Like Naylor and Santoni (2003), Zhao (1995), and Eckel and Egger (2009), in the present paper, intra-industry RFDI and the presence of unions in the labor market is accommodated.
The parameter's restriction on trade costs is given by t [less than or equal to] 0.310 because, at this level, any union wage combination in the two-stage sub-game "unions' wage determination-firms' quantity choices" is consistent with every configuration of the firms' strategies involving international activities (IIT, RFDI, and asymmetric regimes).
It follows that the specification of the firms' profit functions in the presence of RFDI is exactly as indicated in Equations 3 and 4.
Comparing [w.sub.IIT] and [w.sub.RFDI], and production outcomes (and thus employment levels), it is immediately clear that, in the case of international production, both achieve higher values, and therefore, unions in equilibrium have higher utility levels in the RFDI regime than in IIT.
Firms' payoffs in the RFDI regime depend on wage levels set by unions and the amount of sunk costs.
Table 4a shows the normalised coefficients of the explanatory variables on RFDI from the Johansen cointegration test.
[DELTA] RFDI = [[beta].sub.0] + [[beta].sub.1] [DELTA] TARIFF + [[beta].sub.2] [DELTA] PCGNP + [[beta].sub.3] [DELTA] CCFA + [[beta].sub.4] [DM.sub.96] + [[beta].sub.5] [EC.sub.(-1)] (13)
Table 2 PP Unit Root Test for Stationarity (1) (2) (3) (4) (5) Without With Variable Level/First Difference Trend Trend Conclusion RFDI ADF(1) Level -0.8910 -2.4891 First -7.4543* -7.3283* I(1) Difference PCGNP ADF(1) Level -0.8572 -2.7958 First -6.6249 * -0.9131 * I(1) Difference CCFA ADF(1) Level -0.646 -0.8452 First -4.8893 * -5.3683 * I(1) Difference TARIFF ADF(1) Level -0.6447 -2.706 First -5.3404 * -5.2284 * I(1) Difference CRGDP ADF(1) Level -0.5795 -3.7244 ** First -8.2323 * -8.1925 * 1(0) Difference REXPTC ADF(1) Level -0.2318 -2.1561 First -4.5147 * -4.4851 * I(1) Difference DM ADF(1) Level -1.6282 -2.5319 First -5.5527 * -5.4559 * I(1) Difference Note: PP tests were performed using Eviews 3.0.