HARD CURRENCIES FOR HARD TIMES. TERROR ATTACKS AND THE CHOICE OF MONETARY ANCHORS

Abstract

As terror’s victims increase, hard currency commitments gain effectiveness in reducing inflation, and central bank independence loses its effectiveness, because terror reduces transparency and the number of veto players in domestic politics. PCSE (Panel‐Corrected Standard Error) estimations of inflation are run on pooled cross‐section time‐series sample of 87 countries from 1975–2005. When the trend level rises to 100 victims annually a currency board reduces inflation by up to 7.5%, and an independent bank raises inflation by up to 8%. When victims exceptionally exceed the trend by 100, a currency board reduces inflation by 2.5%, and an independent bank raises it by 2%.

Keywords:

• Terror
• Currency boards
• Central banks
• Transparency

JEL Codes:

• E31
• E58

ACKNOWLEDGMENTS

The author acknowledges the helpful suggestions of William Bernhard, Alex Cukierman, John Freeman, David Leblang, Leo Leiderman, Gerald Schneider, an anonymous referee, participants in a workshop on ‘Political Events, Financial Markets, and Trade’ held in Konstanz, Germany in January 2007, and participants in the 2007 Israel Economic Association Annual Meeting. The author also acknowledges the support of the Kurt Lion Foundation in organizing the workshop and a research grant from the Leonard Davis Institute for International Relations. The author is grateful to Amir Kazir and the Bank of Israel for granting him access to the Bank’s library. Nizan Feldman and Tsvika Machlof provided excellent research assistance.

Notes

¹ Following Bernhard et al. (2002) ‘credibility’ is understood in this study to involve ‘persuading private agents that the monetary policymaker will not exploit the flexibility inherent in a fiat standard to achieve short‐run output gains.’

² See Blomberg and Hess (2002); Blomberg et al. (2004); Burgoon (2006); Ferrero (2006); Frey and Luechinger (2004); Garfinkel (2004); Sandler (2000); Schelling (1991); Siqueira and Sandler (2006); and Wilkinson (2001).

³ See Berrebi and Klor (2006); Enders and Su (2007); Hess and Orphanides (1995, 2001); and Lichbach (1987).

⁴ See Arunatilake et al. (2001); Brück and Wickström (2004); and Walkenhorst and Dihel (2002).

⁵ See for example Blomberg et al. (2004); Bloom (2005); Brauer et al. (2004); Buckelew (1984); Garfinkel (2004); Gupta et al. (2004); and Sandler and Enders (2004).

⁶ See Abadie and Gardeazabal (2003); Acemoglu and Robinson (2001); Addison (2003); Alesina and Perotti (1996); Alesina et al. (1996); Berrebi and Klor (2005); Eckstein and Tsiddon (2004); Rodrik (1999); and Stewart and FitzGerald (2001).

⁷ See Arora and Bayoumi (1993); Arunatilake et al. (2001); Bayoumi et al. (1993); Blomberg et al. (2003); Knight et al. (1996); and Shieh et al. (2002).

⁸ Barbieri and Schneider (1999); Brück and Wickström (2004); Nitsch and Schumacher (2004); and Reuveny (1999–2000).

⁹ As in Broz (2002) the term ‘effectiveness’ in this context relates to the magnitude of decline in the rate of inflation that is associated with adopting a certain monetary anchor.

¹⁰ According to Broz (2002) ‘transparency is the ease with which the public can monitor the government with respect to its commitments’.

¹¹ In Cukierman’s original study a behavioral measure of central bank independence was preferred over the legal measure in the case of developing countries. However, such a measure is endogenous, and the control for transparency and the rule of law in this study improves the legal measure’s relevance, even for developing countries. I thank Alex Cukierman for his helpful comments on this matter.

¹² This formula is based on Rae’s (1971) fragmentation index.

¹³ Ethnic and religious composition data is based on CIA Factbook, while linguistic composition data, which relates to mother tongue languages, is based on Gordon (2005). For lack of better data, the same LANGUAGE and FRAGMENT values are used for each country during the entire sample period. However, in the majority of cases ethnic, religious and linguistic stability seems anyway plausible.

¹⁴ See: http://www.cidcm.umd.edu/inscr/polity/index.htm. Democracy is also used as an instrument in two‐step estimation in the next section.

¹⁵ See: http://www.transparency.org/index.php/policy_research/surveys_indices/cpi. Because this index is available only since 1995, and for many of the sample countries only later, CORRUPTION is calculated for each country as a temporally‐constant, multi‐annual average.

¹⁶ See: http://www.freedomhouse.org.

¹⁷ Thus, the formula used is victims/(22+victims) and TERROR, which ranges between 0 and 1, returns the value 0.5 for observations with 22 victims.

¹⁸ Although CORRUPTION’s values are temporally fixed, BANK’s are not.

¹⁹ TERROR’s values are trended with the Hodrick‐Prescott filter (with a factor of 100). Using a polynomial fit in time with four trend terms instead of the Hodrick‐Prescott filter did not yield meaningfully different results.