depends on its own lags and on the lags of the growth rates of industrial production, oil price and M3. Speci cally, it is: h i;t+h = i;0 + + i;1 (L) i;3 (L) i;t + M 3it + i;2 (L) i;4 (L) IPit Oilt + "i;t+h The experiment is conducted in a \pseudo real-time" framework with all models re-estimated at each step using only information up to time t. We choose the lag length to be one or two for the AR component of each model, while xing to four the number of lags of ft . The evaluation and comparison are made over three forecasting periods, 1980-2004, 1980-95 and post 1995, and for eight forecasting horizons (quarters). We report results only for the last subsample at the one-year ahead horizon. The choice of the subsamples is motivated by the issue of unpreditability over the last 15-20 years, and the choice of the horizon by the policy relevance.34 Tables 8 report the RMSE of our preferred speci cation (2) relative to the RMSE of the four competing models. Clearly, our speci cation is preferred in a forecasting sense if the reported statistic is lower than one. A rough comparison across the three benchmarks can also be made: the bigger the reported statitics for a model the better its performance with respect to the others. So, for instance, if the reported statistics for RW is greater than the one for AR, then the former is preferred. The signi cance of these ratios is checked with a simple test on the di erence between two competing Mean Squared Forecast Errors, adjusting the statistic when models are nested (Clark and West 2007). Bold entries in the table denote 10 percent signi cance. Overall results show that our model can outperform the competing models in forecasting in ation on average, across forecast horizon and over evaluation periods, and for most countries. Improvements are of the order of up to 16 percent with respect to the augmented Phillips curve speci cation, 15 percent with respect to the RW and 20 percent with respect to the standard AR. Our model seems to perform particularly well on the 4-quarter-ahed horizon, which is the most relevant one for policymaking, and over the last 10 years, where the unpredictability related to the Great Moderation should be more evident. These conclusions are consistent both with the fact that the Global In ation works as an anchor for national in ations and with a somewhat expected greater commonality among 34 Results for other subsamples and horizons are not reported because qualitatively very similar. A version of them are available in Ciccarelli and Mojon (2005). 19 in ations from the nineties (e.g. Rogo , 2003). Our interpretation of the results is that the unpredictability of in ation is related to the use of \local" vs. \global" models, and might not necessarily be true for all industrialized countries. Our ndings are remarkable, and have the potential features of a new benchmark for forecasting in ation, based on the incorporation of global information in the standard models. Note nally that over the evaluation sample 1995-2004 there is indeed an issue of unpredictability for the average (or median) country, in that the reported statistics for RW are greater than the one for AR and Phillips curve. However this ratio is lower than 1 (and signi cantly so for many countries), meaning that our preferred speci cation is able to beat on average the random walk. Our preliminary conclusion, then, is that a simple parsimonious extension of a standard AR model, where we consider the attraction role of the Global In ation, outperform robust predictors of in ation. The results con rm also the importance of exploiting the international links and commonalities as advocated by the recent empirical Factor-Model literature. What makes our contribution particularly valuable is the search of the factors in a global rather than a domestic information set, and the interpretation of the common factor as an attractor of national in ation. Our parsimonious speci cation, where the global factor is a simple average of 22 national in ation series, seems to forecast well future developments of national in ation. The latter result, which holds across countries, samples periods and forecasting horizons, is obviously one of the main contributions of our current research that deserves further investigation. 