The Elusive Predictive Ability of Global Inflation

• Author: Pablo M. Pincheira,Michael Pedersen,Carlos A. Medel
• DOI: http://doi.org/10.1111/infi.12087
• Published date: 01 June 2016
• Date: 01 June 2016

The Elusive Predictive Ability of

Global Inflation

Carlos A. Medel

y

, Michael Pedersen

z

and Pablo M. Pincheira

§

y

Financial Stability, Central Bank of Chile,

z

Economic Research, Central

Bank of Chile, and

§

School of Business, Adolfo Ib

a~

nez University, Chile.

Abstract

In this paper we analyse the utility of international measures of

inflation in predicting local ones. To that end, we consider a set of 31

OECD economies for which monthly inflation data are available. Three

main conclusions emerge. First, there is an important share of coun-

tries for which relatively robust evidence of predictability is found for

both core and headline inflation. Second, the share of countries for

which there is evidence of robust predictability is about the same for

core and headline inflation, although gains in root-mean-squared

prediction error are higher for headline inflation. Third, while the

evidence indicates that an international inflation factor may be a

useful predictor for several countries, it also indicates that, for many

countries as well, predictability is either questionable, undetectable,

non-robust or simply non-existent.

We thank the comments and suggestions of Claud io Raddatz and two anonymous referees . We also

thank colleagues at Economi c Research at the Central Bank of Ch ile for helpful discussi ons and the

participants of two conferences: the SSEM Euroconference 201 4 in Budapest, Hungar y, and the

Twenty-First International Conference on Forecasting Finan cial Markets in Marseille, France. We also

thank Consuelo Edwards for editin g services. The vi ews expressed in this paper do not necess arily

represent those of the Central Ban k of Chile or its authorities.

International Finance 19:2, 2016: pp. 120–146

DOI: 10.1111/infi.12087

© 2016 John Wiley & Sons Ltd

I. Introduction

During the last decade one direction of research has focused on how international

inflation may help forecast domestic inflation. Inthe present paper we contribute to this

line of research using monthly data from 31 OECD countries in the quest for links

between global and future local inflation. In particular, we ask whether forecasts of the

consumer price index (CPI) inflation rate can be improved with information on

international price movements. We use a broad class of benchmarks as a simple way

to deal with model uncertainty: in our baseline econometricapproach we use a family of

times series models which has been shown to be highly accurate when predicting

inflation rates. We also consider the benchmarks used by Ciccarelli and Mojon (2010)

(henceforth CM), as well as an unemployment Phillips curve (henceforth PC).

1

Three main conclusions e merge. First, there is a n important share of countries for

which relatively robust evid ence of predictabilit y is found for both core and headli ne

inflation. Second, the share of countr ies for which there is evide nce of robust

predictability i s about the same for core and headline i nflation (40%), thou gh gains

in root-mean-squa red prediction error ( RMSPE) are higher for headline inflation.

Third, while the evidence indicates that an international inflation factor (IIF) may

be a useful predic tor for several countr ies, it also indicates that, for many countries

as well, this predictabili ty is either questi onable, undetec table, non-robust or simply

non-existent. This las t point is important because t he influential paper by CM claims

that inflation is a global phenome non, at least for industri alized economies. Ou r

findings only pa rtially support th at claim. For instance , Sweden and the Netherl ands

are two industrialized countr ies for which no (or ver y little) eviden ce of predictabil-

ity was found for the hea dline IIF (HIIF).

Our results for the core IIF (CIIF) are to some extent unexpected, because one

might think that a n important drivin g force of the results ob tained for headl ine

inflation comes from commodit y price fluctuations, which may directly affect the

food and energy compone nts of the CPI. Our predict ability finding s for the CIIF

provide evide nce of a deep predic tive linkage b etween intern ational and l ocal

inflation, at least for some countries.

We also explore the stability of our results by dividing the sampl e into two

subsets, before and after June 2007. Our analysis indi cates that in the most recent

period the HIIF had a larger marginal predic tive contribution amongst a l arger share

of countries, whereas the CII F had only a large r marginal predic tive contribution

amongst a relatively si milar share of countries .

1

The use of multiple bench marks limits the possi ble critique of the hyp othesis that the forecasti ng

performance of a relat ively poor benchmark can be improved by addin g extra variables, even in out-

of-samp le exercis es. As Stoc k and Watson (200 9) menti on, ‘in some cases, apparently good

performance of a predictor for a par ticular inflation se ries over a particular peri od can be the result

of a large denominator, not a small numerator’.

The Elusive Predictive Ability of Global Inflation 121

© 2016 John Wiley & Sons Ltd