CAPITULO II: MARCO DE REFERENCIA
2.1 Marco Histórico
2.1.7 Beneficios a Empleados
2.1.7.4 Granja
Given that both the BMZ and DGSX methodologies require modelling occupational attainment as being endogenously determined, it cannot be discarded that workers are non-randomly allocated into different occupations. To analyse the effect that self- selection into each occupational category has on the on previous findings, this section performs decompositions based on selectivity corrected wage regressions following the Lee (1983) methodology.8
Before proceeding to the decomposition results, wage regression estimations and the intuition behind them are briefly discussed.9 Following the related literature (see, e.g.
Gyourko and Tracy, 1988; Reilly, 1991), the impact of the selection term in the wage
8 Nevertheless, a couple of caveats arise with the use of this procedure. First, the literature has commonly
found that the results are quite sensitive to the specification of the MNL model in the first stage (see, e.g. Miller, 1987; Kidd and Shannon, 1994). Second, the arbitrariness of the chosen instruments implies that the identifying assumptions in the two-stage procedure are commonly only weakly satisfied.
regression can be computed by multiplying the selection coefficient, by the mean value of the selection variable. The non-statistical significance of the correction term may be interpreted as the occupational choice being largely random in the sample.
Among the main findings, it is observed that in the “Managerial and Professional Specialty” category, the selectivity bias term is positive for FG and SG-II workers and not statistically significant for other cohorts. This indicates that the unobservable characteristics that predict attachment into the highest paying occupation are positively correlated with the wage levels. The result is not surprising considering the substantial barriers to entry into high-paying occupations that FG and SG-II workers face, suggesting that only those who perform best in the labour market are able to gain entry. At the other end of the distribution, the negative selection term observed for whites in the “Operators, Fabricators and Labourers” category portrays the opposite picture. This denotes that on average, the wages of whites employed in the lowest-paying occupation are lower than those obtained by an average worker drawn at random from the population. However, the high values of the estimated selectivity term coefficients suggest that the results require a cautions interpretation.
In the forthcoming decomposition analysis, the correction term is not considered as constituting part of the explained or unexplained components and instead is examined separately. Furthermore, the four previously defined terms of the BMZ and DGSX methodologies are grouped and demarcated as representing the wage offer gap or unconditional wage differential. This is equivalent to setting the selectivity effects to zero in each equation, where the wage offer gap is taken to mean the wage a worker randomly drawn from the population would receive if selected into the occupational category in question (Gyourko and Tracy, 1988).
Table 2.13 BMZ decomposition:
Blacks and Mexican-Americans with Lee (1983) correction
FG SG-II SG-I TG Total log wage differential .361*** .120 .003 .014
(.120) (.100) (.110) (.057)
Explained: Differences in average characteristics
Within .189*** .121*** .046*** .047***
(.014) (.006) (.005) (.005)
Between .122*** .045*** .026*** .033***
(.016) (.007) (.004) (.004)
Unexplained: Differences in coefficients
Within .162 .020 .148 .158*
(.130) (.115) (.124) (.081)
Between .062** -.058*** -.065*** -.069***
(.023) (.014) (.012) (.014)
Wage offer gap .537*** .129 .156 .170***
(.134) (.116) (.125) (.083)
Selection term differential -.175 -.009 -.153 -.156***
(.115) (.098) (.109) (.056)
***p<.1, **p<.05, *p<.01
Note: the OLS coefficients of blacks are taken as the non-discriminatory vector.
Source: Author’s elaboration based on the CPS March Supplement 1994-2012. Standard errors are in parentheses. Decomposition based on OLS regressions presented in Table A.2.4.
Table 2.14 BMZ decomposition:
Whites and Mexican-Americans with Lee (1983) correction
FG SG-II SG-I TG Total log wage differential .599*** .358*** .241** .252***
(.123) (.106) (.116) (.067)
Explained: Differences in average characteristics
Within .255*** .151*** .077*** .073***
(.006) (.002) (.001) (.001)
Between .053*** .060*** .039*** .040***
(.008) (.002) (.002) (.002)
Unexplained: Differences in coefficients
Within .396*** .201* 325*** .340***
(.123) (.107) (.117) (.069)
Between .141*** .026*** .023*** .026***
(.010) (.001) (.002) (.001)
Wage offer gap .846*** .439*** .466*** .480***
(.124) (.107) (.117) (.069)
Selection term differential -.247** -.080 -.224* -.227***
(.121) (.106) (.116) (.067)
***p<.1, **p<.05, *p<.01
Note: the OLS coefficients of whites are taken as the non-discriminatory vector.
