Education Flashcards
Human capital theory
- Private returns (income) to education. Positive correlation
- Still depends on quality of education
Signalling theory
- Schooling = signal (to show that you have skills). NOT necessarily a way to acquire valuable skills.
How does signalling theory work?
- Problem: information asymmetry between employer and job candidate
- During the interview, the employer does not know the candidate’s productivity. E hires C blindly (lottery)
- At time +1, the employer hires another C or not based on the SAME CV if the C proved to be productive (statistical discrimination, not taste-based discrimination)
- For a C, certain characteristics cannot be changed, but education can.
- Students chose to study if the costs to acquire this signal are less than the wage potentially earned.
- Therefore, schooling is only a SIGNAL.
Implications of human capital theory and signalling theory?
Increase in education for all is associated with:
HCT: GDP increases
Signaling theory: GDP stays the same
How to calculate? Empirical impact of education: micro studies
Estimate (for a bunch of individuals i):
Log(Wage i) = a + b * School i + c *Experience i + d * Experience i 2
- All studies find a b between 0.05 and 0.15 so:
- 1 more year of education brings between 5% and 15% more salary
- Strongest correlation in the US. Weakest in Sweden.
- Low impact in OECD/high income countries (many years of schooling, low rates of return.) High impact in Sub-Saharan/low income Africa (fewer years of schooling give very high returns).
Problem of microstudies?
Econometric problems:
- reverse causality
- omitted variable bias:
1) abilities
2) entrepreneurial spirit
3) curiosity - measurement error:
1) authors use qty of education (number of years)
2) but what matters is quality
Solutions?
- Control for other factors (but what is ability?)
- Use twins
- use randomised experiments
- use natural experiments
Angrist et al. (1991) experieme nt
Impact of month of birth in education?
2 laws worth considering:
1) you must be 6 yrs old on the 1st of January of the year you start school
2) the minimum legal age to stop is 16
Examples of Angrist et al. experiment:
- A born Feb 1990: A can start in Sep 1996
- B born Dec 1989: B can start in Sep 1995
If both drop out at 16 yrs old:
- A can stop Feb 2006: 9 yrs, 5 months of schooling
- B can stop Dec 2005: 10 yrs, 2 months of schooling
Conclusion:
A and B similar individuals, but B gets 10 months more schooling
Results?
- compare wage of A and B: 1 more year brings 7% increase in salary (only valid for USA when students drop out at 16)
Problems with micro studies in measuring impact of education?
- We’re only measuring private returns, not social returns
- Maybe social returns more than private if positive externalities for society:
- Peer effect
- Technological progress
- Decrease in crime
- Informed political decisions
- Decrease in fertility
- Improvement in health
- Increase in labour force participation
- Maybe social returns less than private if education is just a signal (Spence, 1973)
- Social returns can only be captured by macro studies
Estimates (for a bunch of countries i) on a macro-scale:
Log(Yi )= a + b * School i
This will capture direct and indirect effects of S (society) on Y
(income)
Problems?
- Reverse causality
- OVB (good policies)
- Measurement error of S
- Measurement error across countries (not the same methodology)
Conclusion midway education lecture?
Micro studies: \+ = good identification strategy - = does not measure externalities (social returns) Macro studies: \+ = measure externalities - = no identification strategy But both give results: 1 year -> 5,15% wage, GDP Education matters for growth
Camera study Duflo et al. (2005)
- Rural India
- Teacher absenteeism of 44%
- Intervention in 60 (randomly) out of 120 schools:
- Give a camera to teachers
- Teachers have to take 1 picture with all students at the beginning of the day, and 1 at the end
- Date and time is written on these pictures (not possible to tamper with the system)
- A valid day: a day with these 2 pictures
- 1 valid day=50 Rs (=6$)
- Wage may vary between 500 Rs And 1300 Rs
- In other schools, wage = 1000 Rs
Researchers organized surprise visits to measure absenteeism
