1 Introduction
2 Empirical methodology
2.1 Modelling inefficiency
2.2 Modelling determinants of inefficiency
2.3 The Empirical Model
3 Data
Variable | Min | Mean | Median | SD | Max | |
---|---|---|---|---|---|---|
y1 | Average exam score | 0.242 | 0.664 | 0.673 | 0.091 | 1 |
y2 | Total presence in class | 0.405 | 0.761 | 0.777 | 0.099 | 1 |
y3 | Participation rate in exam | 0.010 | 0.544 | 0.560 | 0.213 | 1 |
x | Cost per learner per session (euros) | 0.635 | 9.097 | 7.648 | 8.365 | 110.468 |
z1 | Gender balance | 0.000 | 0.491 | 0.480 | 0.340 | 1 |
z2 | Age of the learner | 19.875 | 30.822 | 30.345 | 4.634 | 47.5 |
z3 | Low educated (%) | 0 | 0.101 | 0.088 | 0.089 | 0.667 |
z4 | Sessions per program | 0.693 | 3.832 | 3.761 | 0.990 | 7.366 |
z5 | Learners per program | 1.386 | 4.541 | 4.543 | 0.582 | 6.045 |
z6 | Age of the teacher | 29.365 | 46.448 | 46.648 | 6.124 | 72.795 |
z7 | Teacher age composition | 0 | 8.816 | 8.958 | 2.960 | 20.506 |
z8 | Teacher hours composition | 0 | 3.941 | 3.976 | 0.582 | 5.425 |
Number of programs | 1200 |
Type | Count |
---|---|
Design | 230 |
Food | 160 |
Management | 210 |
Metal and wood | 180 |
Environment | 210 |
Technology | 210 |
Total | 1200 |
4 Results
4.1 Education production technology
Full sample | Restricted sample | |
---|---|---|
Variable | Coefficient (SE) | Coefficient (SE) |
(1) | (2) | |
(Intercept) | 0.180*** (0.018) | 0.138*** (0.037) |
log(y2/y1) | 0.685*** (0.039) | 0.769*** (0.049) |
log(y3/y1) | − 0.039*** (0.014) | − 0.054*** (0.020) |
log(x) | 0.023** (0.012) | 0.025* (0.014) |
t | − 0.000 (0.006) | 0.010 (0.012) |
\(\{ \log (y2/y1)\}^{2}\) | 0.577*** (0.029) | 0.574*** (0.037) |
\(\{ \log (y2/y1)\}^{2}\) | − 0.004 (0.003) | − 0.000 (0.005) |
\(\{ \log (x)\}^{2}\) | − 0.006** (0.003) | − 0.005*** (0.003) |
\(t^{2}\) | − 0.000 (0.001) | − 0.001 (0.001) |
log(y2/y1)*log(y3/y1) | − 0.032** (0.013) | − 0.023 (0.021) |
log(y2/y1)*log(x) | − 0.119*** (0.016) | − 0.128*** (0.019) |
log(y2/y1)*t | − 0.002 (0.004) | − 0.009*** (0.005) |
log(y3/y1)*log(x) | 0.011** (0.005) | 0.013** (0.006) |
log(y3/y1)*t | 0.003*** (0.001) | 0.004*** (0.002) |
log(x)*t | 0.001 (0.001) | 0.001 (0.001) |
With respect to | 1st Q. | Median | Mean | 3rd Q. |
---|---|---|---|---|
x | − 0.0298 | − 0.0173 | − 0.0171 | − 0.0037 |
y1 | 0.2736 | 0.3974 | 0.3953 | 0.5114 |
y2 | 0.4875 | 0.6075 | 0.6087 | 0.7371 |
y3 | − 0.0133 | − 0.0036 | − 0.0040 | 0.0056 |
4.2 Efficiency
4.3 Determinants of efficiency differentials
Full sample | Restricted sample | |
---|---|---|
Variable | Coefficient (SE) | Coefficient (SE) |
