Unequal Gains, Prolonged Pain
Protectionism and Democracy in Changing World
Emily J. Blanchard
Tuck @ Dartmouth College
Gerald Willmann
U. Bielefeld & IFW Kiel
Villars-sur-Ollon
February 2017
Motivation
Adjustment Costs Matter... Especially in Politics
Structural change is slow and costly – abundant evidence
that workers face large and long-lasting adjustment costs:
e.g. Artuc, Chaudhuri, McLaren (AER ‘10); Autor, Dorn, Hanson (AER ‘13
+ more); Dix-Caniero (Econometrica ‘14), etc.
Sticky adjustment → even potential “winners” from change
can be losers in the short run → political ramifications
⇒ How does a democracy with slow economic adjustment
respond to unanticipated macroeconomic shocks?
⇒ Do economic frictions carry an additional welfare burden
via political distortions? (Brexit? Trump?... Le Pen?)
Economic Stickiness Shapes Political Outcomes
A crucial distinction:
Winners and losers in the short run:
• Dynamics are key; we need to take time seriously.
Most political economy models are static, steady state, or
rigged to ensure “smooth” adjustments, and thus miss a key
feature of dynamic adjustment.
Winners and losers in the long run:
• Long run outcome depends on distribution of long run gains
from realized adjustment
• Need to recognize endogeneity of structural change,
including the influence of endogenous politics.
Overview of Approach
This paper
1
Develop a new dynamic political economy framework in
which policy can adjust quickly, but economic adjustment
takes more time.
2
Characterize a salient political environment:
• Labor market adjustment to macroeconomic shocks;
• Endogenous policy tool is a tariff – highlights tradeoff
between aggregate efficiency and redistribution.
3
Demonstrate the potential for Political Overshooting and
Permanent Retrenchment in response to shocks.
4
Examine key data markers suggested by the model
Modeling Populist Surges: Brexit/Trump Dynamics
Heterogeneous workers make lifelong (“putty-clay”)
decisions over education, accumulated skills
If expectations are correct, these human capital
investments are ex-post optimal ⇒ steady state policy
under majoritarian voting
Introduce an unanticipated global shock: an aggregate
terms of trade improvement that reduces the returns to
unskilled workers... or SBTC
Skills are stuck, at least for a while, but policy can change.
If median voter is relatively less skilled than ‘average’
⇒ Populist Protectionist Surge
Preview of Results
Permanent Protectionism or Temporary Setback?
Two possibilities over the longer run:
Protectionist Overshooting: Protectionist surge followed by
gradual liberalization
• Voters’ gradually increase skills in response to increasing
skill premium
? Overshooting occurs even if new steady state trade policy is
more liberal; a rosy long-run is preceded by rocky transition.
Permanent Protectionism: Protectionist surge followed by
continuing increase in tariffs:
• Median voter upgrades skills in response to global terms of
trade shock, but “average” skill level increases more
• Skewness in the returns to human capital increases →
median voter increasingly protectionist, skills gradually
deteriorate
Preview of Results
Implications of the Theory
1
Unequal Gains ⇒ Prolonged Pain. Protectionist surge
arises when the median voter bears a disproportionate
burden of the shock.
⇒ Important data marker: mean vs median income changes
2
Efficiency costs of democracy (or limited instruments) are
both static and dynamic.
3
Long run outcome hinges on rising or falling inequality
(skewness) in individual-level returns to openness.
Extensions:
• SBTC → (almost) isomorphic
• Developing country case → ‘liberal overshooting’
4
(Preliminary) U.S./U.K. data suggestively different from
other OECD countries.
