The name is absent



section. The household fixed effects regression, by controlling for household level unobservable
factors, measures the impact of fostering on school enrollment, conditional on the household having
optimized its decision of which child to send. This contrasts with the child fixed effects regression,
which goes further and conditions on a given child’s unobserved type.

This is the first time these strategies have been used to address the endogeneity of child fostering
and the unobserved factors influencing fostering and school enrollment. If fostering is correlated
with household characteristics such as wealth or network quality, which are also important deter-
minants of school enrollment, then failing to control for these factors can yield biased estimates
of the fostering impact on school enrollment. The household fixed effects regression compares the
school enrollment for a foster child and the host siblings, within the same household, before and
after the fostering episode, and the household fixed effect captures any time-invariant household
characteristics that influence school enrollment.

The identification strategy can be illustrated using a two-by-two difference in differences table.
Panel A of Table 3 shows average school enrollment rates for foster children and the host siblings
they live with for the year before the fostering and the year after the fostering. The results are
imprecise because not all available information is used, in particular children who were fostered
in 1998 are excluded from this table.9 The cross-sectional results indicate that foster children, in
the year prior to the fostering, are much less likely to be enrolled compared with the host siblings
they are currently living with. Average enrollment for host siblings is 35.0 percent while only 12.4
percent of foster children are enrolled. These results are consistent with previous research that
uses cross-sectional data. For both foster children and host siblings, average enrollment increased
after the fostering, but it increased more for the foster children. The difference in differences result
9For households fostering a child in 1999, I use 1998 enrollment as pre-fostering enrollment and 1999 enrollment
as post-fostering enrollment. For households fostering a child in 2000, I consider 1999 as pre-fostering enrollment and
2000 as post-fostering enrollment. Results in Table 3 are similar if instead I exclude the children fostered in 2000 and
use enrollment in the year of fostering as pre-fostering enrollment and post-fostering enrollment is the following year.

11



More intriguing information

1. Weather Forecasting for Weather Derivatives
2. Protocol for Past BP: a randomised controlled trial of different blood pressure targets for people with a history of stroke of transient ischaemic attack (TIA) in primary care
3. Trade Liberalization, Firm Performance and Labour Market Outcomes in the Developing World: What Can We Learn from Micro-LevelData?
4. The name is absent
5. Labour Market Flexibility and Regional Unemployment Rate Dynamics: Spain (1980-1995)
6. WP RR 17 - Industrial relations in the transport sector in the Netherlands
7. The name is absent
8. The name is absent
9. he Effect of Phosphorylation on the Electron Capture Dissociation of Peptide Ions
10. Outsourcing, Complementary Innovations and Growth
11. Volunteering and the Strategic Value of Ignorance
12. Road pricing and (re)location decisions households
13. Long-Term Capital Movements
14. Spectral density bandwith choice and prewightening in the estimation of heteroskadasticity and autocorrelation consistent covariance matrices in panel data models
15. Integration, Regional Specialization and Growth Differentials in EU Acceding Countries: Evidence from Hungary
16. Internationalization of Universities as Internationalization of Bildung
17. EMU's Decentralized System of Fiscal Policy
18. European Integration: Some stylised facts
19. ‘Goodwill is not enough’
20. Standards behaviours face to innovation of the entrepreneurships of Beira Interior