5 Conclusions In this paper, we have shown that the in ation of the OECD countries have moved together over the last 45 years. This comovement accounts for 70 percent of the variability of country in ation, on average. Moreover, there is a powerful and robust \error correction mechanism" that brings national in ation rates back toward the level of their long term projection on Global In ation. As a rst practical application of the idea of Global In ation, we present a fairly parsimonious model of in ation forecast. The preliminary ndings suggest that the new speci cation beats standard competitors. The main open question is to assess whether these results re ect some sort of statistical \return to the mean" phenomenon or whether some deeper endogenous economic adjustments are at work. For example, some determinants of in ation are global: the price of commodities is the same for all countries; KOW have shown that there is a global business cycle; last but not least, it seems that monetary policy concepts are e ectively spreading among central banks. In 20 some periods, bad monetary policy strategies are dominating for a majority of countries. At other times, good strategies appear dominant. We show that Global In ation does not result from countries spillovers but rather from common shocks. It responds to commodity prices, the global business cycle and the global trends of liquidity. We further qualify that real developments are more relevant at short horizons and monetary developments matter at longer horizons. Our paper has two important policy implications. First, given the importance of Global In ation for local in ation, the nature of Global In ation brings support to the monetary policy strategies that give importance both to real and monetary developments in their assessment of in ationary pressures. Second, there may be a useful informational content in the average of foreign in ation records, even for countries that, like Switzerland, were on average less a ected by common developments in in ation. Future research to which the authors will contribute should follow mainly three directions. The rst one is to extend the sample of countries and regions to emerging markets, and assess the importance of global, regional and domestic mechanisms that help explain in ation developments. The second one is to explore more systematically the forecasting performance of the Global In ation Model, and compare it with the performance of other univariate and multivariate speci cations, across other samples and cross sections of countries. Finally, we should try to gain insights on the nature of the shocks that drive Global In ation and their transmission to country in ations. Our belief is that to a large extent the results reported in this paper may re ect the importance for central bankers of exchanging views and cooperating in the design of their monetary policy concepts. Hence, paraphrasing the conclusion of the 1848 Communist Manifesto we would like to invite: \central bankers of all countries: unite!" 21 References [1] D'Agostino, A. D. Giannone and P. Surico (2005), (Un)predictability and Macroeconomic Stability, Mimeo, ECARES. [2] Andrews, D. and W.K. Chen (1994), Approximately median-unbiased estimation of autoregresive models, Journal of Business and Economic Statistics 12, 187-204. [3] Angeloni, I., L. Aucremanne and M. Ciccarelli (2006), Price setting and in ation persistence: did EMU matter?, Economic Policy 21, 353-387. [4] Atkenson, A. and L. Ohanian (2001), Are Phillips Curves Useful for forecasting in ation?, Federal Reserve Bank of Mineapolis Quarterly Review 25, 2-11. [5] Bai, J. (2003), Inferential theory for factor models of large dimensions, Econometrica 71, 135-171. [6] Bai, J., and S. Ng (2002), Determine the Number of Factors in Approximate Factor Models, Econometrica 70, 191-221. [7] Ball L. (2006), Has Globalization Changed In ation?, NBER Working Paper 12687. [8] Ball L. and N. Sheridan (2003), Does in ation targeting matter?, NBER Working Paper 9577. [9] Banerjee A., M. Marcellino and I. Masten (2003), Leading indicators of euro area in ation and GDP growth, CEPR WP 3893. [10] Banerjee A. and M. Marcellino (2003), Are there any reliable leading indicators for the US in ation and GDP growth?, IGIER WP 236. [11] Baxter, M. and R. King (1999), Measuring Business Cycles: Approximate Band-Pass Filters for Economic Time Series, Review of Economics and Statistics 81, 575-93. [12] BIS (2005), Annual Report 2005, Bank of International Settlements, www.bis.org. [13] Besley T. and A. Case (1995), Incumbent Behavior: Vote-Seeking, Tax-Setting, and Yardstick Competition, American Economic Review 85, 25-45. [14] Borio, C. and A. Filardo (2007), Globalisation and In ation: New Cross-Country Evidence on Global Determinants of Domestic In ation, BIS Working Paper 227. 22 [15] Borio, C. and P. Lowe (2002), Asset prices, nancial and monetary stability: exploring the nexus, BIS Working Papers 114. [16] Canova F., M. Ciccarelli and E. Ortega (2007), Similarities and convergence in G7 cycles, Journal of Monetary Economics 54, 850-78 . [17] Calvo G. and C. Reihnart (2002), Fear of Floating, Quaterly Journal of Economics 117 (2): 379-408. [18] Cecchetti, S., P. Hooper, B. Kasman, K. Schoenholtz and M. Watson (2007) Understanding the Evolving In ation Process, Proceedings of the U.S. Monetary Policy Forum 2007. [19] Chen, N., J. Imbs and A. Scott (2004), Competition, Globalization and the Decline of In ation, CEPR Discussion paper 4695. [20] Ciccarelli M. and B. Mojon (2005), Global In ation, ECB WPS no 537. [21] Cogley T. and T. J. Sargent (2005), The Conquest of US In ation: Learning and Robustness to Model Uncertainty, Review of Economics Dynamics 8, 