We investigate the behaviour of the well-known HEGY (Hylleberg, Engle, Granger and Yoo, 1990, Journal of Econometrics, vol.44, pp.215-238) regression-based seasonal unit root tests in cases where the driving shocks can display periodic non-stationary volatility and conditional heteroskedasticity.Our set up allows for periodic heteroskedasticity, non-stationary volatility and (seasonal) GARCH as special cases. We show that the limiting null distributions of the HEGY tests depend, in general, on nuisance parameters which derive from the underlying volatility process. Monte Carlo simulations show that the standard HEGY tests can be substantially over-sized in the presence of such effects. As a consequence, we propose wild bootstrap implementations of the HEGY tests.Two possible wild bootstrap re-sampling schemes are discussed, both of which are shown to deliver asymptotically pivotal inference under our general conditions on the shocks. Simulation evidence is presented which suggests that our proposed bootstrap tests perform well in practice, largely correcting the size problems seen with the standard HEGY tests even under extreme patterns of heteroskedasticity, yet not losing finite sample relative to the standard HEGY tests.Keywords: seasonal unit roots, (periodic) non-stationary volatility, conditional heteroskedasticity, wild bootstrap.JEL Codes: C12, C22. 1 roots at the seasonal frequency components of the data. This question is important because incorrect modelling of the seasonality has serious implications for the statistical validity of any subsequent procedures. Moreover, most available seasonally adjusted data are based on filtering methods which imply the application of seasonal differencing to the data. If the data do not contain seasonal unit roots then the resulting seasonally adjusted data will contain moving average unit roots, rendering standard autoregressive modelling methods invalid. In order to formally investigate this issue, in the seminal paper in this literature, Hylleberg, Engle, Granger and Yoo (1990) [HEGY] propose a seasonal generalisation of the augmented Dickey-Fuller [ADF] unit root test under the assumption of homoskedastic innovations. This procedure allows the practitioner to test for unit root behaviour at each of the zero and seasonal frequency components of the data, either separately or via a joint test.A large body of recent applied work has grown suggesting that the assumption of constant unconditional volatility is at odds with what is observed in time series data for many macroeconomic and financial series. In particular, a general decline in the unconditional volatility of the shocks driving macroeconomic series in the twenty years or so leading up to the recent financial crisis has been a relatively commonly observed phenomenon. This feature is known as the "great moderation"; see, inter alia, Kim and Nelson (1999), McConnell and Perez-Quiros (2000), Sensier and van Dijk (2004), and references therein. In the non-seasonal case it is well documented that permanent chan...