37
8/13/2019 HC-Lecture 3 Returns to Education http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 1/37 Economics of Human Capital Lecture 3 Education Ian Walker  Lancaster University Management School [email protected]

HC-Lecture 3 Returns to Education

Embed Size (px)

Citation preview

Page 1: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 1/37

Economics of Human Capital

Lecture 3

Education

Ian Walker  

Lancaster University Management [email protected]

Page 2: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 2/37

Last week

• Returns to education – Ability bias

 – Twins

 – IV methods – Matching

• See

 – Card chapter in Handbook of Labour Economics – Ashenfelter et al meta-analysis

 – Goldin and Katz NBER WP

2

Page 3: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 3/37

Overview of this lecture

• Wider effects of education – Clear that beneficial effects go beyond

earnings

• Inequality – Changes in D and S

 – Other factors

• Signalling• Education and growth

Page 4: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 4/37

Non-pecuniary effects

• Oreopolous JEP 2012

4

Page 5: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 5/37

5

Page 6: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 6/37

6

Page 7: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 7/37

Page 8: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 8/37

Some facts about returns

over time• Returns to education in the U.S. fell

during the 1970s

 – when there was a very sharp increase inthe supply of educated workers.

• Returns to education then began a

sharp rise in 80s – Rise slowed in the 1990s

• Race between technology and

education 8

Page 9: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 9/37

Some facts

• This conclusion is robust to manysensible ways of measuring educationreturns

 – In a standard OLS Mincer wage equation,return to year of education rose from about7.5% in 1980 to 10% in 1990.

 – Large increase in wage differentialbetween college and HS grads

 – And between HS graduates and HS

dropouts 9

Page 10: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 10/37

Some facts

•  Also an increase in the estimated returnto “experience” (for younger cohorts).

 – Note: 'returns to experience' typically

estimated x-section data, so this patternprobably has no meaningful predictivepower for lifecycle earnings.

• Many other economies experienced arise in earnings differentials in1980s

 – but only in the UK was the rise as

pronounced as in the U.S. 10

Page 11: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 11/37

Some facts

• Very large increases: US and UK – And Germany

 – recent work by Dustmann, et al using

Germany social security records

• Modest increases: Australia, Canada,Japan, Spain, and Sweden

• No noticeable changes: France andItaly

• Modest falls: Netherlands 11

Page 12: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 12/37

Some facts

•  After the 1990s, rise in overall inequality inthe US gave way to a 'polarization‘ 

 – wage inequality rising in the upper-half of the

wage distribution – but attenuating or even declining in the lower

half.

• Some share of the rise in overall inequality isdue to a rise in 'residual' inequality

 – That is, the inequality remaining after partiallingout the estimated contribution of observables

(such as education and experience). 12

Page 13: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 13/37

Page 14: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 14/37

Theory of skill premia

• Demand and supply – Competitive market

• Two types of workers (L, H), imperfect subs

• CES

• Wages equal marginal products so

14

Page 15: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 15/37

Page 16: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 16/37

Theory of skill premia

• Expansion in education (rise in H/L) reducesreturns to education

• But large expansion in H/L has occurred in

most countries since 1945• Yet ω has risen in most countries

• Something must be happening to AH/AL to

counteract the rise in H/L …. • Labour saving technical progress

 – AH rising faster than AL 16

Page 17: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 17/37

Katz and Murphy

• KM estimate

where γ2 is 1/σ 

• That is, they assume that AH/AL can becaptured by a time trend

• KM uses CPS 63-87 (25 obs) and find

• So σ=1/0.71=1.4

• See also Katz and Goldin’s “race” book 17

Page 18: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 18/37

Other evidence

• Blau and Kahn JPE 1996 – found that labour market institutions (unions and

wage centralization) are much better predictorsof cross-country wage inequality than are supplyand demand indices

• They find little evidence that D and S explainany of the cross-country differences in the

relative earnings of the less-skilled

18

Page 19: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 19/37

Other evidence

• Leuven, Oosterbeek and van Ophem, EJ 2004 use IALS data rather tha education

 – Weak link between IALS scores and schooling

• Estimate relative wages of skill j acrosscountries as a fn of relative D and Sdifferences

• They find β<0 and support D and S story

19

Page 20: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 20/37

Page 21: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 21/37

Page 22: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 22/37

Human Capital and GrowthMacro evidence

•  Aggregating a Mincer HCEF to the economylevel we get 

• Differencing removes technologicaldifferences that are part of the error termterms to give 

• so the S coeff shows how returns havechanged over time, while the S coefficient

gives the (social) rate of return in  j  at time t .22

ln  jt jt jt jt w r S e

ln   j j jt jt j jt w r S r S e

Page 23: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 23/37

Evidence

• O’Neill (JPE 1995)

 – (social) return rising

• The idea that growth rates should converge

is in a feature of many macro-studies – those below their steady-state growth rate

should catch up with those above.