528-563. [22] Cristadoro R., M. Forni, L. Reichlin and G. Veronese (2005): A core in ation indicator for the euro area, Journal of Money, Credit and Banking 37, 539-60. [23] Corvoisier S. and B. Mojon (2005), Breaks in the mean of in ation: How they happen and what to do with them, ECB Working Paper 451. [24] Darby, M. and J. Lothian (1983), The International Transmission of In ation, National Bureau of Economic Research, University of Chicago Press. [25] Doz, C. D. Giannone and L. Reichlin (2006), A Maximum Likelihood Approach for Large Approximate Dynamic Factor Models, ECB Working Paper 674. [26] Forni, M. and L. Reichlin, (2001), Federal policies and local economies: Europe and the US, European Economic Review 45(1), 109-134 [27] Forni, M., M. Hallin, M. Lippi, and L. Reichlin (2000): The generalized factor model: identi cation and estimation, The Review of Economics and Statistics 82: 540-554. [28] Gerlach S. (2003), The ECB's two pillars, CEPR Discussion paper 3689. [29] Goodfriend, M., (2007), How the world achieved consensus on monetary policy, Journal of Economic Perspectives 21, 47{68. 23 [30] Kose A,. C. Otrok and C. Whiteman (2003), International business cycles: World, region and country-speci c factors, The American Economic Review 93, 1216-39. [31] Ihrig J., S. Kamin, D. Lindner and J. Marquez (2007), Some simple tests of the Globalization and In ation Hypothesis, Federal Reserve Board, International Finance Discussion Papers 891. [32] Levin A.T. and J.M. Piger (2004), Is in ation persistence intrinsic in industrial economies?, ECB Working Paper 334. [33] McKinnon R. I. (1982), Currency substituion and in instability in the World Dollar Standard, American Economic Review 72, 320-333. [34] Monacelli T. and L. Sala (2007), The International Dimension of In ation: Evidence from Disaggregated Consumer Price Data, IGIER mimeo. [35] Mumtaz H. and P. Surico (2008), Evolving International In ation Dynamics: World and Country Speci c Factors, CEPR Working Paper 6767. [36] Nicoletti-Altimari S. (2001), Does money lead in ation in the euro area?, ECB Working Paper 63. [37] Reinhart C. and K. Rogo (2002), The Modern History of Exchange Rate Arrangements: A Reinterpretation , NBER WP 8963. [38] Robalo Marques C. (2004), In ation persistence: facts or artefacts?, ECB Working Paper 371. [39] Rogo K. (2003), Globalization and Global Disin ation, Federal Reserve Bank of Kansas City Economic Review, 4th Quarter. [40] Rogo K. (2006), Impact of Globalization on Monetary Policy, Proceedings of the Federal Reserve Bank of Kansas City Jackson Hole Conference. [41] Stock J. and M. Watson (1999), Forecasting in ation, Journal of Monetary Economics 44, 293-335. [42] Stock J. and M. Watson (2002), Macroeconomic Forecasting Using Di usion Indexes, Journal of Business and Economic Statistics 20, 147-162. [43] Stock J. and M. Watson (2005), Understanding Changes in International Business Cycle Dynamics, Journal of the European Economic Association 3, 966-1006. 24 [44] Stock J. and M. Watson (2007), Has in ation become harder to forecast?, Journal of Money, Banking and Credit 39, 3-33. [45] Wang P. and Y. Wen (2007), In ation Dynamics: A cross-country investigation, Journal of Monetary Economics 54, 2004-2031. [46] Woodford M. (2007), Globalization and Monetary Control, in J. Gali and M. J. Gertler (editors), International Dimensions of Monetary Policy, University of Chicago Press. 25 APPENDIX. Data source and transformation Definition Source Transformation Consummer price indices OECD Main Economic Indicators y-o-y growth rates Hourly earnings OECD Main Economic Indicators y-o-y growth rates Industrial production IMF International Financial Statistics y-o-y growth rates Short-term interest rate (3-month) OECD Economic outlook level Long-term interest rate (10-year) OECD Economic outlook level GDP Eurostat and OECD Economic outlook y-o-y growth rates Commodity prices Bridge/Commodity Research Bureau; Spot market price index: All commodities; www.freelunch.com y-o-y growth rates Oil price Fed St Louis Oil price: Domestic West Texas Intermediate y-o-y growth rates US government fiscal deficit Net lending or net borrowing (-); Table 3.2. Federal Government Current Receipts and Expenditures; Bureau of economic analysis BIS unpublished data base, Borio and Lowe (2002). level Stock prices Real estate prices, housing indices BIS unpublished data base, Borio and Lowe (2002). Broad money (M3) y-o-y growth rates y-o-y growth rates euro area countries (Eurostat Balance sheet items); Canada, Denmark, Sweden and United Kingddom y-o-y growth rates (OECD MEI); Australia, Japan, New Zeland, Norway Switzerland and United States (OECD Economic Outlook); for Austria, Belgium, Finland, France, Germany, Ireland Table 1. Share of inflation variance explained by alternative measures of Global Inflation Average OECD Greece Switzerland Japan Netherlands Germany New Zeland Portugal United States Norway Australia Denmark Austria Spain Sweden Luxembourg United Kindom Finland Canada Belgium