Remainder of Talk
Related Literature
Model
Case 1: Protectionist Overshooting
Case 2: Increasing Protectionism
((SBTC; Developing Country Case))
Empirical Implications; Suggestive Evidence
Conclusion
Related Literature
Trade Liberalization and Labor Adjustment Costs:
Matsuyama (1992); Davidson, Matusz (JIE 2008); Artuc, Chaudhuri,
McLaren (AER 2010); Autor, Dorn, Hanson (AER 2013);
Dix-Caniero (Econometrica 2014); Krishna, Poole, Senses (JIE 2014);
Kondo (2013 WP); etc
Dynamic Trade Policy:
Staiger and Tabellini (AER 1987); Fernandez and Rodrik (AER
1991); Brainard and Verdier (JIE 1997); Blanchard and Willmann
(JIE 2011); many others
Economic-Political Feedback; Democracy, Inequality, and SBTC
Acemoğlu and Robinson (JEP 2013); Hassler, Rodrı́guez-Mora,
Storesletten, & Zilibotti (AER 2003); Acemoğlu, Gancia, and Zilibotti
(WP 2012); Acemoğlu, Naidu, Restrepo, Robinson (Handbook 2013)
Sketch of the Model
Individuals and Education
Continuum of heterogenous agents live for 2 periods
Agents born with innate ability, a ∈ [0, 1]
When young, choose optimal educational investment, e.
Cost of education is foregone wages as a young, unskilled
worker. Time constraint:
l+e=1
In second stage of life, education and ability → human
capital, h ≡ h(a, e) s.t.:
ha > 0
he > 0
hee < 0
hae > 0
Sketch of Model, cont.
Production and Trade
Small open economy, Home
Two goods: U , the numeraire and S, a skill-based good
• U : one-for-one in unskilled labor → unskilled wage =1
• S: x(h) ≡ bh where b > 0 (↑ b ≈ SBTC)
Return to acquiring h: bhp, where p ≡
pS
pU
is rel. price of S
Home has comparative advantage in S.
⇒ Liberalization increases relative price of skill-based good;
protectionism decreases it
Educational Investment
Optimal educational attainment maximizes lifetime
indirect utility. For the young voter at time t:
max
e
o
V (pt , Ity (e; pt )) + βV (pt+1 , It+1
(h(a, e); pt+1 )
where V (p, I) ≡ v(p)I.
⇒ Optimal education level, e(a; pt , pt+1 ) is
• increasing in ability level (single crossing)
• increasing in current & future price of S
I decreasing in current and future tariff, all else equal
Politics
Median Voter Model
Majority voting. Median voter is decisive.
Only the old vote.
Individual tariff preference depends on a and education
Individually optimal tariff given by the FOC:
dEts
s
s ∂pt
Vτ (a) = vI [Et (a) − Ēt ]
+ tp
= 0.
{z
} ∂τt
|
dτt
|{z}
|
{z
}
≡∆t (a)
= 0 @ t=0
(−)
| {z }
|
{z
}
individual bias
std optimal tariff
Lower (higher) ability/education ⇔ ∆(a) < 0 (∆(a) > 0)
(1)
Politics
M
Equilibrium Trade Policy, τt = τ (aM ; eM
t−1 (a ), ēt−1 )
Determined by education of median voter born in previous
generation;
Decreasing in median voter education level (holding
remaining old generation’s education fixed);
Depends critically on ∆(aM ); i.e. the median relative to the
average net skill position (thus, human capital, education).
Individually Optimal Tariff Derivation
Solving the Model
Solution Strategy
1
Define political equilibrium using median voter rule and
rational expectations.
2
Steady state defined by τ (eM ; ē), eM (τ ).
3
Adopt ‘nice’ case conditions: unique, interior steady state
4
Shock the economy with a TOT improvement [or SBTC];
study dynamics
Political Equilibrium
Definition
A rational expectations political equilibrium is defined by a
sequence of tariff and education rule pairs, (τt , et (a))t∈N such
that the following hold for all t ∈ N:
1
τt maximizes indirect utility of the median voter at time t;
2
et (a) is optimal for every agent given rational expectations.
Political Steady State
Definition
Political Steady State. A political steady state is reached
when τt ≡ T (eM
t−1 ) = τt−1 ∀t. A political steady state can be
summarized by the steady state education level of the median
voter and concomitant policy outcome pair, {ẽM , τ̃ }:
!
τ̃
ẽM = eM (τ̃ ) = h−1
aM ,
e
βbpw
τ̃
= T (ẽM ; ē˜) = arg max V o (τ ; aM , ẽM , ē˜).
τ
Unique Stable Steady State
Permanent Terms of Trade Shock
Our thought experiment
Consider a permanent TOT improvement (pw ↑) [or b ↑].
For given tariffs, the incentive to acquire education would
rise in the long run, but
In the short run, the median voter will become more
protectionist iff her net export position is below ‘average’.
