The ERM breakdown: a spatial econometric approach

Abstract

This study proposes a spatial autoregressive model to investigate whether pure contagion effects occurred during the European exchange market turmoil in 1992. It defines pure contagion as spatial autocorrelation between countries in excess to that which can be explained by economic fundamentals. By doing so, it is possible to predict the effectiveness of rescue packages and foreign supports for each country of the dataset.

Acknowledgements

The authors would like to thank André Cartapanis and Anne Péguin-Feissolle for their helpful comments.

Notes

¹ Calculated from the IMF's ‘Direction of Trade Statistics’.

² See Kaminsky et al . (1998).

³ Defined as in Glick and Rose (1999) as the percentage change in the real exchange rate index between the average in the 12 months before the crisis and the average in the previous 3 years. A negative value indicates, here, that the real exchange rate is overvalued.

⁴ All macroeconomic data are extracted from IMF's International Financial Statistics Database. Since the first devaluation occurred at the end of August 1992, annual data from 1992 are used.

⁵ A GWR model has been used for testing the spatial stationarity of the relationship under study. As its results support the OLS model, the GWR results are not presented here, but are available on request.

⁶ This statistical result supports, to some extent, the view that the SAR approach is more relevant than the SEM one.

⁷ The decision rule states that if the p-value of the LMLAG (resp LMERR) test is less than the p-value of the LMERR (resp. LMLAG) test and RLMLAG (resp. RLMERR) is significant but LMERROR (resp. RLMLAG) not, then the appropriate model is the spatial autoregressive model (resp. spatial error model).