PANEL A. Persistent inefficiency component, LHS variable is \(\log \sigma_{{u_{0i} }}^{2}\) | ||
(Intercept) | − 10.313*** (3.646) | − 10.044*** (3.504) |
Type1 Design | 6.200* (3.591) | 5.970* (3.414) |
Type2 Food | 1.479 (3.900) | 0.127 (4.834) |
Type3 Management | 6.842* (3.616) | 6.643* (3.465) |
Type4 Environment | 5.348 (3.589) | 5.103 (3.420) |
Type5 Technology | 5.525 (3.603) | 5.304 (3.450) |
PANEL B. Transient inefficiency component, LHS variable is \(\log \sigma_{{u_{it} }}^{2}\) | ||
(Intercept) | − 0.090 (0.825) | 0.464 (1.029) |
z1 (Gender balance) | − 0.333** (0.170) | − 0.421** (0.200) |
z2 (Age of the learner) | − 0.038** (0.016) | − 0.056*** (0.020) |
z3 (% low educated) | − 5.276 (3.825) | − 6.463 (4.900) |
z4 (Sessions per program) | − 0.774*** (0.071) | − 0.881*** (0.097) |
z5 (Learners per program) | − 0.155* (0.094) | − 0.143 (0.113) |
z6 (Age of the teacher) | − 0.002 (0.009) | 0.007 (0.011) |
z7 (Teacher age composition) | 0.045** (0.018) | 0.064*** (0.021) |
z8 (Teacher hours composition) | 0.062 (0.100) | 0.050 (0.119) |
t | − 0.044* (0.026) | − 0.100*** (0.034) |
z2*z3 | 0.132 (0.119) | 0.191 (0.152) |
Variable | Min | Mean | Median | Max | |
---|---|---|---|---|---|
z1 | Gender balance | − 0.166 | − 0.082 | − 0.080 | 0 |
z2 | Age of the learner | − 0.895 | − 0.378 | − 0.388 | 1.080 |
z3 | Low educated (%) | − 0.748 | − 0.065 | − 0.049 | 0.132 |
z4 | Sessions per program | − 0.387 | − 0.387 | − 0.387 | − 0.387 |
z5 | Learners per program | − 0.077 | − 0.077 | − 0.077 | − 0.077 |
z6 | Age of the teacher | − 0.061 | − 0.039 | − 0.039 | − 0.024 |
z7 | Teacher age composition | 0 | 0.200 | 0.203 | 0.464 |
z8 | Teacher hours composition | 0 | 0.030 | 0.030 | 0.031 |
4.4 Input requirement function
Variable | Coefficient (SE) |
---|---|
PANEL A. Persistent inefficiency component, LHS variable is \(\log \sigma_{{u_{0i} }}^{2}\) | |
(Intercept) | − 2.676* (1.542) |
Type1_Design | 2.859** (1.422) |
Type2_Food | 0.977 (1.408) |
Type3_Management | 3.355** (1.483) |
Type5_Environment | 0.810 (1.465) |
Type6_Technology | 2.142 (1.381) |
PANEL B. Transient inefficiency component, LHS variable is \(\log \sigma_{{u_{it} }}^{2}\) | |
(Intercept) | − 7.774*** (2.193) |
z1 (Gender balance) | − 0.779 (0.530) |
z2 (Age of the learner) | − 0.073 (0.044) |
z3 (% low educated) | 12.950 (9.580) |
z4 (Sessions per program) | 0.771*** (0.145) |
z5 (Learners per program) | 1.083*** (0.295) |
z6 (Age of the teacher) | − 0.038 (0.025) |
z7 (Teacher age composition) | 0.019 (0.048) |
z8 (Teacher hours composition) | 0.018*** (0.004) |
t | 0.028 (0.227) |
z2*z3 | − 0.519 (0.358) |