Characterize transition dynamics and new steady state.
Focus first on the ‘overshooting’ case, then consider
alternative.
Steady state response to ↑ pw
Steady state response to ↑ pw
e(τ ) shifts right/up
Steady state response to ↑ pw
˜ M ) < 0)
τ (eM ) shifts up if median voter relatively less skilled (∆(a
Steady state response to ↑ pw
ẽM ↑; Net effect on τ̃ ambiguous – here, we depict case in which τ̃ ↓
Time Path of Adjustment to pw ↑ at time T
? Policy rule adjusts immediately – Education takes time
Immediate jump in τ
Tariff locus shifts up: τ (eM ; pw1 ) > τ (eM ; pw0 )
Because eM fixed at T ⇒ τT >> τ̃ : tariff jumps at T
w1
Thereafter, τt+1 = τ (eM
t ; p , ēt ).
Median voter’s education stuck at T, then rises
Education Rule et = e(aM ; pt , pt+1 )
Can prove that tariff spike offers partial protection, i.e.
pT > p̃, then eT +1 (a) > ẽ(a) ∀a.
Overshooting Case: If ∆t+1 (aM ) > ∆t (aM ) → τt+1 < τt
∀t ≥ T : tariff gradually diminishes over time
Time Path of Adjustment: Overshooting Case [∆(aM ) ↑ing]
Time Path of Trade Policy Adjustment
Time Path of Trade Policy Adjustment
Note: “Overshooting” can occur with new SS tariff above or below old SS
Time Path of Human Capital Adjustment
Gradual Skill Upgrading
Time Path of Price Adjustment
Policy as Shock Absorber
Efficiency Cost of Overshooting
Or, continuing value of the WTO
Alternative Case
Increasing Protectionism
[∆(a) ↓ing – growing gap in returns to skill]
Alternative Case
Increasing Protectionism
Increasing Protectionism Case:
Time Path of Trade Policy Adjustment
Increasing Protectionism Case:
Welfare Implications
One last possibility
Pit-stop Summary
Protectionist Overshooting or Rising Protectionism?
When human capital gains are skewed toward the top, ↑ pw
causes median voter to become more protectionist, τT ↑
Initial tariff spike will not fully offset the increase in pw , so
education increases among the young generation
In the long run, 2 possibilities:
• If median voter’s human capital (net export position) rises
faster than the average, ∆M ↑ ⇒ tariffs fall and education
continues to rise. τ̃˜ < τT .
• If median voter’s human capital (net export position) rises
slower than the average, ∆M ↓ ⇒ tariffs continue to rise
and educational gains reverse. τ̃˜ > τT .
SBTC (almost) isomorphic in political consequences
Empirical Implications
Conditions for Protectionist Surges and Overshooting
1
The shock makes the median voter more protectionist in
the short-run:
• Median wage is initially below the mean wage. i.e. returns
to openness/skill are skewed toward the top.
• Median real wage falls relative to the mean at the time of
the shock.
2
For Overshooting: Through education/skill acquisition,
skewness in returns to human capital will fall over time.
i.e. the majority can ‘catch up’.
3
For Permanent Protection: the continue to fall behind.
[? ] Are voters are politically enfranchised? (Is median voter
rule a fair approximation?]
The Majority Seems Vulnerable
Median Voter Source: US Census via Haskel, Lawrence, Slaughter JEP 2012
U.S. Mean and Median (Real) Income Since 1975
Mean-Median Gaps (∆) Across Countries
The $64K Question
Can the Majority Share in Globalization’s Gains (eventually)?
A Pessimistic View
Artuc et al (2010); Autor et al (2013+); Dix-Caniero (2014)
Winner-take-all/superstar returns to human capital
Stolper-Samuleson + dist’n of capital [Piketty: r > g?]
Conversely
Broadbased or UBTC could ↓ skewness
Educational institutions designed to enable ‘catching up’
If transfers/educational investment are democratically
determined, inequality could be self-correcting...
Stickiness Index (intergenerational earnings ε)
British and American Exceptionalism? (Let’s hope...)
Protectionist Sentiment on the Rise
Survey Question:
“In general, do you think that free trade agreements between the
United States and foreign countries have helped the United
States, have hurt the United States, or have not made much of a
difference either way?”
