Evidence from Five Countries

Report
Economic Impacts of Immigration and Population Diversity, University of Waikato, April 2012
Scientific Mobility and Knowledge
Networks in High Emigration Countries:
Evidence from the Pacific
John Gibson, University of Waikato and Motu
(with David McKenzie, World Bank)
Competing Narratives
• Brain Drain – decimates the human capital
and fiscal revenues of sending countries
Migration → countries without a local research
base, weak universities, few doctors etc
• Brain Gain – potent force for developing local
economy through knowledge transfers,
remittances, trade, FDI, incentive effects
Migration → diaspora help bring world scientific
frontier closer and eventually return to develop
your own Silicon Valley
Issue is also of interest to (potential)
destination countries
• Confusion in potential destination countries about
whether it is a good or bad thing to recruit skilled
workers (or retain skilled foreign students) from highemigration developing countries
• Mayor of New York City (2011 speech to US Chamber of Commerce):
“It’s what I call national suicide… We are investing millions of
dollars to educate these students at our leading
universities, and then giving the economic dividends back
to our competitors – for free. … So instead of staying here
to contribute to our economy, they go home and go to
work for companies that compete with our own. That
makes no sense.”
• New York Times (March 7, 2012)
“America is Stealing the World’s Doctors”
What does this paper contribute?
• empirical micro evidence on knowledge flows from
the international mobility of researchers
– Focus on the places with highest skilled emigration rates,
which are small island countries
– Most previous case studies for India, China, Taiwan, which
are atypical
• Find evidence contrary to two claims in the literature
– Too few researchers ever return to have much impact
– No productivity gain to researchers from migration
• Migration is important channel for knowledge
acquisition and knowledge transfer for researchers
from small island countries
Prior approaches
• focus on individuals from a selected profession
–
–
–
–
–
highly cited physicists (Hunter et al, EJ)
ISI highly cited (Weinberg, JDE), with e-mail address (Trippl, RegStud)
Israeli economists (Ben-David)
Non-US faculty of US PhD-granting chemistry departments (Gaule)
Italian researchers and scientists (Constant and D’Agosto, AIEL)
• Problems with this for looking at scale and consequences of
scientific mobility:
1.
2.
3.
4.
Initial decision to enter discipline may be tied to desire to migrate
Training for this occupation/discipline may occur through migration
Whether remain in the occupation may depend on migration
Frames based on revealed career performance may draw stayers
from more extreme end of the tail
 Right counterfactual for a high-skilled individual abroad not
necessarily someone in the same job at home
Our approach
• Define a target sample of interest whose
members can be identified before migration
has occurred
– ex post define the sample frame in terms of ex
ante characteristics
• Then survey these individuals regardless of
their subsequent emigration and occupational
choices
– No survivorship bias and can test for, and adjust,
for differential survey rates
Highly Skilled Migrants
• Target sample of interest: “The best and the brightest”
– individuals who were the best students in their countries
at the end of high school from 1976-2004 in three Pacific
countries: Tonga, PNG and NZ
– Approx one-quarter are ever researchers (self-identified)
• Natural counterfactual for measuring impacts
– Outcomes for the high skilled who did not migrate
• Comparison of returnees and migrants sifts between
two theories of migration
– local attributes of the country of work make individuals
more productive when they are abroad
– workers learn how to be more productive when working
abroad and can bring that knowledge back with them
Country choice
• Skilled emigration rates are highest for small island
states (Docquier-Marfouk, Beine et al)
– These are the places where concerns about brain drain
most vivid, so seem the right place to look at the
consequences
– Many more countries like these than like China, India and
Taiwan
• We chose:
– New Zealand: 4.3 million, $28,000 GDP per capita, 15.8%
brain drain rate (with Ireland, equal highest in OECD)
– Papua New Guinea: 6.4 million, $1010 GDP per capita,
19.8% brain drain rate
– Tonga: 110,000, $2560 GDP per capita, 65.1% brain drain.
Tracking and surveying
• Extensive tracking effort attempted to locate
these individuals wherever they are today and
survey them
– Used visits to schools and home communities, school
networks, online search, phone book searches, peer
search, etc.