Source: Author’s elaboration based on the CPS March Supplement 1994-2012. Standard errors are in parentheses. Decomposition based on OLS regressions presented in Table A.2.4.
Tables 2.13 and 2.14 present the results of the BMZ decomposition corrected for self- selection between Mexican-Americans and blacks and whites, respectively. It can be seen
that the selection term differential is always negative, implying that the wage offer gap is larger than the total wage gap and that wage differentials between natives and Mexican- Americans are larger after correcting for self-selection. Relative to blacks, the selection term differential is only significant for the case of TG workers. Focusing on the BU component, it is observed that for FG immigrants this represents 11.5% of the wage offer gap, while for second and third generation Mexican-Americans the contribution of this term to the unconditional wage differentials is negative. With respect to whites, the selection term differential is significant for all cohorts except SG-II workers. The magnitude of the BU term in accounting for the wage offer gap ranges from 16.7% for FG immigrants to 4.9% for SG-I workers.
Finally, Tables 2.15 and 2.16 present the results of the DGSX decomposition between Mexican-Americans and blacks and whites, respectively. Given that the MNL model estimated in the first stage and the selectivity corrected wage regressions are the same as the ones employed in the BMZ methodology, the unconditional wage differentials do not change. With respect to blacks, the biggest change is observed for FG immigrants, where the contribution of the BU term to the observed wage differential is now larger. For subsequent generations the effect of this term is small and not statistically different from zero. With respect to whites in the case of FG workers the BU component accounts for 17.5% of the wage offer gap, yet this term is not significant for any of the three generations of Mexican-Americans.
Table 2.15 DGSX decomposition:
Blacks and Mexican-Americans with Lee (1983) correction
FG SG-II SG-I TG Total log wage differential with respect to whites .361*** .120 .003 .014
(.093) (.076) (.065) (.070)
Explained: Differences in average characteristics
Within -.043** .036*** -.007 -.023**
(.021) (.011) (.010) (.011)
Between .122*** .045** .026** .033***
(.039) (.018) (.011) (.011)
Unexplained: Differences in coefficients
Within .294*** .066 .157 .154**
(.113) (.103) (.114) (.073)
Between -.163* -.019 -.019 .005
.084 (.065) (.054) (.047)
Wage offer gap .537*** .129 .156 .170*
(.148) (.124) (.127) (.088)
Selection term differential -.175* -.009 -.153 -.156***
(.115) (.098) (.109) (.054)
***p<.1, **p<.05, *p<.01
Note: the OLS coefficients of whites are taken as the non-discriminatory vector.
Source: Author’s elaboration based on the CPS March Supplement 1994-2012. Standard errors are in parentheses. Decomposition based on OLS regressions presented in Table A.2.4.
Table 2.16 DGSX decomposition:
Whites and Mexican-Americans with Lee (1983) correction
FG SG-II SG-I TG Total log wage differential with respect to whites .599*** .358*** .241*** .252***
(.065) (.025) (.019) (.016)
Explained: Differences in average characteristics
Within .211*** .197*** .132*** .149***
(.009) (.003) (.003) (.003)
Between .053*** .060*** .039*** .040***
(.020) (.007) (.004) (.005)
Unexplained: Differences in coefficients
Within .432*** .216** .300*** .300***
(.096) (.094) (.104) (.059)
Between .148 -.034 -.006 -.010
(.096) (.053) (.054) (.035)
Wage offer gap .846*** .439*** .466*** .480**
(.138) (.109) (.118) (.069)
Selection term differential -.247** -.080 -.224* -.227***
(.121) (.106) (.116) (.067)
***p<.1, **p<.05, *p<.01
Note: the OLS coefficients of whites are taken as the non-discriminatory vector.
Source: Author’s elaboration based on the CPS March Supplement 1994-2012. Standard errors are in parentheses. Decomposition based on OLS regression results presented in Table A.2.4.