Lant Pritchett's presentation - OECD - PGD Expert Meeting

The feasibility and potential impacts of free human mobility

Lant PritchettHarvard Kennedy School and Center for Global Development

Feb. 24, 2015At

OECD first expert group meeting:Perspectives on Global Development

International Migration and Development

Outline

• What would we expect the impact to be? A dizzying and dazzling tour of caricatures of different theories of cross-national income

• What do we find the marginal impact to be:– Micro-economic estimates

– Macro-economic estimates

• How does labor mobility affect the “A” of growth theories?

Massive cross national inequality: incomes from all of modern history exists in countries today

More inequality, and more of it dues to location

The (statistically) rich of poor countries are poorer than the poor of rich

countries

Branko Milanovic’s 21st century trilemma

We are connected instantaneously by internet and cell phone with, and are only a 6 ½ hour flight from Paris to, countries with levels of per capita income last experienced in the West in the Middle Ages—even a country like Morocco has a level of income lower than prior to the Franco-Prussian war.

What we expect the impact of human mobility to be? Depends on our explanation of why incomes are

different

Manna from Heaven (Modern) economic production

Endowed with natural resources that yield income—either concentrated (oil, diamonds, gold, copper) or diffuse (wheat, rubber)—with just sector specific institutions (e.g. enclave) and effort (fixed assets—e.g. holes in the ground)

Example: Kuwait is rich because it has low marginal cost oil

Stuff (K, HK) plus A as “technical knowledge

“A” as “institutions” or “capabilities”—tacit, non-transferrable, augmentersof productivity

(Augmented) Solow model with “capital” and “human capital” and an “A” term interpreted as “technology” or “blueprints” or “codifiedknowledge”

“Institutions” (of the right,creative destruction inclusive market supporting) type are the causal key to prosperity (e.g. AJR, NWW, RS)

A large number of “Capabilities” as non-tradable collective inputs that require tacit knowledge is key to prosperity (Hausmann)

In rich “manna” countries migration welcome—as long as no manna for (most) migrants

On mover On host country On sending country

Mover gets access to manna

Moves to more manna—better off

Less manna per person—worse off

More manna per person—better off

Mover has access to manna in home country does not get access to manna in host country

Mover only moves if wages/incomes are higher in modern sector are higher

Doesn’t have to share manna—depends on impact on modern sector, if manna is abundant (manna/person>reservation wage) then host better off, indeed rely on migrant labor (e.g. Gulf States) for both unskilled and skilled labor

More manna per person—better off (or with only consequences on modern sector)

General approach of augmented Solow-Swan or “neo-classical” growth model

• A is T—the “A” of the production function was “knowledge” that was common to all countries (e.g. laws of physics) and/or would diffuse rapidly across borders so technical progress was a global externality

• K and HK as stocks had to be accumulated by savings/investment today for higher stocks tomorrow and hence growth dynamics were driven by capital(s) accumulation dynamics

• Wage differences across countries driven by HK differences did not lead to migration motivation (“skill price” equalized across countries with same A and K/L

• If K/L drove higher wages then migration had “dilution” effects as workers arriving with no capital drove down average K/L.

• If A(T) converged and capital markets worked then incomes and wages should converge (with dynamics driven by investment) and no need for labor mobility to equalized incomes.

• (Plus models with trade and “factor price equalization” by exchanging goods which embody different K/L ratios equalize w and r via trade)

Nearly everything about neoclassical model as model of “development” is wrong

• There hasn’t been (much) convergence across countries and over the long-horizon massive divergence (Pritchett 1997, Milanovic 2012, Bourginonand Morrison 2002)

• This is in spite of massive convergence in K and HK (Grier and Grier 2007)—so rather than A(T) rapidly converging and convergence dynamics limited by K and HK dynamics A (measured as residual) has been (until perhaps recently) diverging (e.g. Bosworth and Collins ).

