Trade and Development: The Growth Dimension L Alan Winters - - PowerPoint PPT Presentation

Trade and Development: The Growth Dimension

L Alan Winters Professor of Economics University of Sussex CEPR, IZA and GDN L.A.Winters@sussex.ac.uk

Preliminaries

• Growth is a (the?) fundamental question for

development

• Intuitively trade is a major influence

– Think how an autarkic village would fare – Theory and empirical evidence support this view

• One determinant of trade is simple and cheap

to change (technically) – trade policy

17th September 2015 WIDER 2

Why trade?

Net Exports = Production - Consumption

– Specialization

• Economies of scale

– Inputs and technologies from abroad – Competitive pressures

• Selection, X-inefficiency, innovation

– All these allow governments/countries to grow

• utput without worrying about demand
• Britain (1700) vs. China (1983)

17th September 2015 WIDER 3

The Long Run: Income or Growth?

(1) ln y = f(o) → Δln y = f(Δo) (2)Δln y = f(o) → Δ2ln y = f(Δo)

17th September 2015 WIDER 4

Theory

• Continuing interest
• Neo-classical model:

– Accumulation – Determinant of technology/efficiency

• Multi-sectoral Models

– ‘right’ sectors – Innovation sector – competes for resources, comparative advantage rules – Trade and spill-overs in innovation – Scale (+) and competition (-)

17th September 2015 WIDER 5

Theory

• Continuing interest
• Neo-classical model:

– Accumulation – Determinant of technology/efficiency

• Multi-sectoral Models

– ‘right’ sectors – Innovation sector – competes for resources, comparative advantage rules – Trade and spill-overs in innovation – Scale (+) and competition (-)

17th September 2015 WIDER 6

The Long Run: Income or Growth?

(1) ln y = f(o) → Δln y = f(Δo) (2)Δln y = f(o) → Δ2ln y = f(Δo)

(2) y” Liberalisation (Δo) y (1) y’ time ln(y)

17th September 2015 WIDER 7

Empirics I: Causation - Instruments

• Early empirics debunked by Rodrik and Rodriguez
• Frankel and Romer (AER, 1999) gravity instruments
• Two issues (Bazzi and Clemens, AEJ-Mac, 2013)
• Instruments are weak, except country size - treacherous
• Be serious about the exclusion restriction

– Many studies in the literature take the form: growth = f(x, W, u); x=g(Z) – If you estimate: growth = h(y, W*, v); y=m(Z) – You strictly have to reject every one of them!

17th September 2015 WIDER 8

Other instruments

• Feyrer (2009, NBER WP)

– Time varying instrument for trade – Importance of difference in sea distance and air distance becomes more significant as air travel cheapens

• Romalis (2007, NBER WP)

– US tariffs (also time varying), but

• Commodity structure;
• Relates trade policy level to output growth

17th September 2015 WIDER 9

II: Conditions for trade to boost income

• Heterogeneity of country studies
• Weaker effects of liberalisation in poor

countries

• Benefits for low-income Africa uncertain
• Interactions of openness with various

conditions

– Chang, Kaltani and Loayza (2009, JDE) and – Bolaky and Freund (JDE, 2008)

17th September 2015 WIDER 10

17th September 2015 WIDER 11

What does all this mean for policy?

• Policy is a decision, not a hypothesis test

– Have to decide (doing nothing is a decision) – Balance of evidence and priors

• Evidence never fits your case precisely
• Parameters imprecise (standard errors)
• Model may also be wrong (more complex)

– Consider the whole distribution of possible

• utcomes

– Costs of different errors – Cost of uncertainty per se

Brock and Durlauf (WBER, 2001)

17th September 2015 WIDER 12

Distribution of growth increments

Percent change in growth over liberalisation 2 4 6 8 10 12

• 2
• 1

1 2 3 4 5 6 7 More

percent p.a. number

Kneller et al (2008) Sample

• f 47

liberal- isations

Less than

Mean effects is 2.38 (3.88)

17th September 2015 WIDER 13

Productivity

• Trade liberalisation raises productivity

– Basis of higher incomes – But recall structural change lit.’s concern

• Several possible mechanisms

– Competition/selectivity – Melitz – Imported inputs – Changes in extensive margin (product/market mix) – Learning from foreign interactions

• Firm-level data, increasingly with trade data

Thank you

WIDER 17th September 2015 14

Supporting Slides

17th September 2015 WIDER 15

Slower and Smaller

• Doubling of GDP pc from $1300 PPP to $2600; Maddison

Date of doubling Initial

• Popn. Mn.

Years to double

Britain 1700-1855 9 155 USA 1820-1873 10 53 Germany 1830-1894 28 64 Japan 1906-1939 47 33 China 1983-1995 1023 12 India 1989-2006 822 17

WIDER 17th September 2015 16

Openness and Income

• Long-run growth
• Transmission/shocks
• Productivity
• Theory
• Cross-country panels
• Structural models
• Firm-level data

17th September 2015 WIDER 17

Fundamental question for development

Theory

• Static trade model – opening trade increases

the value of output at world prices

• Neo-classical growth model y = ALαHβK(1-α-β)

– Trade may affect accumulation

• Via factor returns – K and H
• By affecting prices – K
• Recent revival via structural modelling, e.g.

