APPROPRIATE TECHNOLOGY AND INCOME DIFFERENCES

• Date: 01 August 2016
• Published date: 01 August 2016
• Author: Dozie Okoye
• DOI: http://doi.org/10.1111/iere.12182

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 3, August 2016

APPROPRIATE TECHNOLOGY AND INCOME DIFFERENCES∗

BYDOZIE OKOYE1

Dalhousie University, Canada

This article studies the relative productivity of skilled to unskilled workers across countries. Relative productivities

are broken down into the human capital embodied in skilled workers and relative physical productivities (reflecting

production techniques). I find that skilled workers from poorer countries embody less human capital and are also

relatively less physically productive. Furthermore, results show that production techniques are inappropriate for most

low-income countries, and these countries experience large increases in GDP per capita by increasing the relative

physical productivity of skilled to unskilled workers. This suggests that there are significant barriers to the adoption of

skill-complementary technologies.

1. INTRODUCTION

An important problem in development economics, and economic growth, is understanding

how economies use available factors of production. The conventional view is that low-income

countries use available factors unproductively compared to richer countries. Recent studies have

found significant productivity differences between sectors in low-income countries, and there

is an emerging view that low-income countries are not unproductive at everything. Examples

include productivity differences between agricultural and nonagricultural sectors (Restuccia

et al., 2008; Gollin et al., 2011), differences in productivity by sectoral skill intensity (Acemoglu

and Zilibotti, 2001; Ciccone and Papaioannou, 2009), and sectoral differences in productivity

based on tradability of the final good (Herrendorf and Valentinyi, 2012).

Looking across factors of production, Caselli and Coleman (2006) find that in comparison

to high-income countries, low-income countries use unskilled workers relatively more produc-

tively than the skilled. In fact, they find that that low-income countries use unskilled workers

absolutely more productively than richer countries under their preferred set of parameters.

However, the methodology in Caselli and Coleman (2006) cannot distinguish the hypothesis

that skilled workers in low-income countries are less productive because they have less em-

bodied human capital from the hypothesis that skilled workers are less productive because

low-income countries adopt technologies that are (more) complementary with unskilled work-

ers.2For example, are doctors in the United States more productive than doctors in Liberia

because they are generally better trained (higher embodied human capital), or because they

∗Manuscript received June 2014; revised April 2015.

1I am grateful to Elizabeth Caucutt, Igor Livshits, James B. Davies, the editor, and two anonymous referees for

insightful comments and suggestions that greatly improved the article. I also wish to thank participants at the UWO

Macro Lunch and Dalhousie University’s Macro and Economic Development group for remarks on earlier drafts.

Comments from participants at the 2013 Canadian Economics Association meetings are also acknowledged. All errors

are mine.

Please address correspondence to: Dozie Okoye, Department of Economics, Dalhousie University, 6214 University

Ave., Halifax, NS, B3H 4R2, Canada. Phone: +902-494-4453. E-mail: cokoye@dal.ca.

2They fully acknowledge this possibility and use information on the elasticity of human capital to measures of

schooling inputs to argue that human capital quality differences are small and can only explain a small fraction of the

differences in physical productivities. However, recent studies, using different data to back out human capital quality,

have found considerable cross-country differences in the quality of skilled labor. In many cases, differences in the

quality of skilled labor are as large as differences in the quantity of skilled labor (Manuelli and Seshadri, 2005; Erosa

et al., 2010; Schoellman, 2012).

955

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

956 OKOYE

have access to better equipment such as MRI machines (higher physical productivities), or

both?3

To answer the question mentioned above, I first extend the methodology in Caselli and

Coleman (2006) and explain how cross-country differences in the productivity of skilled workers

(relative to the unskilled) can be decomposed into differences in the amount of human capital

embodied in skilled workers and differences in production techniques, or what I call physical

productivities. This is done using estimates of skill premiums for immigrants from different

countries living in the United States and estimates of skill premiums across countries. The

intuition is that since all skilled workers in the United States work with the same technologies,

any differences in skill premiums by country of origin within the United States must be due to

differences in embodied human capital.4Further, controlling for cross-country differences in

the supply of skilled workers, any differences in skill premiums for immigrants in the United

States and those left behind in their home countries must be due to differences in physical

productivities (production techniques).

I find that (compared to richer countries) skilled workers in low-income countries have

significantly less embodied human capital. The physical productivity of skilled workers is also

higher in high-income countries, implying that they are used more productively relative to

unskilled workers. This result holds for various plausible parameterization of the production

function. Compared to the productivity of skilled workers relative to the unskilled in Caselli

and Coleman (2006), controlling for embodied human capital means that differences in the

physical productivity of skilled workers are smaller. However, large differences in physical

productivities still remain because differences in embodied human capital are small, compared

to differences in physical productivities. For example, the ratio of embodied human capital in

Ghanaian skilled workers relative to their American counterparts is about 3.6, which is small

compared to the 50-fold difference in the physical productivity of skilled relative to unskilled

workers.

I then investigate the appropriateness of physical productivities. For example, are skilled

workers relatively unproductive in Ghana because Ghanaian firms appropriately choose tech-

nologies which make abundant unskilled workers more productive? I argue that the estimates

imply that this is unlikely: In Ghana, the estimates imply that skilled workers are 50 times less

physically productive compared to unskilled workers, and in Venezuela they are 100 times less

productive. However, the data show that there are just as many skilled and unskilled workers

in Ghana and Venezuela. It is unlikely that unskilled workers being 50 times as physically

productive as skilled workers is appropriate, given that skilled workers are just as numerous,

possess more human capital, and are substitutable with unskilled workers.