• Empirical growth decompositions find that K and HK account for a quarter to at best a third of cross-national differences even as proximate determinants (Casselli 2005, Inklaar and Timmer 2013) (and if K and HK are endogenous to A(I) or A(C) then causally even less)

• The price dynamics of interest rates are wildly inconsistent with equalized A and K dynamics (King and Rebelo 1989) and MPK appears to be already equalized (Casselli and Feyrer 2007)

• Factors flow to rich areas (Easterly and Levine 2001)

Two new models of what the “A” is that makes equivalent L, K and HK more productive

“Institutions” (A is I) • “Institutions” (Acemoglu,

Johnson, Robinson (2000), North, Wallis, Weingast (2012) or “social infrastructure” (Hall and Jones 1999) account for the bulk of the differences in cross-national income

• “Institutions” are “norms” or “patterns of behavior” that structure relationships among actors and “market supporting” institutions that are restraining on the “grabbing hand” and “inclusive enough” for creative destruction are needed

“Capabilities” (A is C)• “Capabilities” (Hausmann and

Hidalgo 2009, Hausmann et al 2011) are the key to prosperity

• “Capabilities” can be generated from a product specific Leontief production function with many inputs and “simple” products require few inputs and “complex” products many inputs so countries with many available inputs—including collectively produced inputs (like public goods, infrastructure, rule of law, specific policies)—are able to produce complex products

Levels of GDP per capita and “institutions”

Higher “economic complexity” as a measure of the countries sophistication of exports in the product

space is highly correlated with level of GDP

Source: Hausmann et al 2011

“Manna” countries(nat’l resource exports>10% of GDP)

Non-resourceexporters

GDP per capita

Economic Complexity of Exports

What would we expect to see at the margin if A as I or A as C explained most cross-national

income/labor productivity/wage dispersion?

Growth model ofA as I or A as C drives levels

For movers For host countries For sending countries

A (I or C driven) is place/countrySpecific, determines the productivity of all factors, and is fixed at the margin

Massive wage gains for movers as their labor productivity (for given HK) is place dependent and the additional productivity is “in the air” (of I and C) and hence rapid and near complete wage convergence for given skills (including language)

Almost no impact at all as at the margin “A” as I or C is fixed and public good (non-rival and non-excludable)Some relative price impacts on types of L and HK and K depending on whether migrants are substitutes or complements

Almost no impact at all, for the same reason.

What properties the new A as “institutions” or “capabilities” has to have (versus the old A as T)

Properties of the new A• “A” does not diffuse rapidly

across national borders (perhaps not even regionally if “capabilities” are place specific due to IRS)

• “A” has a dynamics in which “poverty traps” are possible (lower A leads to less growth of A)—versus “advantages of backwardness” dynamics

• “A” capable of reversals within countries in which “A” deteriorates

Facts to accomodate• Long-run historical and post

war divergence across countries not due to factors

• Countries in long-run poverty/slow growth traps

• Massive and sustained reversals in output per worker at same (or rising) K and HK inputs (Liberia, Venezuela, Cote d’Ivoire, Zambia (‘67-’94))

We see in the data exactly what we expect to see based on A as I or A as C macroeconomic

theories of income (or manna)

• Microeconomic estimates of wage gains

– Data on observational equivalent workers

– Data on differences across occupations

– Experiments (and quasi-experiments)

• Macroeconomic

– Modeling

– Experiences

Wages of observationally equivalent workers between USA and 41 other

countriesWages

Worker characteristics(e.g. schooling, residence, age)

USA born, educatedWorkers in USA

Foreign born, foreign educated workers in USA (wage profile estimated with US Census)

Foreign born, foreign educated workers in foreign country (estimated with country data)

Place premium: Same worker characteristics (for all available observables) just different places

X

Estimated average wage differences of observationally equivalent low skill (9 years schooling), urban, male, formal

sector, young (35 year old) workers between USA and source country is (PPP adjusted) $15,000 a year

Simple arithmetic for 35 year old, male, urban, formal sector, 9 years of schooling:

Wage in Haiti: 80 cents/hrWage in USA: $8.25 /hrAnnual hours 8hrs/day,

22/days month, 12 months year:

(8.25-.80)*(8*22*12)=$15,738

Average (of 41 countries):Wage in foreign: $2.53Wage in USA: $9.83Annual wage gap:

$15,411

0

5000

10000

15000

20000

25000

Haiti

Ghana

Indon

esia

Bang

ladesh

Guatemala

An

nu

al e

an

rin

gs

(in

PP

P)

In home

In USA

Wage gap is absolutely higher the more education (if the proportionate return to schooling is the same in both markets)

Table 1: Estimates of the annual gain in earnings from moving to the USA for a typical

male wage earner from a sample of 41 developing countries

Estimates 9 years 12 years (assuming 10

percent Mincer return)

