Co-integration, error correction and the Fisher effect: A clarification

The presence of co-integration between interest rates and inflation implies the existence of an error-correction model and the possibility of two sources of causation. Causality testing which does not account for feedback through the error-correction mechanism as well as through the lagged changes in the variables can produce misleading results. Reinterpreting Atkins' error-correction model and causality tests in this framework points to a feedback relationship between inflation and post-tax nominal interest rates. These findings are consistent with previously published results but are in contrast to Atkins' conclusion of one-way causality from inflation to interest rates.