Central bank independence and ageing

Abstract

We contrast the influence of demography and Central Bank Independence (CBI) on inflation. The recent demographic trends in developed countries are shown to weight more on inflation than CBI, while the contrary stands for the period from 1960 to 1979.

Acknowledgements

For remarks and suggestions on a previous version, we would like to thank Alain Ayong Le Kama, King Banaian, Vincent Bouvatier, Alex Cukierman, Jakob de Haan, Andreas Freytag, Carsten Hefeker, Jeroen Klomp, Stéphane Lambrecht, Donato Masciandaro, Valérie Mignon, Marc Quintyn, Pierre Siklos, Stéphane Vigeant and participants to Bocconi University's Paolo Baffi Center Conference on Central Bank Independence, Rennes 24th Symposium on Money and Banking and ICMSE conference, and to the Banque de France and Paris School of Economics joint seminar. Any remaining errors are ours.

Notes

¹ Shiller (1997) even shows that intergenerational differences in inflation aversion are more important than international ones.

² Data for France available from the INSEE website (http://www.insee.fr/fr/ppp/ir/accueil.asp?page=transm04/doc/documentation.htm); data for the US from the 2004 Survey of Consumer Finances. See Bucks et al. (2006).

³ Motivations for the suppression of the 0–14 year old share are provided by Lindh and Malmberg (1998).

⁴ Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Spain, Portugal, Sweden, Switzerland, UK and the US.

⁵ Results are available from the authors on request.

⁶ An observation on nonstationarity concerns. The current panel unit root tests (Maddala and Wu, 1999; Levin et al., 2002; Im et al., 2003) often disagree on the persistence of the data, so that it would not be clear whether the variables in Equation 1(1) and 2(2) are indeed integrated or co-integrated, whatever Pedroni's (2001) panel cointegration test would tell. Besides, the sub-periods we are considering are too short (5–14 years) for imposing any meaningful restrictions. We can therefore perform estimations in levels, since we are de facto interested in the short-term relationships between inflation and right-hand side variables.

⁷ Results are available from the authors upon request.

⁸ Moreover, a relatively small number of lags for the instruments minimizes the potential small sample bias that may arise when too many over-identifying restrictions are imposed.

⁹ Hansen overidentification test did not reject our sets of instruments. Complete results available upon request.

¹⁰ This phenomenon probably contributes to explain their subsequent fear of inflation.

¹¹ Intra-group correlation means that the error terms are correlated within groups, but not across groups.