Observationally equivalent, 100 percent of

spending in USA (PPP adjusted)

$15,298 $20,361

Observationally equivalent, adjust “real”

consumption wage upward for 40 percent of

spending in home (remittances or savings) at lower

prices $23,130 $30,785

Adjust downward by 1.25 (maximum empirically

demonstrated positive selection of low skill

migrants on unobserved characteristics) with 40

percent spent at home$18,504 $24,628

Source: Based on Clemens, Montenegro and Pritchett 2008

The observed wage gap across countries for workers in the same occupation is higher for medium skill (construction workers) than low

skill (waiters)

$13,111

$34,824

$0

$5,000

$10,000

$15,000

$20,000

$25,000

$30,000

$35,000

$40,000

$45,000

Waiters Construction workers

An

nu

aliz

ed

ear

nin

gs, i

n P

PP

Wage in bottom 30 reporting countries inOWW (in PPP)

USA reported wages

Gap between bottom 30 and USA

Source: Pritchett and Smith, forthcoming, based on OWW data (Oostendorp, 2013)

Gains from temporary Tonga-New Zealand migration for agriculture—using lottery to

control of selection

$7,509

$26,381

$18,872

$0

$5,000

$10,000

$15,000

$20,000

$25,000

$30,000

Applicants earnings in Tonga Estimated earnings (using lotteryoutcome for identification)

Migrant gain (corrected for selection)

An

nu

aliz

ed

(5

2 w

ee

k) e

arn

ings

in U

S$

McKenzie, Gibson and Stillman, 2010

Temporary workers on H-1B visas. Exchange-rate dollar annual wage differences for observably and unobservably identical workers producing seamlessly tradable good (software) interpreted as productivity differences. Source: Michael A. Clemens, 2013, “Why Do Programmers Earn More in Houston than Hyderabad? Evidence from Randomized Processing of U.S. Visas”, American Economic Review Papers & Proceedings, 103 (3): 198–202.

Computer programmers (India to US)—

identified from random access to US visas

0

20000

40000

60000

India US

An

nu

al

wa

ge

(e

mp

loy

ed

), X

R U

S$

Temporary workers on 3-year labor card. Exchange-rate dollar annual wage differences for observably and unobservably identical workers, who remit about 85% of these earnings to India, spent at Indian prices. Source: Michael A. Clemens (2015), “Household effects of temporary low-skill work visas: Evidence from the India-Gulf corridor”, Working Paper, Washington, DC: Center for Global Development.

Construction workers (India to Gulf)

0

1000

2000

3000

4000

5000

India UAE

An

nu

al

wa

ge

(e

mp

loy

ed

), X

R U

S$

Experiment: Tonga-New Zealand seasonal mobility for agricultural work with applicants

chosen by lottery

$7,509

$26,381

$18,872

$0

$5,000

$10,000

$15,000

$20,000

$25,000

$30,000

Applicants earnings in Tonga Estimated earnings (using lotteryoutcome for identification)

Migrant gain (corrected for selection)

An

nu

aliz

ed

(5

2 w

ee

k) e

arn

ings

in U

S$

What should be completely uncontested at this point based on our “best available” theories and

evidence• Macro/growth: Most of (upwards of 2/3) of differences in

productivity per worker across countries in the world are driven by “in the air” elements at the national level which diffuse only slowly and can be stuck or reverse

• Micro/wages: The annual wage/labor productivity gain to a low skill (<=12 years schooling) worker from moving from a poor country to rich country of unskilled laborers is P$10,000-P$30,000 per worker—factor multiples (4 to 5) of earnings in their home country—and higher for richer workers

• Legal barriers to mobility imposed by rich countries prevent people from moving in response to these potential gains.

• At the margin (e.g. “A” unchanged) labor mobility has small net positive impacts on the recipient countries with also small distributional impacts on citizens—and possibly, depending on structure, eligibility, etc. some fiscal issues.

Open Borders and the Pushback to Open Borders

• One can do calculations of the total gains from free labor mobility

• If one assumes that country “A” is unaffected by population mobility the gains are in the tens of trillions—a rough doubling of global GDP

109.25

1.20

20

40

60

80

100

120

Median of 4 estimatesof complete labor

mobility

Median of 7 estimatesof policy barriers tomerchandise trade

Gai

n a

s p

erc

en

t o

f w

orl

d G

DP

The key question about the gains from open borders is the responsiveness of the new “A” to

flows of migrants

Change in country specific A(as T? as I? as C?)

