Credibility for stationarity