– Cost about US$200 for each respondent to track,
locate and survey, excluding time cost for PIs
• Survey then administered online (using
SurveyMonkey), in-person, or by phone
– of those in the survey frame, 44% responded
How good is this survey rate and what
non-contact/non-response bias is likely
• Overall survey rate of 44% compares well to other studies:
– Trippl gets 25%, conditional on having an e-mail address
for survey of ISI highly cited scientists
– Constant et al. get 28% despite using pre-existing
database of Italian scientists and researchers
• Our surveys made repeated passes through the
frame, searching with greater intensity for those
not found and this gives a test for response bias
– supplement all the regressions in our EJ paper with indicator for
whether respondent was surveyed latter in the year for each
country and find no evidence of differences, suggesting that those
not found/not surveyed may not differ from those whose
responses we obtained
Understanding the determinants
(J. Development Economics, 2011)
• Returnees leave behind much higher incomes
and return for mainly non-economic reasons
– Family commitments
– Scholarship/bonding commitments
• Returnees do not earn significant income
premium over the similarly skilled nevermigrants
– supports view that local (non-portable) attributes of
the country of work make individuals more
productive when they are abroad
High rates of migration and of return
migration in the overall sample
% Ever Migrate
% Current Migrants
100
90
80
70
60
50
40
30
20
10
0
NZ
PNG
Tonga
 Mistake to think in terms of irreversible “brain drain”
Even higher mobility rates for those who
are ever researchers
(but lower return rates)
% Ever Migrate
% Current Migrants
100
80
60
40
20
0
NZ
PNG
Note: based on those age 22+, moving for work or study
Tonga
Measuring the consequences
(Economic Journal, 2012)
• largest gains to the migrants themselves, US$20,000-$40,000
per year but often ignored in the migration literature
– Labour/immigration literature looks at impact on natives in
destination (or compares with natives)
– Development literature looks at impact on those left behind
• Also significant human capital gains
• Annual remittances exceed fiscal costs for Tonga
– Lean public sector, flat income tax and comprehensive
consumption tax are good settings for high emigration countries
– While there are high remittances from skilled PNG migrants, fiscal
costs are higher because of progressive income tax
• Very little business investment or trade facilitation
– Remoteness and low economic density that make emigration
attractive, also limit scope for business development
Gains and losses (annual) from high
skilled migration
Net Income Gain
Net Remittances
Net Fiscal Cost
50
US$ (000)
40
30
20
10
0
NZ
PNG
Tonga
Researcher sub-sample
• Self-defined “ever been a researcher – including doctoral
and post-doctoral students”
– Highest for the NZ sample – 43%
– Ca. 10% for PNG and Tonga, which we combine into one group
• Individuals who could, potentially, publish scientific articles
although not all did
– Ex ante, policy makers who tinker with innovation and training
policy to restrain emigration or attract returnees don’t know
which researchers will be successful
– Therefore a sample from a relevant population
• Early career, with average age of 35
– Use Harzing Publish or Perish rather than Web of Science to look
for impacts, since many citations are to working papers
Researcher activity
• Current migrants are more active and have larger
impacts than the returnees and never-migrants
– 50% higher odds of publishing and unconditional
number of articles is twice as high as for the returnees
(NZ sample)
– Total citations almost twice that of the returnees and
10x higher than the never migrants (NZ sample)
– Returnees to PNG+Tonga more similar to the current
migrants in activity and impacts
• Reflects some highly productive medical researchers who
returned to PNG
• More constrained return to PNG from bonding provisions
Research activity (risk of publishing academic
article in last five years) by migration status
Current migrants
Returnees
Never-migrants
1
1
1
articles
1
0
0
0
0
0
NZ
PNG+Tonga
Research activity (unconditional number of
articles in last five years) by migration status
Current migrants
Returnees
Never-migrants
6
5
articles
4
3
2
1
0
NZ
PNG+Tonga
Scientific impacts (h-index) by
migration status
Current migrants
Returnees
Never-migrants
9
8
7
h-index
6
5
4
3
2
1
0
NZ
PNG+Tonga
Apparent returnee premium just
reflects age differences (NZ sample only)
• Number of papers, total citations and h-index
show no significant difference between returnees
and never-migrants once condition on age
h  0.4
(3.0)**
Age  4.2 Migrant 0.9 Returnee 10.5
(4.3)**
(0.6)
(2.4)**
 Returnees have no significant productivity
premium over the similarly skilled never-migrants
– supports view that local (non-portable) attributes of
the country of work make individuals more productive
when they are abroad
Researcher networks, dissemination
and funding
• Current migrants have larger co-author networks
but little overlap with home country researchers
(NZ sample)
• NZ expatriate researchers have significantly
higher grant funding and much higher fraction
(30%) are funding from international grants
– No significant differences between returnees and
never migrants in terms of funding
• Returnees to PNG+Tonga have largest co-author
network and more active conference networks than
current migrants
Location of co-authors by migration status
Current migrants
Returnees
Never-migrants
Number of co-authors (last 5 years)
12
New Zealand
PNG + Tonga
10
8
6
4
2
0
Home country
Other country
Home country
Other country
Research Dissemination
• Rare for expatriates to return to present at home
country academic conferences or to home country
audiences in government/university seminars
– Once every five years, on average
– Only 10% of conferences that expatriate researchers
attended in the past year were in the home country
• Return migrants may be conduit for dissemination
– Attend conferences with similar frequency to expatriates
but with 50/50 split between home country and
international conferences
– Much more active than the never migrants
Research funding (US$, all grants, last 3 years)
by migration status
Current migrants
Returnees
Never-migrants
120,000
100,000
h-index
80,000
60,000
40,000
20,000
0
NZ
PNG+Tonga
Conclusions
• High rates of mobility, with significant return
– Contrary to results from a sample frame based on
those from developing countries who started their
research career in the US
• Large positive impact of migration on researcher
productivity and impact
– Contrary to results from a sample frame of highly
cited researchers, where the stayers were already
from the scientifically advanced countries
• Return migrants are the ones actively engaged in
maintaining international knowledge networks

similar documents