The Balassa-Samuelson effect reversed: new evidence from OECD countries

Swiss Journal of Economics and Statistics

Table 5 The Balassa-Samuelson effect in recent times

Notes: See Table 1 for the definitions of the variables. Panel DOLS estimates in (1)–(3), and in (5)–(7): all FE estimator regressions include country-specific and time-specific dummy variables as well as the first differences of each explanatory variable (1 lead/lag in (1)–(2) and (5)–(7), and 3 leads/lags in (3)). Group-mean panel FMOLS estimate proposed by Pedroni (2001) in (4). Sample period 1984–2008 in (1)–(6), and 1970–2008 in (7). Country sample (Appendix 1.1): sample (ii), in (4) Spain is excluded due to the small sample size. The productivity data stem from the PDBi in (1)–(5) and the STAN database in (6)–(7). The standard errors are reported in parentheses (robust standard errors proposed by Driscoll and Kraay (1998) in (1)–(3), and in (5)–(7)). LLC test: cointegration test following MacDonald and Ricci (2007): t statistic of Levin et al. (2002) (lag length selection by SIC; Bartlett kernel, Newey-West bandwidth). Kao test: cointegration test proposed by Kao (1999): t statistic (lag length selection by SIC; Bartlett kernel, Newey-West bandwidth). *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively