Financial time series are typified by nonstationarity. This is especially true of the exchange rate, whether nominal and real. The nonstationarity of the real exchange rate has also an added significance: it means persisting deviations from the value reflecting purchasing power parity.

The cause of these deviations may lie in, among other things, an expectation of future deviations, errors in forecasting inflation differentials for two countries or inaccurate prediction of movement of the exchange rates.

There is vast theoretical and empirical literature on purchasing power parity, its rate of decay, conditions for occurrence and causes of deviation from the trend. One of the methods to test purchasing power parity involves cointegration of the time series in question. Long-term equilibrium relations for nonstationary variables can be expressed by models with cointegration. Those require that the integration of the variables in the model be tested.

In the case of the equation describing the $/PLN exchange rate, the paper uses quarterly data. Deterministic and stochastic seasonality was tested by applying Hyllenberg, Engle, Granger and Yoo tests, both for the real and nominal exchange rate series of observations.

Due to the seasonality of the data, seasonal integration analysis was conducted, based on Dickey's, Hasza and Fuller tests. Integration tests are carried out with the help of ADF and Dickey-Pantuli tests.

The results of the HEGY test suggest that real exchange rate is not characterised by stochastic seasonality. ADF tests do not justify rejecting the null hypothesis concerning the possible occurrence of a single unit root. This means that the real exchange rate is nonstationary, while its first differences are stationary.

With respect to the nominal exchange rate, non-seasonal integration was tested with the ADF test, the Dickey - Pantuli test and, to verify the findings, the Phillips-Perron test. There are no grounds to reject the null hypothesis of nonstationarity of the nominal exchange rate; its first differences are stationary. Moreover, based on the HEGY test we draw a conclusion as to the lack of stochastic seasonality, which could be removed by calculating first seasonal differences.

To recapitulate, the real exchange rate reflects the characteristics of the nominal exchange rate, i.e. its nonstationarity and order of integration. In addition, the nominal exchange rate and price indices for both countries are not cointegrated with respect to the parameters reflecting purchasing power parity. This confirms the view expressed in the literature that parity could not be observed in the period under review.

Besides the empirical findings, the paper includes a description of all the tests applied.