Economic Diversification and Human Development in South Asia

Abstract

The paper examines the effect of diversification on social welfare in South Asia using the macroeconomic data for the period 1996–2011, with export diversification as a proxy of economic diversification. In this paper, three types of diversification are assessed (i.e. related, unrelated, and overall variety). While unrelated variety and overall variety show increasing trend over the years, related variety seems to show a non-linear inverted U-shaped curve. Estimation results reveal that all three types of variety have a positive and significant relationship with human development in South Asia, which shows that diversification is beneficial for human development in South Asian countries. It is also found that the existing level of human capital significantly moderates the relationship between related variety and social welfare.

Keywords

Introduction

Literature of economic diversification shows that a diversified economic structure is beneficial for growth, as it allows the system to absorb external economic shocks as well as results in innovation through knowledge spillovers across sectors. Differentiating between related and unrelated variety, Frenken et al. (2007) show that related variety has immediate positive relationship with growth (measured by employment growth), while unrelated variety affects growth after a significant lag. Similar conclusions are found by Saviotti and Frenken (2008) using data for exports. Their findings illustrate that diversification is beneficial for growth. However, when it comes to human development, it is unclear whether diversification improves or deteriorates human development (Hartmann and Pyka, 2013). The ambiguity of the relationship arises from the fact that diversification strategies may result in obsolescence of existing sectors, which may result in unemployment at least in the short run. Moreover, GDP per-capita is used as a proxy for social welfare in neoclassical economics with an assumption that higher per-capita income growth also results in social welfare. However, in the presence of income inequality, per-capita income at aggregate level is likely to provide a misleading picture of the welfare status. Moreover, analysis of per-capita income in neoclassical tradition considers factors such as labor inputs, capital inputs, and technological change as the determinants of per-capita income growth; however, these factors are mostly production oriented and not necessarily related to the improvement of social welfare of the society. Due to the above-mentioned caveats, an analysis of growth may not necessarily give the same implications for human development. Similarly, in the context of economic diversification, a positive relationship between diversification and economic growth may not necessarily imply the same for human development. Despite the contradiction mentioned above, only few studies attempted to conceptually establish the relationship between the economic diversification and human development (Hartmann and Pyka, 2013).

This paper, therefore, aims to examine the effect of diversification on social welfare in South Asia (i.e. Bangladesh, India, Maldives, Nepal, Pakistan, and Sri Lanka). Since most developing economies specialize in primary goods and are increasingly following diversification strategies to facilitate the transition toward manufacturing sector, the implications of diversification on social welfare are especially important for these economies. The choice of South Asian region for the analysis is based on the relatively similar growth trajectories of these countries. The time period covered in the empirical analysis ranges from 1996 to 2011. This study takes its premise from Ali and Cantner (2015), which covers 20 European countries. This study contributes to the literature by zooming in to the South Asian economies, which allows for a detailed analysis of how economic composition affects human development in developing countries.

Following the current section, the section “Economic growth, diversification, and human development” describes the interrelationship among growth, human development and economic diversification, followed by the section “Export composition in South Asian countries” which provides descriptive analysis of diversification trends in South Asian countries, the section “Data and methodology” discusses data and methodology, “Econometric model” develops and describes the econometric model, “Results and discussion” discusses the key findings of the research, and the final section concludes.

Economic growth, diversification, and human development

Neoclassical theory of growth represents economic growth in the form of output per-capita and defines it as a function of capital, labor, and total factor productivity (Solow, 1956; Swan, 1956). An important feature of the neoclassical formulation of growth processes is the assumption that technological progress that shapes total factor productivity is determined outside the economic system (i.e. it is exogenous to economic activities). Contrary to the neoclassical growth theory, endogenous growth theory posits that technological progress is determined within the economic system primarily through the factors related to accumulation and diffusion of knowledge. In particular, it postulates that innovations, whether they are product or process innovations, are derived from technological progress. Such innovations are generally the outcome of economic activities resulting from, for example, learning-by-doing strategies. Firms learn from their experiences as well as from the progress of their competitors, resulting in a higher incentive to innovate in order to gain or sustain market shares. By bringing technological progress and innovation to the center of the discussion, endogenous growth theory also brings factors that influence innovation such as R&D expenditures, taxes, subsidies, trade, human capital, and property rights to the center of the discussion on the growth policy.