Results:
December 1999: 39% Helped vs. 30% Hurt
March 2007: 28% Helped vs. 46% Hurt
September 2010: 17% Helped vs. 53% Hurt
Key feature: The recent converts have college + education
–Wall Street Journal, “Americans Sour on Trade,” 10/4/10 (pg. A1)
Summary
Transitions are important, not obvious, and depend on
differential stickiness.
• The more unequally a shock is felt, the greater and more
persistent the political response
Long-Run democratic response to shocks depends on
skewness in net export position (thus, human capital)
• If shocks exacerbate inequality, protectionism will rise.
• Simply increasing (all) education may not be enough to get
globalization buy-in; need to reduce skewness.
In context of the model, U.S., U.K. data look different and
give pause.
Key question: how flexible are workers in the long run?
Crucial measure is potential and differential labor
market/human capital adjustment.
Broader Contribution
Introduce ‘Policy Overshooting’
Tractable model of political adjustment process based on
simple insight: policy may respond faster than structural
change
Broad range of applications: trapped factors, social
security, alternative fuel adoption; etc.
Thank You!
Conditions for Uniqueness and Stability
Assumption 2
Sufficient Conditions for e locus to cross τ locus once and only
once from below:
lim he (aM , e) = ∞,
e→0
lim he (aM , e) = 0
e→1
Vτoe − o < βbhee pw aM
V |
{z
}
| {zτ τ} aM
dτ
dτ o
de
deM
Alternative Case: Rapid Liberalization
Median voter becomes less protectionist due to shock
Back to Baseline Overshooting Case
Voters’ Trade Policy Preferences
Income:
Ito (a) = |{z}
1 + xs (h(a, et−1 (a)))pt + R(τt )
|
{z
}
| {z }
base rate
skill premium
tariff revenue
Optimal Policy:
τ o (a; et−1 (a)) = arg max
τt
V o pt , Ito (a, et−1 )
(2)
FOC:
Vτ (a) = vI
|
|
Back to Model
∂pt
dEts
−
+ tp
= 0.
{z
} ∂τt
dτt
|{z}
|
{z
}
≡∆t (a)
= 0 @ t=0
(−)
| {z }
{z
}
[Ets (a)
Ēts ]
individual bias
std optimal tariff
(3)
Vox Populi:
Near Universal Protection for Lower-Wage Workers
Source: Lu, Scheve, Slaughter AJPS 2012
Caveat: Rhetoric vs. Policy in Practice
Despite impassioned speeches from the House floor...
“We can’t continue to sit on our hands while Chinese businesses
undercut American workers and our manufacturing base
continues to drift overseas.”
–Representative Bill Pascrell Jr. March 6, 2012 (H1169)
...the data suggest other forces at play...
Despite the Rhetoric...
5 18.00 4.5 16.00 4 14.00 3.5 12.00 3 10.00 2.5 8.00 6.00 2 Manufacturing Imports (% of GDP) 1.5 Manufacturing Value Added (% of GDP) 4.00 2.00 Average Applied Tariff Rate (All Products) Average Applied Tariff Rate (Manufacturing) -­‐ 1 0.5 0 Source: World Bank Statistics (DataBank)
Weighted Average Tariff 20.00 Ye
ar
19 80
19 81
19 82
19 83
19 84
19 85
19 86
19 87
19 88
19 89
19 90
19 91
19 92
19 93
19 94
19 95
19 96
19 97
19 98
19 99
20 00
20 01
20 02
20 03
20 04
20 05
20 06
20 07
20 08
20 09
20 10
Manufacturing as Percent of Imports/GDP Decreasing Output, Falling Tariffs, Rising Imports of U.S. Manufacturing
Protectionism Since the 2008 Crisis
Not a return to Smoot-Hawley... but only thanks to WTO bindings, etc.?
Source: Bown (2011)
Back
Alternative Case
Increasing Protectionism
Alternative Case
Increasing Protectionism
Developing Country Case:
Comparative advantage in U; human capital returns skewed to top as in North
Developing Country Case:
Response to ↑ pw : eM (τ ) shifts right/up
Developing Country Case:
Response to ↑ pw : τ (eM ) pivots counterclockwise
Developing Country Case:
Transition Dynamics for Liberal Overshooting
Developing Country Case:
Time Path of Policy Response