Ireland Italy France 0.41 0.46 0.55 0.58 0.60 0.63 0.65 0.69 0.70 0.73 0.73 0.74 0.75 0.76 0.76 0.82 0.83 0.83 0.84 0.85 0.86 0.89 0.65 0.24 0.29 0.25 0.33 0.61 0.65 0.72 0.57 0.70 0.53 0.40 0.58 0.63 0.48 0.66 0.58 0.75 0.56 0.61 0.81 0.73 Static factor first second 0.37 0.16 0.41 0.16 0.52 0.15 0.57 0.20 0.57 0.16 0.61 0.14 0.63 0.09 0.67 0.02 0.68 0.03 0.71 0.06 0.71 0.00 0.72 0.12 0.74 0.03 0.71 0.02 0.77 0.02 0.80 0.00 0.81 0.01 0.81 0.04 0.84 0.03 0.86 0.00 0.86 0.03 0.89 0.00 mean median 0.71 0.73 0.56 0.59 0.69 0.71 0.07 0.03 euro area 0.95 0.76 0.95 0.00 Note: 1961:1-2007:2. The euro area aggregate inflation is not included in the pool of 22 countries used to estimate Global Inflation. Table 2: Share of inflation variance explained by average inflation for a selection of six countries 1961-2007 1975-2007 1984-2007 Australia Canada Germany UK Japan US 0.71 0.83 0.63 0.88 0.61 0.78 0.70 0.86 0.63 0.91 0.83 0.81 0.37 0.69 0.26 0.83 0.66 0.67 mean median 0.74 0.75 0.79 0.82 0.58 0.67 Note: Global inflation is here defined as in column 1 of Table 1, i.e. as the unweighted average of the inflation rates of the six countries of this table. Table 3. Share of detrended inflation variance explained by alternative measures of Global Inflation Portugal Spain New Zeland Norway Greece Sweden Netherlands Denmark Germany Australia Canada Finland Luxembourg United Kingdom United States Switzerland Austria Japan Ireland France Italy Belgium Average 0.03 0.10 0.10 0.15 0.27 0.28 0.29 0.30 0.32 0.35 0.36 0.39 0.40 0.40 0.43 0.44 0.47 0.54 0.57 0.61 0.61 0.63 OECD 0.01 0.00 0.01 0.02 0.29 0.08 0.08 0.21 0.08 0.20 0.20 0.15 0.05 0.24 0.57 0.22 0.12 0.49 0.20 0.49 0.41 0.24 Static factor 0.02 0.03 0.05 0.09 0.19 0.17 0.29 0.21 0.26 0.26 0.29 0.34 0.40 0.35 0.42 0.33 0.43 0.46 0.57 0.64 0.60 0.64 mean median 0.37 0.37 0.20 0.20 0.32 0.31 Euro area 0.83 0.34 0.84 Note: 1961:1-2007:2. The inflation series are detrended by applying the band pass filter of Baxter and King (1999). The euro area aggregate inflation is not included in the pool of 22 countries used to estimate Global Inflation. Table 4: Decomposition of inflation dynamics into common shocks, international spill overs and countries own shocks 1961-2007 Canada horizon 0 4 16 (1) global shock 0.71 0.69 0.70 (2) (3)=(1)+(2) spillovers 0.00 0.14 0.19 common 0.71 0.83 0.89 1984-2007 1961-1983 (4) (2) (3)=(1)+(2) 0.29 0.17 0.11 (1) global shock 0.34 0.51 0.60 spillovers 0.00 0.09 0.14 common 0.34 0.60 0.74 own shock (4) (2) (3)=(1)+(2) 0.66 0.40 0.26 (1) global shock 0.43 0.47 0.47 spillovers 0.00 0.12 0.15 common 0.43 0.59 0.61 own shock (4) own shock 0.57 0.41 0.39 France 0 4 16 0.37 0.57 0.64 0.00 0.07 0.11 0.37 0.64 0.75 0.63 0.37 0.25 0.22 0.51 0.60 0.00 0.07 0.11 0.22 0.58 0.71 0.78 0.42 0.29 0.57 0.47 0.45 0.00 0.16 0.24 0.57 0.63 0.70 0.43 0.37 0.30 Germany 0 4 16 0.31 0.41 0.46 0.00 0.04 0.07 0.31 0.45 0.52 0.69 0.55 0.48 0.32 0.38 0.42 0.00 0.04 0.06 0.32 0.42 0.48 0.68 0.58 0.52 0.84 0.66 0.65 0.00 0.22 0.24 0.84 0.88 0.88 0.16 0.12 0.12 Italy 0 4 16 0.51 0.62 0.67 0.00 0.11 0.17 0.51 0.74 0.85 0.49 0.26 0.15 0.91 0.80 0.79 0.00 0.14 0.18 0.91 0.95 0.97 0.09 0.05 0.03 0.39 0.42 0.43 0.00 0.05 0.06 0.39 0.47 0.49 0.62 0.53 0.51 Japan 0 4 16 0.30 0.39 0.43 0.00 0.06 0.08 0.30 0.46 0.51 0.70 0.54 0.49 0.32 0.34 0.35 0.00 0.10 0.10 0.32 0.44 0.46 0.68 0.56 0.54 0.87 0.78 0.76 0.00 0.14 0.16 0.87 0.92 0.92 0.13 0.08 0.08 UK 0 4 16 0.87 0.78 0.76 0.00 0.14 0.18 0.87 0.92 0.95 0.13 0.08 0.05 0.96 0.81 0.79 0.00 0.17 0.19 0.96 0.98 0.98 0.04 0.02 0.02 0.23 0.36 0.37 0.00 0.13 0.16 0.23 0.49 0.54 0.77 0.51 0.46 US 0 4 16 0.42 0.46 0.54 0.00 0.07 0.11 0.42 0.53 0.64 0.58 0.47 0.36 0.67 0.66 0.69 0.00 0.05 0.07 0.67 0.72 0.76 0.33 0.28 0.24 0.61 0.60 0.60 0.00 0.06 0.06 0.61 0.66 0.66 0.39 0.34 0.34 Australia 0 4 16 0.09 0.25 0.39 0.00 0.08 0.13 0.09 0.33 0.52 0.91 0.67 0.48 0.10 0.32 0.45 0.00 0.09 0.13 0.10 0.41 0.58 0.90 0.59 0.42 0.16 0.26 0.27 0.00 0.08 0.11 0.16 0.35 0.37 0.84 0.65 0.63 average 0 4 16 0.45 0.52 0.57 0.00 0.09 0.13 0.45 0.61 0.70 0.55 0.39 0.30 0.48 0.54 0.59 0.00 0.09 0.12 0.48 0.64 0.71 0.52 0.36 0.29 0.51 0.50 0.50 0.00 0.12 0.15 0.51 0.62 0.65 0.49 0.38 0.35 Note: 1961:1-2007:2. Values in the table represent the fraction of forecast error variance due to each shock at the one-, four- and sixteen-quarter horizon. The results are based on a structural FAVAR model estimated with maximum likelihood methods and a factor model approach a la Stock and Watson (2005) and Forni et al (2000), where the k common factors are generated by q shocks, and q<k. Likelihood tests for overidentification restrictions reject the null of up to 2 international factors and one shock against the unrestricted alternative full rank. The results are therefore based on a specification with three international factors generated by one common shock. Table 5: BMA Posterior probabilities and estimates, dependent variable is Global Inflation 1 step ahead lag of own comm. price oil price GDP ULC Wages Stock prices House