Working Paper Series: Special Edition of 2016 to 2018 Interns

∆Ç 6 = ∆Ñ 6 + ∙ 8 ∆Ö 6 + ∆Üá 6 : + (1 − ) ∙ 8 ∆7 6 + ∆há 6 : Ç 6É= Ñ 6É= Ö 6É= Üá 6É= 7 6É= há 6É=

(18)

where " and ℎ " are the quality indices of capital and labour. The analysis by a second level of decomposition, which disaggregates, by the three different education levels. Equation (19) incorporates primary, secondary and tertiary education: (19) ∆Ç 6 = ∆Ñ 6 + ∙ 8 ∆Ö 6 : +∑ D T ∙ â ∆3 ä,6 ã + (1 − − ∑ D T ) ∙ 8 ∆7 6 : Ç 6É= Ñ 6É= Ö 6É= TI% 3 ä,6É= TI% 7 6É= T the shares of human capital. These values are sourced from the regression analysis. Since the framework assumes that human capital share contributes % , these coefficients are scaled to ensure that they sum to 0.333: primary – 0.194, D secondary 0.094 and tertiary – 0.045. Results are reported for total average years of schooling and by education level in (See Appendix Table 3). The Augmented Dickey-Fuller test confirmed that variables are non-stationary (I1). With respect to total average years of schooling specification, like Loening (2005) and Moore (2006), the loading coefficient is highly significant and suggests a speed of adjustment towards the long- run growth path equal to about 44.0 per cent of the deviations per year. This result is higher than the 24.0 per cent and 25.0 per cent obtained for Guatemala and Barbados respectively. This may be due to the use of employed workforce rather than the economically active population, as well as the relatively greater contribution of human capital to GDP given that the economy is driven more by labour than the use of capital when compared to Barbados and Guatemala. where T," disaggregates for the 3 levels of education and 4.0 Results and Analysis 4.1 Error Correction Model

215