To formally investigate, I compare the estimated relative physical productivity of skilled

workers to what they would be if technologies were chosen by profit-maximizing firms in each

country, following models of appropriate technology.5For most of the 49 countries in the data

set, the estimated physical productivities of skilled workers are significantly lower than what

is optimal. This is true under various definitions of skilled–unskilled labor and values of the

skilled–unskilled labor elasticity of substitution. Furthermore, the distance between estimated

and optimal physical productivity of skilled workers is decreasing with GDP-per-capita. For

3This is an important distinction because the source of the higher productivity of skilled workers has different policy

implications. If doctors in high-income countries are primarily more productive because they are better trained, this

would imply important differences in the quality of doctors across countries. On the other hand, if doctors are more

productive in high-income countries because of access to better medical equipment, it would imply significant barriers

to technology adoption.

4Although immigrants are positively selected on schooling, Schoellman (2012) provides evidence that the return to

schooling for immigrants is robust to selection problems. Similar patterns are found when the sample is restricted to

refugees and asylum seekers who are a less-selected group of immigrants.

5The shape of the technology frontier available to firms is assumed to be the same across countries, but the height of

the frontier or number of available technologies is allowed to vary by country. This is in contrast to Caselli and Coleman

(2006), who estimate the technology frontier assuming that firms in every country choose technologies appropriately,

but also allow the shape and the height to vary across countries.

TECHNOLOGY AND INCOME DIFFERENCES 957

richer countries such as South Korea, Japan, Israel, the Netherlands, and Australia, we cannot

reject that they use skilled and unskilled workers appropriately. For other countries, such as

Thailand, India, Venezuela, Ghana, and Kenya, the estimated physical productivity of skilled

workers is four times less than what is appropriate.

It is important to see that the finding that the estimated mix of skilled–unskilled physi-

cal productivities is inappropriate is independent of the exact shape of the world technology

frontier. This is because many countries could increase output by using the estimated mix of

skilled–unskilled physical productivity of other countries in the data. For example, Ghana could

increase GDP per capita by a factor of 2.5 simply by using the mix of skilled–unskilled physical

productivities used by Ecuador or Greece, leaving all else the same. This implies that with-

out knowing the country-specific technology frontier, there is a possible mix of technologies

that would increase income in many low-income countries by making skilled workers more

productive relative to the unskilled.

This finding also differs from regular notions of appropriate technology, which argue that poor

countries use technologies that are appropriate for a high-skilled workforce, but inappropriate

for their relatively unskilled workforce. The result here is in fact the opposite; controlling for

the human capital embodied in skilled workers, low-income countries tend to use technologies

that make skilled workers very unproductive and unskilled workers too productive. This sug-

gests significant income gains by adopting more technologies that make skilled workers more

productive and unskilled workers less productive.

To get a sense of potential income gains from using more appropriate and skill-complementary

technologies, I compare income per capita with estimated physical productivities to income per

capita under the optimal physical productivity of skilled workers. Note that in this exercise,

all factors of production are held constant, and only the relative productivity of skilled and

unskilled workers is changing. Under the preferred set of parameters, the average country in

the data increases its income per capita by a factor of 2 from using its appropriate technology

and increasing the relative physical productivity of skilled workers.

There is some variation in income gains from adopting appropriate technologies, depending

on a country’s income relative to the United States. Countries in the lowest income quartile

experience a sevenfold increase in income, because they are farthest away from their optimal

mix of skilled–unskilled physical productivities. More than 50% of all countries in the data set

could increase incomes by a factor of 4 from adopting the appropriate mix of skilled–unskilled

worker complementary technologies. Countries in the top income quartile only experience a

23% increase in income by adopting their appropriate technologies, and most of the gains are

driven by France and Greece.

These results could be interpreted as the result of significant barriers to technology adoption,

in a world in which new technologies are complementary with skilled workers.6As tradi-

tional technologies are complementary with unskilled workers, the adoption of newer skilled-

complementary technologies are readily blocked by vested interests, rendering skilled workers

relatively less productive than they could be.7A report by McKinsey (2001) documents the

prevalent nonadoption of more productive technologies in many Indian sectors. A recent ex-

ample is the resistance to Walmart’s entry into the Indian retail market by small-scale retailers

(see the Bloomberg news report by Pradhan and MacAskill, 2011).

The inefficient use of skilled workers could also be a result of financial frictions, which may

prevent firms from adopting modern technologies (Greenwood and Jovanovic, 1990; Buera

and Shin, 2010; Greenwood et al., 2010). In particular, Midrigan and Xu (2014) find that the

effect of financial frictions on productivity is primarily due to its distortionary effects on the

decision to enter the modern sector and adopt modern technologies. Furthermore, the lack

6See Acemoglu (2002), Berman et al. (1998), and Ciccone and Papaioannou (2009) for evidence that recent tech-

nologies have been skilled biased.

7For examples of how vested interests can block the adoption of new technologies, see Bridgman et al. (2007) and

Bellettini and Ottaviano (2005).