prices U.S. fiscal deficit Dollar effective x rate Taylor residual M3 1970-2006 1970-1990 1990-2006 prob. 1.00 0.98 0.29 0.90 0.11 0.29 0.10 0.09 0.08 0.21 0.08 0.42 b 0.90 0.03 0.00 0.13 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.02 std 0.06 0.01 0.01 0.07 0.03 0.04 0.00 0.01 0.02 0.01 0.01 0.03 prob. 1.00 0.95 0.17 0.38 0.15 0.47 0.10 0.10 0.08 0.41 0.13 0.40 b 0.81 0.04 0.00 0.03 0.01 0.09 0.00 0.00 0.00 0.01 0.00 0.02 std 0.13 0.02 0.00 0.05 0.04 0.11 0.00 0.01 0.02 0.01 0.02 0.04 prob. 1.00 0.51 0.11 0.14 0.10 0.17 0.15 0.10 0.19 0.14 0.16 0.08 b 0.86 0.01 0.00 0.01 0.00 0.02 0.00 0.00 0.02 0.00 -0.01 0.00 std 0.08 0.01 0.00 0.03 0.02 0.07 0.00 0.01 0.05 0.00 0.02 0.01 1.00 0.23 1.00 0.85 0.17 0.10 0.09 0.13 0.08 0.19 0.21 1.00 0.61 0.01 0.06 0.23 0.02 0.01 0.00 0.00 0.00 0.00 0.02 0.26 0.09 0.01 0.01 0.13 0.07 0.04 0.00 0.02 0.04 0.01 0.04 0.05 0.20 0.32 0.39 0.98 0.98 0.70 0.10 0.18 0.16 0.17 0.76 0.50 0.02 0.02 0.01 0.42 0.56 0.23 0.00 -0.01 -0.02 0.00 0.13 0.08 0.12 0.03 0.01 0.14 0.16 0.19 0.00 0.03 0.08 0.01 0.10 0.09 0.13 0.19 0.12 0.14 0.47 1.00 0.26 0.18 0.23 0.98 0.97 0.56 0.01 0.00 0.00 -0.01 -0.10 0.49 0.00 -0.01 0.02 -0.03 -0.21 0.07 0.07 0.01 0.00 0.05 0.13 0.10 0.00 0.02 0.06 0.01 0.07 0.08 0.21 0.36 0.20 0.17 0.13 0.10 0.93 0.22 0.30 0.95 1.00 1.00 -0.03 0.02 0.00 0.02 -0.01 0.00 0.021 -0.02 -0.08 -0.06 0.41 0.58 0.08 0.03 0.01 0.07 0.04 0.04 0.009 0.04 0.14 0.02 0.06 0.06 0.18 0.62 0.10 0.15 0.18 0.19 0.67 0.09 0.09 0.82 1.00 1.00 0.02 0.04 0.00 0.02 0.02 0.03 0.01 0.00 -0.01 -0.04 0.39 0.43 0.08 0.04 0.01 0.08 0.08 0.09 0.01 0.02 0.06 0.03 0.08 0.08 0.98 0.26 0.11 0.90 0.53 0.48 0.25 0.92 0.12 0.24 0.92 0.41 0.58 0.01 0.00 -0.33 -0.13 -0.15 0.00 0.16 -0.01 0.00 -0.33 -0.06 0.26 0.02 0.00 0.17 0.16 0.19 0.00 0.07 0.04 0.01 0.14 0.09 4 steps ahead own comm. price oil price GDP ULC Wages Stock prices House prices U.S. fiscal deficit Dollar effective x rate Taylor residual M3 8 steps ahead own comm. price oil price GDP ULC Wages Stock prices House prices U.S. fiscal deficit Dollar effective x rate Taylor residual M3 Note: The three columns of numbers for each sample-panel report the probability that the corresponding variable help predict Global Inflation, the estimated coefficient and its standard deviation respectively. Probability higher than 0.5 and significant coefficients are in bold. The dependent variable is Global inflation. Potential explanatory variables enter with one lag. The estimation and search technique are explained in Ciccarelli and Mojon (2005). Table 6: Persistence of the Global and the National Components of Inflation 1960-2007 ρ stderr Euro area 1960-1980 ρ stderr 1980-1990 ρ stderr 1990-2007 ρ stderr 0.48 0.06 0.23 0.11 0.36 0.14 0.55 0.10 United States Canada United Kingdom Japan Germany France Italy 0.63 0.35 0.45 0.54 0.54 0.55 0.44 0.06 0.07 0.07 0.06 0.06 0.06 0.07 0.68 0.30 0.34 0.37 0.47 0.50 0.43 0.09 0.11 0.11 0.11 0.10 0.10 0.10 0.47 0.61 0.52 0.33 0.40 0.24 0.03 0.12 0.12 0.11 0.14 0.14 0.15 0.15 0.15 0.28 0.17 0.03 0.42 0.19 0.52 0.12 0.12 0.12 0.12 0.11 0.12 0.11 median mean 0.54 0.50 0.06 0.06 0.43 0.44 0.10 0.10 0.40 0.37 0.14 0.13 0.19 0.25 0.12 0.12 Other Euro/EU Austria Belgium Denmark Finland Greece Ireland Luxembourg Portugal Spain Sweden The Netherlands 0.18 0.42 0.10 0.44 0.73 0.19 0.48 0.05 0.37 0.18 0.41 0.07 0.07 0.07 0.07 0.05 0.07 0.06 0.07 0.07 0.07 0.07 0.10 0.42 -0.05 0.42 0.41 0.02 0.44 -0.14 0.39 0.02 0.16 0.11 0.10 0.11 0.10 0.10 0.11 0.10 0.11 0.10 0.12 0.11 0.32 0.52 -0.03 0.37 0.39 -0.34 0.48 0.61 0.16 0.23 0.07 0.15 0.13 0.16 0.15 0.14 0.14 0.13 0.11 0.15 0.14 0.16 0.17 -0.04 0.30 0.37 0.44 0.61 0.03 0.22 -0.13 0.10 0.42 0.12 0.12 0.11 0.11 0.11 0.10 0.12 0.12 0.12 0.12 0.11 median mean 0.37 0.32 0.07 0.07 0.16 0.20 0.11 0.11 0.32 0.25 0.14 0.14 0.22 0.23 0.12 0.11 0.28 0.59 0.17 0.61 0.07 0.06 0.07 0.06 -0.09 0.57 0.09 0.63 0.11 0.09 0.11 0.09 0.32 0.51 0.44 0.33 0.15 0.13 0.14 0.15 0.20 0.45 -0.16 0.29 0.12 0.11 0.13 0.12 0.43 0.41 0.07 0.07 0.33 0.30 0.11 0.11 0.38 0.40 0.14 0.14 0.24 0.20 0.12 0.12 0.94 0.02 0.92 0.04 0.88 0.05 0.69 0.08 G7 Others Australia New Zeland Norway Switzerland median mean Global Inflation Note: The table reports the estimates of the first autoregressive coefficient of national inflations (defined as πit-λift) and their standard errors over four samples. An AR(1) is asssumed. Estimation technique is OLS. The factor is estimated with a simple average. Table 7: Loadings of National Inflation Rates on Global Inflation λ Euro area 1960-2007 stderr λ 1960-1980 stderr λ 1980-1990 stderr λ 1990-2007 stderr 0.76 0.02 0.73 0.03 1.01 0.03 0.85 0.05 United States Canada United Kingdom Japan Germany France Italy 0.71 0.84 1.34 0.91 0.42 1.04 1.48 0.04 0.04 0.06 0.08 0.03 0.03 0.05 0.85 0.84 1.54 0.81 0.37 0.93 1.62 0.07 0.07 0.11 0.14 0.05 0.06 0.10 0.90 1.00 1.10 0.61 0.70 1.44 1.78 0.13 0.08 0.14 0.10 0.07 0.09 0.12 0.80 0.91 1.24 0.65 0.86 0.60 1.03 0.10 0.16 0.14 0.15 0.12 0.06 0.09 median mean 0.91 0.96 0.04 0.05 0.85 0.99 0.07 0.08 