October 11, 2010

Related with the topics discussed, I was able to look for a paper (as the title suggests) testing the PPP through the VAR model. Even though it has some degree of complication compared to some formula lectured, it is still a good reference on how PPP is being tested empirically.

Paper: Miyakoshi, Tatsuyoshi (2004) A testing of the purchasing power parity hypothesis using a vector autoregressive model, Empirical Economics, 29(3), 541-552.

The objectives of the paper are the following:

1. empirically examine the hypothesis of PPP relation and the hypothesis of uncovered interest rates parity (UIRP is discussed in Chapter 3 of Copeland’s book) relation between Japan and the USA by using cointegration approach.

2. use an error correction model (ECM is introduced in Chapter 1 of Enders’s book) with the long-run relations and evaluate the ECM in a one-step prediction (whoooh! quite lost, huh! Sorry!)

According to the paper, in the long-run, the exchange rate is determined by the symmetry restriction on the PPP relationship is stationary/steady. Or, the exchange rate is represented by the UIP is stationary. Traditional exchange rate theories are still valid in the long-run concept by cointegration.

The paper also assessed the ECM with such long-run relations in a one-step prediction. The fluctuations of the exchange rate were well predicted in the sense that the prediction errors fall in the prescribed range of the confidence bands. Moreover, the ECM dominated the benchmark prediction. That’s why the paper concluded that the one-step prediction based on this ECM with such long-run relations is useful.

狄 強

9704630