Income maximization and the selection and sorting of international
Oct 25, 2016 - c) Positive sorting: countries with large absolute skill-related wage .... final sample of 15 (OECD) destination countries and 102 source countries.
Income maximization and the selection and sorting of international migrants
by Jeffrey Grogger and Gordon H. Hanson (2011) Presented by Pablo Jaramillo and Arnaud Wolff
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
1 / 28
I. Introduction a) Context
n
n
Most international migrants leave home bound for rich nations – In 2005, 40.9% of the global emigrant population resided in just eight rich economies, with 20.2% living in the US alone Another striking feature of international labor flows is that the more educated are those most likely to move abroad
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
2 / 28
I. Introduction b) Positive selection: the more educated are more likely to move abroad
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
3 / 28
I. Introduction c) Positive sorting: countries with large absolute skill-related wage differences attract a disproportionate share of more educated emigrants
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
4 / 28
II. What is the aim of the paper? What do they want to prove and how are they going to proceed?
n
n
They want to evaluate the contribution of wage differences to the destination choice of skilled migrants To do that, they construct a theoretical model and then estimate the: – scale of migration – selectivity of migrants in terms of schooling – sorting of migrants in terms of schooling
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
5 / 28
III. Contribution of the paper to the literature
n
n
n
The authors address conflicting results on migrants selectivity – Borjas (1987) predicts negative selection in most South-to-North migration The authors want to reconcile the Roy (1951) model with the strong positive selection seen in the data – Absolute wage differences (linear utility) are more important than relative wage differences (log utility) for migrant selection The authors take the role of taxes into account in the destination decision of the migrants
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
6 / 28
IV. Theoretical comparison of the linear and log frameworks a) The wage and the cost function
n
n
n
Notations: – i = migrant – s = source-country – h = destination-country – j = skill level (primary (1), secondary (2) or tertiary (3) education) Wage function j 2 2 3 3 – W ish = exp[µh + δh Dis + δh Dis ] Cost function j 1 1 2 2 3 3 – C ish = fsh + gsh Di + gsh Di + gsh Di
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
7 / 28
IV. Theoretical comparison of the linear and log frameworks b) Utility functions
n
n
n
Linear utility function j j j j – U ish = α[Wih − Cish ] + ish where α > 0 Log utility function j j j j λ – U ish = [Wih − Cish ] exp[vish ] where λ > 0 The workers choose whether and where to emigrate so as to maximize their utility
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
8 / 28
IV. Theoretical comparison of the linear and log frameworks c) Scale equation: fraction of population that migrates
n
The scale equation under linear specification –
n
ln
j Esh Esj
j = α[Whj − Wsj ] − αfsh − αgsh
The scale equation under log specification –
ln
j Esh Esj
= λ[lnWhj − lnWsj ] − λmjsh with mjsh =
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
j fsh −gsh j Wh
25 October 2016
9 / 28
IV. Theoretical comparison of the linear and log frameworks d) Selection equation: relative emigration of high-skill versus low-skill individuals
n
The selection equation under linear specification –
IV. Theoretical comparison of the linear and log frameworks f) Predictions
n
Their model predicts that an increase in the absolute difference in earnings between high and low-skilled workers in destination countries (Wh3 − Wh1 ) causes: n migration from source countries to rise for the more educated ones n the mix of migrants to become more skilled
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
12 / 28
V. Data and illustrations a) Data description
n
n
n
Data from Docquier et al. (2009) n stock of migrants of 25 years and older, by level of education (primary, secondary and tertiary) n final sample of 15 (OECD) destination countries and 102 source countries Wage data from Luxembourg Income Study (LIS) and World Development Indicators (WDI) th n 20 percentile of income distribution as low-skill wages and 80th percentile of income distribution as high-skill wages Data on taxes only for destination countries
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
13 / 28
V. Data and illustrations b) This graph shows positive selection in the data
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
14 / 28
V. Data and illustrations c) This graph shows relative differences in wages
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
15 / 28
V. Data and illustrations d) This graph shows absolute differences in wages
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
16 / 28
VI. Estimation procedure a) Scale equation: fraction of population that migrates
n
The scale equation under linear specification – –
– n
ˆj E
j j j 3 ln Eˆsh j = α[Wh − Ws ] + xsh β + I(j = 3)xsh β + ηsh s with xsh a vector of characteristics of the source-destination pair (geographic and linguistic distance, etc.) I(A) is a dummy variable equal to 1 if A is true, 0 otherwise
The scale equation under log specification –
ln
ˆj E sh ˆsj E
j = λ[lnWhj − lnWsj ] + xsh θ + ηsh
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
17 / 28
VI. Estimation procedure b) Selection equation: relative emigration of high-skill versus low-skill individuals
n
The selection equation under linear specification –
VI. Estimation procedure c) Sorting equation: skill composition of migrants by destination
n
The sorting equation under linear specification –
ˆ3 E
ln Eˆsh = α[Wh3 − Wh1 ] + xsh γ + τs + ηsh 1 sh
n
The sorting equation under log specification –
ˆ3 E
ln Eˆsh = λδh3 + ρs + ηsh 1 sh
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
19 / 28
VII. Main results a) Regression results from linear-utility model
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
20 / 28
VII. Main results b) Regression results from log-utility model
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
21 / 28
VII. Main results c) Summary
n
The linear specification predicts positive selection, which is consistent with the data illustrations
n
The log specification predicts negative selection
n
Both specifications predict positive sorting. Why?