1.00 1.07 0.10 0.10 0.86 0.87 0.12 0.12 0.54 0.77 0.91 1.12 1.53 1.44 0.70 2.02 1.35 0.96 0.57 0.04 0.04 0.07 0.06 0.14 0.07 0.04 0.15 0.07 0.06 0.05 0.49 0.79 0.74 1.04 1.99 1.46 0.70 2.06 1.30 0.83 0.52 0.08 0.07 0.14 0.11 0.18 0.13 0.06 0.29 0.15 0.10 0.09 0.67 0.93 0.98 0.92 0.87 2.25 1.03 1.24 1.20 0.86 0.90 0.08 0.09 0.14 0.10 0.23 0.18 0.13 0.35 0.11 0.15 0.06 0.67 0.61 0.15 0.92 3.68 0.28 0.62 2.10 1.03 2.09 0.35 0.09 0.09 0.08 0.11 0.32 0.16 0.10 0.16 0.10 0.20 0.09 0.96 1.08 0.07 0.08 0.83 1.08 0.13 0.14 0.93 1.08 0.15 0.17 0.67 1.14 0.16 0.16 1.04 1.26 0.83 0.47 0.06 0.08 0.06 0.05 1.24 1.22 0.67 0.26 0.08 0.10 0.11 0.09 0.39 0.88 0.81 0.52 0.15 0.33 0.14 0.12 0.87 0.59 0.80 1.29 0.21 0.15 0.18 0.08 0.94 0.90 0.06 0.07 0.94 0.85 0.10 0.10 0.81 0.65 0.15 0.18 0.80 0.89 0.16 0.15 G7 Other Euro/EU Austria Belgium Denmark Finland Greece Ireland Luxembourg Portugal Spain Sweden The Netherlands median mean Others Australia New Zeland Norway Switzerland median mean Note: The table reports the estimates of the factor loading in the equation πit = λift+εit and its standard errors for all countries and over four samples. Estimation technique is OLS. The factor is estimated with a simple average. Table 8: RMSE of the Global Inflation model relative to standard benchmarks (1995-2006) Euro area total RW 0.77 1-step ahead AR PHIL 0.76 0.76 RW 0.87 4-step ahead AR PHIL 0.83 0.84 RW 0.86 8-step ahead AR PHIL 0.79 0.78 G7 United States Canada United Kingdom Japan Germany France Italy 0.90 0.81 0.96 1.17 0.78 0.86 1.36 0.96 0.87 0.76 1.06 0.84 0.86 0.95 0.93 0.89 0.81 1.34 0.84 0.87 1.13 0.92 0.77 1.01 1.14 0.81 0.86 1.02 0.97 0.84 0.65 0.96 0.86 0.81 0.64 0.96 0.85 0.69 1.37 0.87 0.85 0.75 0.85 0.76 1.02 1.18 0.82 0.80 0.92 0.89 0.83 0.58 0.89 0.87 0.75 0.52 0.91 0.84 0.61 1.57 0.91 0.82 0.59 median mean 0.90 0.98 0.87 0.90 0.89 0.97 0.92 0.93 0.84 0.82 0.85 0.90 0.85 0.91 0.83 0.76 0.84 0.89 Other Euro/EU Austria Belgium Denmark Finland Greece Ireland Luxembourg Portugal Spain Sweden The Netherlands 0.88 0.69 1.35 0.97 1.06 1.37 0.84 1.30 0.76 1.03 0.87 0.75 0.75 0.62 0.90 0.95 1.25 0.90 0.39 0.66 0.76 0.81 0.76 0.74 0.81 0.91 1.02 1.16 0.91 0.44 0.68 0.73 0.82 0.80 0.77 1.36 0.94 1.03 1.02 0.86 0.98 0.85 0.96 0.78 0.69 0.82 0.59 0.83 0.76 1.07 0.92 0.39 0.66 0.71 0.73 0.73 0.83 0.86 0.82 0.79 0.87 0.91 0.41 0.70 0.70 0.78 0.69 0.64 1.12 0.96 0.98 0.97 0.84 1.11 0.82 1.05 0.88 0.63 0.69 0.53 0.80 0.69 1.05 0.93 0.36 0.59 0.72 0.82 0.66 0.68 0.81 0.82 0.71 0.88 0.92 0.39 0.65 0.71 0.83 median mean 0.97 1.01 0.76 0.79 0.81 0.82 0.94 0.94 0.73 0.74 0.79 0.76 0.96 0.91 0.69 0.71 0.71 0.73 Australia New Zeland Norway Switzerland 0.79 0.85 0.75 1.06 0.85 0.80 0.82 0.94 0.84 0.75 0.81 0.95 0.83 0.83 0.74 1.25 0.88 0.73 0.79 1.01 0.89 0.71 0.78 0.96 0.74 0.83 0.74 1.31 0.81 0.67 0.77 0.98 0.80 0.61 0.77 0.99 0.82 0.86 0.84 0.85 0.83 0.84 0.83 0.91 0.83 0.85 0.83 0.83 0.79 0.91 0.79 0.81 0.79 0.79 0.90 0.96 0.84 0.83 0.84 0.87 0.92 0.93 0.81 0.79 0.83 0.82 0.88 0.91 0.76 0.75 0.80 0.79 Others median mean Overall median Overall mean Note: entries are the ratios of root mean squared errors of the Global inflation forecast model to the one obtained with a random walk (RW), an AR(1) (AR), and a Phillips Curve augmented with industrial production, commodity prices and money (PHIL). Evaluation period: 1995-2006. Figure 1: Measures of Global Inflation United States, Japan, Canada and United Kingdom (1965-2004) 3.5 3 First static factor average OECD aggregate 2.5 2 1.5 1 0.5 0 -0.5 -1 -1.5 6 -0 3 Ju n -0 0 Ju n -0 7 Ju n -9 4 Ju n -9 1 Ju n -9 8 Ju n -8 5 Ju n -8 2 Ju n -8 9 Ju n -7 6 Ju n -7 3 Ju n -7 0 Ju n -7 7 Ju n -6 4 Ju n -6 Ju n Ju n -6 1 -2 Note: Three measure of Global Inflation: a simple cross-country average, the accregate OECD measure and a static factor. Figure 2a : G7 and euro area inflation and their projection on Global Inflation United States, Japan, Canada and United Kingdom (1961-2007) US Canada 16 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 0 -2 -2 -4 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 1960 1964 1968 1972 1976 Japan 1980 1984 1988 1992 1996 2000 2004 1988 1992 1996 2000 2004 UK 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 -5 -5 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 1960 1964 1968 1972 1976 1980 1984 Note: Domestic inflations and their projections on the Global Inflation (measured by simple average). Estimation technique: OLS. Dependent variable is a deseasonalized quarter-on-quarter inflation rate. Figure 2b : G7 and euro area inflation and their projection on Global Inflation United States, Japan, Canada and United Kingdom (1961-2007) Germany Italy 9.6 25 8.0 20 6.4 15 4.8 10 3.2 5 1.6 0 0.0 -1.6 -5 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 1960 1964 1968 1972 1976 France 1980 1984 1988 1992 1996 2000 2004 1988 1992 1996 2000 2004 Euro area 16 12 14 10 12 10 8 8 6 6 4 4 2 2 0 -2 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 1960 1964 1968 1972 1976 1980 1984 Note: Domestic