Abstract
The financial markets in London and Amsterdam were some of the first to develop. Using threshold autoregressive models, we use data on two commonly traded stocks in these cities to show that the joint behaviour of the prices is consistent with the theory of arbitrage in the presence of transportation costs. The results suggest that prices converged more quickly as the price difference between the two markets increased. We also show that the threshold estimates are consistent between assets and across time. These results provide some of the earliest evidence of nonlinear mean reversion in asset prices in geographically separate financial markets.
Notes
1 See Hansen (Citation2011) for a good review of the many ways the TAR model has been used in research.
2 See Miller (Citation2010) for a discussion of the effects of irregularly spaced data when estimating cointegrating relationships.
3 Comparing the estimated variance of the error term in the linear model to that of the EQ-TAR model suggests that the improvement in fit of the nonlinear model compared with the conventional AR model is modest at best. The improvement in fit is best in the case of EIC. This is most likely a result of the fact that there are more observations in the inner regime when compared with BOE. This would imply a greater variability in the rates of reversion that the EQ-TAR model would be able to describe.