– Anderson, Larch, Yotov (2015, NBER 21377) – Eaton, Kortum, Neiman, Romalis (2015)

– Trade usually taken to affect A (levels or changes)

• If introduce distortions, result is ambiguous

17th September 2015 WIDER 18

Determinants of Trade

• Gravity model explains by economic mass and

trade costs (incl. distance)

• Can track back from trade to welfare to

calculate benefits of Δtrade costs d ln W = - (1 – η) d ln λ / ε

− λ is share of domestic goods in expenditure − ε is elasticity of trade wtr trade costs − η determined by changes in mark-ups – small >0

Arkolakis, Costinot, Donaldson,Rodriguez-Clare (NBER 21370)

17th September 2015 WIDER 19

Multi-sectoral models

• Structural models of development
• Trade may → wrong specialisation - level or growth
• Now recognise innovation as being produced

— Competes for resources – trade may shift resources in or out — Trade affects size of market for ‘innovative’ goods — There is comparative advantage in innovation

17th September 2015 WIDER 20

Endogenous Growth

• Knowledge has public elements, i.e. spill-overs

– Between firms, partially between countries – May be conscious – FDI, learning from customers – May depend on trade or FDI (or other) links

• Two effects – scale (+) and competition (-)
• Can combine with comparative advantage

Grossman and Helpman (1991, and 2015, AER P&P)

17th September 2015 WIDER 21

Income, Openness and Labour Regulation

(Bolaky-Freund; JDE 2008)

Less regulated half More regulated half

17th September 2015 WIDER 22

17th September 2015 WIDER 23

• Chang et al:
• Linear interaction

– Effect of opennessi = α+ γCi

• Conditions are highly correlated, tested 1-by-1

– Is it just development that matters? – Is it just catching Africa in 1980s/1990s? – Is Africa different in 2000s/2010s?

Δln y = ..αo+βC+γ(C*o)..

17th September 2015 WIDER 24

Policy-Relevant Growth Econometrics

Brock and Durlauf (

Δln y = Xα + Zγ + βΔtariff

X ‘maintained’ variables; Z ‘optional’ variables

• Hypothesis testing

H0 : β=0 against H1 : β>0

• Trade policy ‘affects’ growth if t-statistic on β > 2
• Issues: (a) uncertainty about β

(specification as well as sampling ), (b) relative size of Δln y and βΔtariff

Transmission/Shocks

• Volatility affects growth (and welfare)
• Trade transmits shocks from abroad
• But also dissipates domestic shocks
• Net effect on volatility uncertain
• Raddatz (JDE, 2007): most shocks are internal
• Caselli et al (2015, CEPR 10775):

– structural Eaton-Kortum model – In 2/3rd countries trade reduced volatility 1972-2007 – Diversifications across partners important

17th September 2015 WIDER 25

Transmission/Shocks

• Volatility affects growth (and welfare)
• Trade transmits shocks from abroad
• But also dissipates domestic shocks

– most shocks are internal – In 2/3rd countries trade reduced volatility 1972-2007 – Diversifications across partners important

• The Great Recession

– Many theories for dip in trade/GDP and recovery – Credit, protection, demand for tradables, GVCs

17th September 2015 WIDER 26

Net effect?

Permanent Shocks

• For example, advent of China as trading power

– Competition effects on consumers and producers; – Demand for commodities → prices – Source of aid and FDI

• Mexican firms (Iacovone et al, JIE, 2013)

– Main locus of competition – US market – Heterogeneous effects – stronger did better

17th September 2015 WIDER 27

Creative Destruction?

How Chinese Competition affects Mexican Manufactures

17th September 2015 WIDER 28

Source: Iacovone, Rauch and Winters (2013)

The Great Recession

• Many theories for dip in trade/GDP and recovery

– Credit, protection, demand for tradables, GVCs – Trade data approaches, e.g.

• with RBC theory → efficiency of investment in durables
• structural model (Eaton, Kortum, Neiman, Romalis, 2015)

– Firm level data – how firms reacted

• Di Ulbado – Slovenia: what explains speed of transmission
• Inventory adjustment – intermediates with largest cost-

shares reacted quickest.

• Related-party trade reacted quicker that arms-length trade

17th September 2015 WIDER 29

Imported Inputs

• Input tariff effect > output tariff effect

– Indonesia 1991-2001 (Amiti, Konings, 2007, AER) – India 1991 (Topolova, Khandelwal, 2011, REcSt) – Panel (Estevadeordal, Taylor, 2013, REcSt)

• New inputs (Goldberg et al, 2009, AER P&P)
• Applies to services too

– Czech Rep (Arnold, Javorcik, Mattoo, 2011, JIE) – US outsourcing to India (Amiti, Wei, 2009, WE)

17th September 2015 WIDER 30

Learning by Exporting

• Exporters have higher productivity

– Learning vs. selectivity? Timing for identification – Indonesia 1990s (Blalock, Gertler, 2004, JDE) – Panel China in Asian crisis (Park et al 2010 REcSt)

• Spillovers to other firms

– China, 2000-06 (Fernandes, Tang 2014, JIE)

• Learning to Export

– Mexico NAFTA (Iacovone, Javorcik, 2012, CEPR)

• anticipate tariffs changes by raising quality (productivity)

17th September 2015 WIDER 31