inflations and their projections on the Global Inflation (measured by simple average). Estimation technique: OLS. Dependent variable is a deseasonalized quarter-on-quarter inflation rate. Figure 3: Measures of Global de-trended Inflation United States, Japan, Canada and United Kingdom (1965-2004) 5 First static factor Average OECD aggregate 4 3 2 1 0 -1 -2 -3 ar -0 5 M ar -0 2 M ar -9 9 M ar -9 6 M ar -9 3 M ar -9 0 M ar -8 7 M ar -8 4 M ar -8 1 M ar -7 8 M ar -7 5 M ar -7 2 M ar -6 9 M ar -6 6 M M ar -6 3 -4 Note: Three measures of detrended Global Inflation. The inflation series are detrended with Baxter and King (1999). Working Paper Series A series of research studies on regional economic issues relating to the Seventh Federal Reserve District, and on financial and economic topics. Firm-Specific Capital, Nominal Rigidities and the Business Cycle David Altig, Lawrence J. Christiano, Martin Eichenbaum and Jesper Linde WP-05-01 Do Returns to Schooling Differ by Race and Ethnicity? Lisa Barrow and Cecilia Elena Rouse WP-05-02 Derivatives and Systemic Risk: Netting, Collateral, and Closeout Robert R. Bliss and George G. Kaufman WP-05-03 Risk Overhang and Loan Portfolio Decisions Robert DeYoung, Anne Gron and Andrew Winton WP-05-04 Characterizations in a random record model with a non-identically distributed initial record Gadi Barlevy and H. N. Nagaraja WP-05-05 Price discovery in a market under stress: the U.S. Treasury market in fall 1998 Craig H. Furfine and Eli M. Remolona WP-05-06 Politics and Efficiency of Separating Capital and Ordinary Government Budgets Marco Bassetto with Thomas J. Sargent WP-05-07 Rigid Prices: Evidence from U.S. Scanner Data Jeffrey R. Campbell and Benjamin Eden WP-05-08 Entrepreneurship, Frictions, and Wealth Marco Cagetti and Mariacristina De Nardi WP-05-09 Wealth inequality: data and models Marco Cagetti and Mariacristina De Nardi WP-05-10 What Determines Bilateral Trade Flows? Marianne Baxter and Michael A. Kouparitsas WP-05-11 Intergenerational Economic Mobility in the U.S., 1940 to 2000 Daniel Aaronson and Bhashkar Mazumder WP-05-12 Differential Mortality, Uncertain Medical Expenses, and the Saving of Elderly Singles Mariacristina De Nardi, Eric French, and John Bailey Jones WP-05-13 Fixed Term Employment Contracts in an Equilibrium Search Model Fernando Alvarez and Marcelo Veracierto WP-05-14 1 Working Paper Series (continued) Causality, Causality, Causality: The View of Education Inputs and Outputs from Economics Lisa Barrow and Cecilia Elena Rouse WP-05-15 Competition in Large Markets Jeffrey R. Campbell WP-05-16 Why Do Firms Go Public? Evidence from the Banking Industry Richard J. Rosen, Scott B. Smart and Chad J. Zutter WP-05-17 Clustering of Auto Supplier Plants in the U.S.: GMM Spatial Logit for Large Samples Thomas Klier and Daniel P. McMillen WP-05-18 Why are Immigrants’ Incarceration Rates So Low? Evidence on Selective Immigration, Deterrence, and Deportation Kristin F. Butcher and Anne Morrison Piehl WP-05-19 Constructing the Chicago Fed Income Based Economic Index – Consumer Price Index: Inflation Experiences by Demographic Group: 1983-2005 Leslie McGranahan and Anna Paulson WP-05-20 Universal Access, Cost Recovery, and Payment Services Sujit Chakravorti, Jeffery W. Gunther, and Robert R. Moore WP-05-21 Supplier Switching and Outsourcing Yukako Ono and Victor Stango WP-05-22 Do Enclaves Matter in Immigrants’ Self-Employment Decision? Maude Toussaint-Comeau WP-05-23 The Changing Pattern of Wage Growth for Low Skilled Workers Eric French, Bhashkar Mazumder and Christopher Taber WP-05-24 U.S. Corporate and Bank Insolvency Regimes: An Economic Comparison and Evaluation Robert R. Bliss and George G. Kaufman WP-06-01 Redistribution, Taxes, and the Median Voter Marco Bassetto and Jess Benhabib WP-06-02 Identification of Search Models with Initial Condition Problems Gadi Barlevy and H. N. Nagaraja WP-06-03 Tax Riots Marco Bassetto and Christopher Phelan WP-06-04 The Tradeoff between Mortgage Prepayments and Tax-Deferred Retirement Savings Gene Amromin, Jennifer Huang,and Clemens Sialm WP-06-05 2 Working Paper Series (continued) Why are safeguards needed in a trade agreement? Meredith A. Crowley WP-06-06 Taxation, Entrepreneurship, and Wealth Marco Cagetti and Mariacristina De Nardi WP-06-07 A New Social Compact: How University Engagement Can Fuel Innovation Laura Melle, Larry Isaak, and Richard Mattoon WP-06-08 Mergers and Risk Craig H. Furfine and Richard J. Rosen WP-06-09 Two Flaws in Business Cycle Accounting Lawrence J. Christiano and Joshua M. Davis WP-06-10 Do Consumers Choose the Right Credit Contracts? Sumit Agarwal, Souphala Chomsisengphet, Chunlin Liu, and Nicholas S. Souleles WP-06-11 Chronicles of a Deflation Unforetold François R. Velde WP-06-12 Female Offenders Use of Social Welfare Programs Before and After Jail and Prison: Does Prison Cause Welfare Dependency? Kristin F. Butcher and Robert J. LaLonde Eat or Be Eaten: A Theory of Mergers and Firm Size Gary Gorton, Matthias Kahl, and Richard