| dc.description.abstract | he first chapter provides the first consistent estimates of intergenerational earnings mobility in Chile which is, based on administrative records that link a child’s and
their parent’s earnings from the formal private labour sector. We estimate that the
intergenerational earnings elasticity is between 0.288 and 0.323, whereas the rankrank slope is between 0.254 and 0.275. We find significant non-linearities in the relationship between parents’ and their children’s earnings, where the intergenerational
mobility is high in the bottom 80% of the parents’ distribution but with extremely
high intergenerational persistence in the upper part of the earnings distribution. In
addition, we find remarkable heterogeneity in intergenerational mobility at the regional level, where Antofagasta, a mining region, is the most upwardly-mobile region.
Finally, we estimate significant differences across municipalities in the Metropolitan
Region, where our estimates suggest that the place of residence makes a significant
difference in intergenerational mobility for children of upper-class families, while it
is less relatively important for children of lower- and middle-class families.
The second chapter estimates intergenerational mobility for Chilean males using
an administrative data set that links parents’ and their sons’ earnings. We find that
intergenerational earnings elasticity (IGE) and rank-rank correlation are 0.282 and
0.239, respectively. Our IGE estimate is about half of the previous estimates for Chile
that have used the Two-Sample Two-Stage Least Squares (TSTSLS) method, where
parents’ earnings must be imputed. We simulate a TSTSLS setting with our data
and recover these past estimates. Then, we show that TSTSLS estimates have two
sources of bias: a projection bias and a variance bias, which are both consequences
of imputing parents’ earnings via Mincer regressions. To improve IGE estimation
under TSTSLS, we provide two steps to reduce these biases: parent fixed effects to
improve the Mincer equation predictions and stochastic imputation to increase the
variance of predicted wages. We show that if both of these corrections are used, we
can recover our original estimates. The results are closer to our measure of IGE, but
only when we have a precise first stage, which requires information beyond what is
usually found in household surveys. We show that rank-rank correlations estimated
using the TSTSLS method are much closer to estimates that comes directly from the
administrative data. Our results suggest that administrative data should be used
to measure intergenerational mobility, however, when linked earnings data between
parents and their children is not available, researchers should focus on rank-rank
correlations for this purpose.
The third chapter investigates two measures of teacher’s quality and their impact
on tertiary education attendance utilizing a novel national administrative data set.
The two alternative instruments that measure teachers’ effectiveness for the same
sample of Chilean teachers and students are: the National Teachers’ Evaluation test
(Evaluaci´on Docente, ED) and the traditional value-added results approach (VA)
used in the literature. We find that the correlation between the measurements of
teachers’ quality from the ED and the VA approach appears to be null, which could
be due to differences in the dimensions of teacher quality measured (as suggested
by previous studies). Our analysis also reveals that both measures, ED and VA,
positively affect the probability of tertiary education attendance, corroborating that
both measures are complementary in measuring teacher quality. Additionally, we
show that two (portfolio and external references) out of four parts of the ED are the
best predictors of graduate students’ tertiary education attendance. These results
suggest that the best approach for evaluating teachers should consider a combination
of the VA and ED, with improved instruments measured in the ED in terms of cost
and teachers’ time spent in the evaluation. | |
