Structural Transformation and Comparative Advantage: Implications for Small Open Economies

• DOI: http://doi.org/10.1111/twec.12398
• Date: 01 April 2017
• Author: Unjung Whang
• Published date: 01 April 2017

Structural Transformation and

Comparative Advantage: Implications for

Small Open Economies

Unjung Whang

1,2

1

Department of International Trade, Korea Institute for International Economic Policy (KIEP),

Sejong-si, Korea and

2

Department of Economics, University of Colorado, Boulder, CO, USA

1. INTRODUCTION

SOME rich countries seem to have relatively high productivity in agriculture than in

non-agriculture (e.g. Australia, New Zealand and the USA). Several Latin American

countries (e.g. Argentina, Brazil, Chile, Mexico and Uruguay), however, failed to achieve a

high level of income during the past half-century despite relatively high agricultural produc-

tivity, whereas some East Asian countries (e.g. Korea and Japan) successfully transformed the

economy into industrialisation despite lower agricultural productivity. To explain these pat-

terns, economists have taken one of two approaches. Most studies consider closed-economy

models where rich countries are relatively more productive in agriculture than in

non-agriculture (e.g. Restuccia et al., 2003; Caselli, 2005; Gollin et al., 2007), whereas a

smaller set of studies consider explicitly an open-economy model where the driving force is

the law of comparative advantage (e.g. Matsuyama, 1992).

1

In this paper, I refine the second

stream of literature and investigate its quantitative implications for open economies.

A starting point for this paper is Matsuyama (1992), which suggests that the relationship

between agricultural productivity and industrialisation in an open economy could be negative

because a low agricultural productivity, ceteris paribus, implies a comparative advantage in

non-agricultural sector.

2

These countries with a comparative advantage in non-agricultural

sector specialise non-agricultural goods so that labour force moves away from agricultural

sector. In contrast, a high agricultural productivity relative to the rest of the world (and hence

a comparative advantage in agriculture) pushes labour force out of manufacturing to agricul-

tural sector due to international trade, through which a country exports agricultural products

in response to importing manufacturing products.

However, Matsuyama’s (1992) model lacks solid empirical support. For instance, McMillan

and Rodrik (2011) show that the labour productivity in agricultural sector relative to non-agricul-

The author thanks Jin-hyuk Kim, James Markusen, Keith Maskus, Eun-Sun Park, Yeo Joon Yoon, Seong

Man Moon and Murat Iyigun for their insightful comments. The paper has also benefited from comments

by seminar participants in the 2012 Midwest Macroeconomics Meeting, Kyunghee University, and KIEP

seminar. The errors remain the author’s responsibility.

1

In these strands of the closed-economy literature, a positive link between agricultural productivity and

economic development can be supported by two reasons, which are associated with the fact that the

income elasticity of demand for agricultural good is less then unity (i.e. Stone–Geary utility function).

First, a low agricultural productivity can impede industrialisation by delaying onset of industrialisation.

Second, an increase in agricultural productivity, other things being equal, pushes labour force out of

agricultural sector to non-agricultural sector.

2

Matsuyama’s argument was based on historical evidence from Belgium and the Netherlands (Mokyr,

1974) as well as New England and the South (Field, 1978; Wright, 1979).

©2016 John Wiley & Sons Ltd 743

The World Economy (2017)

doi: 10.1111/twec.12398

The World Economy

tural sector exhibits a U-shaped pattern as the economy develops. Specifically, they show that

data from India exhibit the downward sloping part of the curve, whereas French data show the

upward sloping part. This paper offers a revised Matsuyama model to show that the predictions

of the open-economy model are more complicated; however, Matsuyama’s theoretical results

continue to hold qualitatively, and more importantly, I show through the model calibration that

the quantitative effects of comparative advantage and learning-by-doing in open economy can

be substantial.

This paper considers a two-sector Ricardian model similar to the one considered by Lucas

(1988) and Matsuyama (1992), where due to learning-by-doing technology the manufacturing

sector’s total factor productivity (TFP) grows with its scale of production. Following Gollin

et al. (2007) and Restuccia et al. (2003), I allow agricultural sector to be modernised using

manufacturing output as an intermediate input to its production. Different from these articles,

however, I focus on the role of comparative advantage in an open economy as the main driv-

ing force behind structural transformation. Indeed, counterfactual analysis in this paper show s

that the role of comparative advantage in an open economy is quantitatively important for

explaining the different patterns of structural change across countries. Besides, the open-

economy aspects of structural change can complement the dominant view discussed in the

literature that the poor countries are relatively unproductive in agriculture, so that it captures

the real world more completely.

In Section 2, to motivate our study, I construct an index of revealed comparative advantage

(RCA) in agriculture as well as manufacturing, as proposed by Balassa (1965) and show that

there is a strong negative (positive) relationship between RCA in agriculture (manufacturing)

and the growth rate of manufacturing labour share in a cross section of countries. Although

this correlation is only suggestive, it is consistent with the theoretical result in Matsuyama

(1992) that there could be a negative relationship between agricultural productivity and speed

of industrialisation.

In Section 3, I present the theoretical model, show conditions for industrialisation to begin

in a subsistence economy and characterise the equilibrium and the time path of an open econ-

omy’s structural transformation. In Section 4, I calibrate the model to match the US and the

UK data on agriculture’s labour share. The calibrated model, despite being simple, accounts

fairly well for the long-run patterns, such as the evolution of agriculture’s share of total

output in the USA and the UK.

Using the 50 largest countries’ labour share data from 1961 to 2003 as a benchmark, I then

quantitatively analyse the effect of comparative advantage on structural changes. The numeri-

cal experiments show that a mere 4 per cent initial comparative advantage in the manufactur-

ing sector leads to a threefold difference in agriculture’s employment share and accounts for

roughly 15 per cent difference in the aggregate output during the sample period. Further, a 5

per cent increase in the speed of learning-by-doing can account for a factor of three difference

in agriculture’s employment share and about 18 per cent increase in aggregate output, while

learning has little effect in a closed economy.

This paper is related to the strand of literature that numerically investigates the effect of

trade on industrialisation (e.g. Echevarria, 1995; Stokey, 2001; Teignier, 2012). For instance,

Teignier simulates a neoclassical growth model to the structural transformation of the USA,

the UK and South Korea and estimates the effect of trade on consumer welfare. The differ-

ence is that in my model, productivities evolve endogenously and I focus on the effect of

comparative advantage as well as learning-by-doing in an open economy, while in the above

©2016 John Wiley & Sons Ltd

744 U. WHANG