Rosen Do Bonds Span Volatility Risk in the U.S. Treasury Market? A Specification Test for Affine Term Structure Models Torben G. Andersen and Luca Benzoni WP-06-13 WP-06-14 WP-06-15 Transforming Payment Choices by Doubling Fees on the Illinois Tollway Gene Amromin, Carrie Jankowski, and Richard D. Porter WP-06-16 How Did the 2003 Dividend Tax Cut Affect Stock Prices? Gene Amromin, Paul Harrison, and Steven Sharpe WP-06-17 Will Writing and Bequest Motives: Early 20th Century Irish Evidence Leslie McGranahan WP-06-18 How Professional Forecasters View Shocks to GDP Spencer D. Krane WP-06-19 Evolving Agglomeration in the U.S. auto supplier industry Thomas Klier and Daniel P. McMillen WP-06-20 3 Working Paper Series (continued) Mortality, Mass-Layoffs, and Career Outcomes: An Analysis using Administrative Data Daniel Sullivan and Till von Wachter The Agreement on Subsidies and Countervailing Measures: Tying One’s Hand through the WTO. Meredith A. Crowley WP-06-21 WP-06-22 How Did Schooling Laws Improve Long-Term Health and Lower Mortality? Bhashkar Mazumder WP-06-23 Manufacturing Plants’ Use of Temporary Workers: An Analysis Using Census Micro Data Yukako Ono and Daniel Sullivan WP-06-24 What Can We Learn about Financial Access from U.S. Immigrants? Una Okonkwo Osili and Anna Paulson WP-06-25 Bank Imputed Interest Rates: Unbiased Estimates of Offered Rates? Evren Ors and Tara Rice WP-06-26 Welfare Implications of the Transition to High Household Debt Jeffrey R. Campbell and Zvi Hercowitz WP-06-27 Last-In First-Out Oligopoly Dynamics Jaap H. Abbring and Jeffrey R. Campbell WP-06-28 Oligopoly Dynamics with Barriers to Entry Jaap H. Abbring and Jeffrey R. Campbell WP-06-29 Risk Taking and the Quality of Informal Insurance: Gambling and Remittances in Thailand Douglas L. Miller and Anna L. Paulson WP-07-01 Fast Micro and Slow Macro: Can Aggregation Explain the Persistence of Inflation? Filippo Altissimo, Benoît Mojon, and Paolo Zaffaroni WP-07-02 Assessing a Decade of Interstate Bank Branching Christian Johnson and Tara Rice WP-07-03 Debit Card and Cash Usage: A Cross-Country Analysis Gene Amromin and Sujit Chakravorti WP-07-04 The Age of Reason: Financial Decisions Over the Lifecycle Sumit Agarwal, John C. Driscoll, Xavier Gabaix, and David Laibson WP-07-05 Information Acquisition in Financial Markets: a Correction Gadi Barlevy and Pietro Veronesi WP-07-06 Monetary Policy, Output Composition and the Great Moderation Benoît Mojon WP-07-07 4 Working Paper Series (continued) Estate Taxation, Entrepreneurship, and Wealth Marco Cagetti and Mariacristina De Nardi WP-07-08 Conflict of Interest and Certification in the U.S. IPO Market Luca Benzoni and Carola Schenone WP-07-09 The Reaction of Consumer Spending and Debt to Tax Rebates – Evidence from Consumer Credit Data Sumit Agarwal, Chunlin Liu, and Nicholas S. Souleles WP-07-10 Portfolio Choice over the Life-Cycle when the Stock and Labor Markets are Cointegrated Luca Benzoni, Pierre Collin-Dufresne, and Robert S. Goldstein WP-07-11 Nonparametric Analysis of Intergenerational Income Mobility with Application to the United States Debopam Bhattacharya and Bhashkar Mazumder WP-07-12 How the Credit Channel Works: Differentiating the Bank Lending Channel and the Balance Sheet Channel Lamont K. Black and Richard J. Rosen WP-07-13 Labor Market Transitions and Self-Employment Ellen R. Rissman WP-07-14 First-Time Home Buyers and Residential Investment Volatility Jonas D.M. Fisher and Martin Gervais WP-07-15 Establishments Dynamics and Matching Frictions in Classical Competitive Equilibrium Marcelo Veracierto WP-07-16 Technology’s Edge: The Educational Benefits of Computer-Aided Instruction Lisa Barrow, Lisa Markman, and Cecilia Elena Rouse WP-07-17 The Widow’s Offering: Inheritance, Family Structure, and the Charitable Gifts of Women Leslie McGranahan WP-07-18 Demand Volatility and the Lag between the Growth of Temporary and Permanent Employment Sainan Jin, Yukako Ono, and Qinghua Zhang WP-07-19 A Conversation with 590 Nascent Entrepreneurs Jeffrey R. Campbell and Mariacristina De Nardi WP-07-20 Cyclical Dumping and US Antidumping Protection: 1980-2001 Meredith A. Crowley WP-07-21 The Effects of Maternal Fasting During Ramadan on Birth and Adult Outcomes Douglas Almond and Bhashkar Mazumder WP-07-22 5 Working Paper Series (continued) The Consumption Response to Minimum Wage Increases Daniel Aaronson, Sumit Agarwal, and Eric French WP-07-23 The Impact of Mexican Immigrants on U.S. Wage Structure Maude Toussaint-Comeau WP-07-24 A Leverage-based Model of Speculative Bubbles Gadi Barlevy WP-08-01 Displacement, Asymmetric Information and Heterogeneous Human Capital Luojia Hu and Christopher Taber WP-08-02 BankCaR (Bank Capital-at-Risk): A credit risk model for US commercial bank charge-offs Jon Frye and Eduard Pelz WP-08-03 Bank Lending, Financing Constraints and SME Investment Santiago Carbó-Valverde, Francisco Rodríguez-Fernández, and Gregory F. Udell WP-08-04 Global Inflation Matteo Ciccarelli and Benoît Mojon WP-08-05 6