•  

 – where W*  is the steady state level of GDP pc

• Then the macro growth equation wouldbecome

23

*

1  j jt j jW W W u  

1 , 1  ......  j jt j t jW W rS e  

Page 24: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 24/37

Tests

• Many unconvincing attempts• See Brown and Sessions paper

 – in G Johnes and J Johnes, Intl Handbook of

Education Economics• For example, compare returns to self

employed with returns to employed

 – But self-employed are self-selected – Unclear if IV available

24

Page 25: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 25/37

Tests

• One idea –signalling says

that relat ive  educationmatters

 –HC saysabsoluteeducation

matters 25

• Chevalier, Harmon,Walker and ZhuEconomic Journal 

2004

Page 26: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 26/37

Better idea

•  A more direct test would be to show thata ‘signal’ has a positive effect onwages even when acquiring that signal

has no impact on productivity.• Tyler et al  QJE 2000 natural experiment

• Does GED raise earnings?

 – GED is a post-school test that is supposedto be equivalent to getting a high schooldiploma

26

Page 27: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 27/37

TMW strategy

• GED costs $50 and takes 20 hours• GED holders earn less than HS graduates

 – But more than HS dropouts

• Selection issue – GED holders, compared to dropouts, have:

• More S, better cognitive skills, better educated parents

27

Page 28: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 28/37

Method

• GED pass level differs by state!• But if GED score is a good measure of

a person’s ability/productivity, then

people with same‘ability

’ (i.e. 40 − 44)

are given a GED in Texas but not in NY.

• This quasi-experiment effectively

randomly assigns the GED ‘signal’ topeople with the same GED scoresacross different U.S. states.

28

Page 29: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 29/37

Result• What does the signalling model predict

in this case? – People with GED scores of 40 − 44 will earn

more if they get the GED certificate than not 

• But Human Capital model implies thatsince ability is comparable among these

groups, wages will be also• MTW found 20% effect after 5 years

 – Note: MTW method assumes that employers

think that a NY GED is not perceived as betterthan a Texas GED29

Page 30: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 30/37

Aggregation

• Signalling depends on relative education• So if education levels rose generally

 – Signalling model would show no rise in incomes

• HC depends on absolute level of education

• So if education levels rise generally

 – HC model would show rise in incomes

• So how do income levels change aseducation levels rise on average?

 – Aggregate HC earnings function30

Page 31: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 31/37

Human Capital and GrowthMacro evidence

•  Aggregating a Mincer HCEF to the economylevel we get 

• Differencing removes technologicaldifferences that are part of the error termterms to give 

• so the S coeff shows how returns havechanged over time, while the ΔS coefficient

gives the (social) rate of return in  j  at time t .31

ln  jt jt jt jt w r S e

  lnw j   =  r  jS  jt  + r  jt   S  j  +   u jt 

Page 32: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 32/37

Evidence

• O’Neill (JPE 1995) and Lindahl and Krueger(JEL 2001)

 – (social) return rising

• The idea that growth rates should convergeis in a feature of many macro-studies

 – those below their steady-state growth rateshould catch up with those above.

 – where W*  is the steady state level of GDP pc

• Then the macro growth equation wouldbecome 32

*1  j jt j j

W W W u  

1 , 1  ......t t 

W W rS e  

Page 33: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 33/37

Page 34: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 34/37

Externalities

• Several studies examine the effects ofschooling of one group on another group’sincomes

• Most convincing work exploits migration• Iranzo and Peri (2009)

 – variation in the push-driven immigration of highlyeducated workers

 – They find that an increase in college education(but not an increase in high school education)had significant positive productivity externalities.

34

Page 35: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 35/37

Externalities

• Peri et al . (NORFACE Migration DiscussionPaper 2013, and IZA WP 2013)

 – exploit the exogenous variation in the stock offoreign-born “STEM” workers induced by visa

policy changes• 1% increase in the foreign-STEM share of a US city’s

total employment in the US increases wages for nativegraduates by 4-6%

• Small positive effects on native employment

35

Page 36: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 36/37

Endogenous growth

• Endogenous growth models with R&D – Romer (NBER WP 1989, NBER WP2000)

• growth rate partly depends on the level  of human

capital.• Policy that incentivizes innovative activity,

such as R&D tax credits

 – Needs to also encourage a more STEM workers•  Goolsbee (AER 1998)

 – government R&D spending raises wages ofscientists

36

Page 37: HC-Lecture 3 Returns to Education

8/13/2019 HC-Lecture 3 Returns to Education

http://slidepdf.com/reader/full/hc-lecture-3-returns-to-education 37/37

Next week

• Intergenerational transmission

37