n
Post-tax wage differences are more powerful to explain positive sorting compared to pre-tax wage differences – this may explain why Canada and US (which have less progressive tax systems) receive such a large share of skilled migrants
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
22 / 28
VIII. Robustness checks a) They first check on alternative wage measures and PPP-adjusted wages
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
23 / 28
VIII. Robustness checks b) They also check for the country where education has been done, quality of university, and networks
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
24 / 28
IX. Decomposition of the immigrant skills gap
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
25 / 28
X. Conclusion
n
n
The authors showed that migrants for a source-destination pair are more educated relative to non-migrants, the larger is the skill-related difference in earnings between the destination country and the source – Positive selectivity is stronger where the reward to skill (in levels) in the destination is relatively large The relative stock of more educated migrants in a destination is increasing in the level earnings difference between high and low-skilled workers – The income maximization hypothesis has been confirmed
Université Paris 1 Panthéon-Sorbonne
M.E.D.E.
25 October 2016
26 / 28
XI. Critical assessment
n
n
n
The generalization of the fact that they obtained meaningful estimates of wage differences from a commonly available data source Due to lack of data, the empirical strategy doesn’t account for individual migratory policies of destination countries Their specification of the log-utility model doesn’t take into account liquidity constraints, which could restrict emigration of low-skilled migrants
i=1, 2, with n1+n2=n, dn(x, y)=Fn(dn1(x1, y1), dn2(x2, y2)) ...... There are some advantages in ... try[ing] to catch the extent of inequality in some objective sense...
Feb 24, 1984 - "Selection and Evaluation of Wood" is an excerpt from the first of a ... content reduces, wood becomes stronger in all respects and tougher or ...
Dec 3, 1999 - Western societies are not 'layer cakes' where the rich stay rich and the poor stay poor. They are more like roller-coasters: people go up and ...
Independent management consultant ..... However, it is essential that they draw on the accumu- .... representatives so that you can discuss the technical speci-.
surrogacy, the surrogate mother carries a couple's embryo; in partial surrogacy she ..... the specific aim of evaluating suitability of both parties in the surrogacy ...
classes of antibiotic, appeared to offer the key to controlling .... Review. Trends in Ecology & Evolution February 2013, Vol. 28, No. 2. 111 ... the same answer: Mosaic > Periodic application. By ..... (1994) Computer simulation of insecticide.
Page 1 .... If all incomes grow but everyone keeps their position (or rank) .... 13 There are not a sufficient number of children of other races to calculate precise ...
advanced economies, has raised concerns about its economic and social consequences (Piketty,. 2014 ... Three theoretical landmark papers explaining the finance-inequality nexus are Banerjee and ... the discussion of the existent theoretical predictio
Aug 4, 2016 - quantify the performance of the algorithm. Then, we ... Therefore, our method appears to be a general, fast and ..... imental protocol in vitro. ..... and is available with its full documentation at http://spyking-circus.rtfd.org. ... w
Department of Neurology, Box 8111, Washington University School of Medicine, 660 S. Euclid Ave.,. St. Louis, MO 63110, ... developed the view of two motor systems, the ... posture of the rest of the body. Prior to the ... are inhibitory and project t
of the striatum results in relatively long latency ... It has long been known that medium spiny neurons ...... can combine with the abnormally reduced contrac-.
5.1 Introduction. Cervical cancer is one of the ... test for cervical cancer is the Papanicolaou (or âPapâ) smear, which involves ..... Genetic algorithms have been widely used for feature selection in the last decade or so. The GA .... developme
Our work has focused on the neural processes that select the target for an eye ... The selection of the target leads to growth of movement-related activity at.
CABG Oriented. â¢Standard multiple views focused on detection of. âsignificantâ stenoses and status of distal vessels (run- off, site anastomosis). â¢No limitations to ...
Jan 8, 2005 - ... was broken. âDiane Ravitch in The Troubled Crusade: American ... an effect is this? We ask the reader to consider the following question. ... argues that signaling could play a role in the expansion of wage inequality in the postw
you, as a private individual, might not want to use. In every .... angle of attack, although you may wonder about some of .... consultant for companies with a desire.
consequences that implicationally related error patterns have for acquisition and clinical treatment are considered. The paper concludes with a brief summary.
perturbations, both natural and man-made. ..... Island (5 female, 4 male), 8 Kanowna Island (5 female,. 3 male), and 10 Phillip Island (5 female, 5 male) indi-.
GRC.NASA.gov/. WWW/K-12/airplane/map.htm, was created by Tom Benson at NASA's ..... consultant for companies with a desire to simulate their aircraft designs.
Mar 16, 2012 - Aix-Marseille School of Economics ... âThis paper is produced as part of the project European Firms in a Global Economy (EFIGE), ... Centre National de la Recherche Scientifique (cnrs) em- .... or the local schooling system. ..... To
on the chosen geometry (subjective choice) of the set of probability measures. ... Bayesian rule after observing the data to give the a posteriori distribution (see ...
3In section V-A, there is an example illustrating the violation of these .... allowed to belong to the whole set ËP and the a posteriori P(p | z) is zero outside the ...