Some measure of the stock of migrants (weighted by characteristics)

One (default) theory: Invariant

Collier/Borjasconjecture:Threshold effect after which host country A starts to converge to sending country A

?

Immigrants good(up to a point)

What is a red herring and what is a red flag?

• For marginal increases in the rate of labor mobility this is a complete red herring—unless and until one is at the threshold—and there is no evidence there is a threshold, much less than any OECD country is near it

• For “temporary” schemes this is a red herring—the Gulf states and Singapore have migrant/citizen ratios over 1 (>50 percent migrants) and have not detectable impact on A

• The argument: “a stock of politically and institutionally relevant migrants that passes a threshold may deteriorate the market supporting institutions that create high productivity hence estimates of ‘open borders’ are exaggerated” has ceded the day to all policy relevant arguments as no one is talking about “open borders” as a near term policy or political agenda—just “relaxation” of existing controls.

High immigration OECD countries are at levels of 20 percent or more

0

5

10

15

20

25

30

Fore

ign

bo

rn a

s p

erc

en

t o

f p

op

ula

tio

n

Order of magnitude variation across US states—is Montana doing better on “institutions” than California or Texas?

Lets distinguish sophisticated economic arguments from garden variety xenophobia

(which may be politically powerful)

Cultural norms

• How people dress , worship, celebrate New Years, marry, socialize

• What actions are “honorable” or “dishonorable”

• How “inside” and “outside” the relevant social unit are treated

Market supporting norms

• Restraints on arbitrary action of governments—including ‘rule of law”

• Legal arrangements that allow making and enforcing private contracts and “thick” financial sectors

• “Markets” that allow liberty people and firms to engage in arms-length transactions

“Culture” as determinant of incomes: Largely Claptrap

North and South Korea at night: culture?

Why “culture” doesn’t work well to explain per capita income differences

• Argued as barrier against many countries—just before they have dramatic acceleration: Japan (Meiji), Russia (pre-revolution), China (pre-Deng)

• Measures like “trust” are endogenous and culture adapts (e.g. Fiji)

• Migrant communities succeed even while host country is poor: Indians, Chinese

• Clear examples of common culture, different outcomes

• Singapore higher GDP per capita than any (non-manna) Western country

Elements of a model of “threshold” impact of stock of migrants on “A”

• Norms are properties of places, not individuals, and individuals understand that.

• Not “cultural” norms by “market supporting norms”—the entire point of market supporting norms, especially norms formalized into organizations and laws, is to allow people to transact who don’t share cultural norms.

Elements of the “immigrants erode institutions” argument

• People who choose to move may seek out places with the desire to adopt the new norms

• People who move may choose to adopt compliance with the “market supporting” norms—even perhaps internalized them—if not immediately then over time—the speed of “assimilation” (not “cultural” but “market supporting”)

• If people move from lots of different norms this may not create any impetus to switch to a new norm

A theory about “A as I” needs to reflect the “pressure on I” from migrants, not “migrants”

Threshold of “pressure”

Stock of migrants,Pressure on A from stock

Rapid expansion, little assimilation/internalization, concentrated migrants

Pressure on market support institutions that sustain high productivity

Time

A theory about “A as I” needs to reflect the “pressure on I” from migrants, not “migrants”

Threshold of “pressure”

Stock of migrants,Pressure on A from stock Slower increase in stock

Depending on pace of immigration versus assimilation/internalization, initial gap, concentration, etc. even huge stocks of “foreign born” are compatible with never reaching threshold pressure on institutions

Time

It’s the stupid politics stupid

• The economics of the benefits of greater labor mobility are, at this stage, largely uncontested: at the margin gaps in “A” that are “in the air” produce massive (PPP$10,000 to 30,000 per mover) gains in labor productivity with little impact on citizens (on aggregate, positive)

• The push back on “general equilibrium” effects on A: (1) cedes the field on the policy relevant issues, (2) is, at this stage, a completely empirically unsupported conjecture

• The politics of migration hinge almost entirely on cultural arguments, which have enormous political traction but which are enormously intellectually problematic as “just” or “legitimate” grounds for discrimination