Endogenous growth theories based on innovation can be divided into two main strands. One strand describes the rate of increase in total factor productivity as a function of product variety, where new varieties are important regardless of their improvements in quality as compared to their predecessors (Romer, 1990). The second strand of theories, generally termed as “Schumpeterian growth theories,” relate to the core ideas of the first one, however, with an explicit focus on quality improvements through which old varieties are outcompeted by the new ones (Aghion and Howitt, 1992; Grossman and Helpman, 1991; Schumpeter, 1942). The process through which old products are replaced by better products is generally termed as “creative destruction” in the Schumpeterian growth literature (Schumpeter, 1942). The general premise of the endogenous growth theory is that the returns from the production should be invested in innovation; for example, in the form of R&D expenditures, in order to innovate and grow rapidly. The process of creative destruction shapes the structure of product variety in an economy through the forces of competition and innovation.

Literature on the relationship between economic structure and economic growth shows that both specialization and diversification strategies can be employed to promote growth. The importance of diversity in development process was introduced by Jacobs (1969), which argues that a variety of activities, instead of concentration, fosters innovation through recombination of ideas across sectors. Jacobs’ view challenges the classical view of efficiency, which is based on the principle of comparative advantage and specialization. The concept of so-called Marshal–Arrow–Romer externalities proposes that specialization results in “efficiency-gains” and fosters innovation through learning-by-doing principles. Recent studies have also shown that diversification can be disentangled into “related” and “unrelated” diversification with significant differences in implications for growth and innovation (Castaldi et al., 2015; Frenken et al., 2007). Even though there seems to be an apparent distinction between the concepts of specialization and diversification, they may not be mutually exclusive. First, in the context of the level of sectoral disaggregation at which the diversification/specialization is measured, a diversified economic structure at a higher degree of aggregation could also have high levels of specialization at lower levels of aggregation, resulting in the coexistence of both mechanisms in an economy at the same time (Hartmann and Pyka, 2013). Second, countries tend to diversify at early stages of development and re-specialize after reaching a threshold level (Imbs and Wacziarg, 2003). It is, therefore, difficult to conclude whether one strategy is better than the other for growth.

Theorization on the relationship between growth and diversification/specialization can be traced back to Adam Smith’s ideas on division of labor force, followed by the Schumpeterian notion of “creative destruction,” where new sectors take over the older ones as a result of innovation. The discussion further stressed that sustainable economic development requires constant internal transformations. Such transformations can be subtle where, for example, little changes in the methods of production would change the ways in which firms operate or they could result in structural changes across the economy by replacing old economic activities with the new ones (Saviotti and Frenken, 2008). In the process of structural change, new sectors create employment opportunities, while the obsolescence of old sectors creates unemployment, at least in the short run (Smith and Gibson, 1988). Moreover, literature on economic diversification also shows that the diversified economic structure is more beneficial for growth than the specialized one, as it allows the system to absorb external economic shocks and leads to technological innovation through knowledge spillovers (Frenken et al., 2007).

Types of diversification: related and unrelated

Recent literature on diversification/variety distinguishes between related and unrelated variety. When diversification is achieved in cognitively related sectors, it is termed as “related variety.” On the contrary, when diversification is achieved in sectors that are cognitively unrelated to each other then it is termed as “unrelated variety.” Empirical evidence has shown that both types are important for the economy and they both bring different types of benefits. Related variety, for instance, induces the knowledge transfer across sectors through spinoff dynamics, labor mobility, and network formation at the regional level (Boschma, 2009; Cantner and Meder, 2009; Frenken et al., 2007). According the evolutionary economic geography, the related varieties of a region condition the scope for innovation (Boschma and Frenken, 2015). The higher level of variety leads to an increase in the inherent capacity for vertical integration that leads to higher levels of innovation (Cainelli and Iacobucci, 2012; Glaeser et al., 1992). On the contrary, since unrelated variety is achieved in cognitively unrelated sectors, the frequency of knowledge spillovers is likely to be lower than related variety. This may be mainly because the unrelated variety, by means of distinct knowledge paths, diverges to technological “breakthroughs.” The unrelated building-blocks connect in a manner that enables the innovation capabilities to build new structural and technological trajectories to bring in the efficiency and improvement in the existing system that not only increases the resistance against economic shocks but also increases the potential for radical innovation (Castaldi et al., 2015; Desrochers and Leppälä, 2011).

Human development approach

The conceptualization of human development in general refers to the concept developed by Amartya Sen in which he criticizes the traditional approach in defining and conceptualizing social welfare by combining the aspects of wealth with the factors related to well-being (Sen, 1983). The human development approach (also known as the capability approach) realizes the importance of expansion of goods and services; however, it proposes to consider the improvements in quality of life and increase in economic opportunities in the analysis of traditional welfare. The underlying argument behind the human development approach, as an alternative to the neoclassical welfare concept, is that a mere count of commodities available in the economy may not necessarily reflect the real welfare state of the economy unless we consider how well the economic agents are able to perform with the available commodities. The capability approach also proposes to consider factors related to rights and economic freedom in the utility analysis in order to include the measure of “ability to perform” in the welfare analysis. Similarly, improvements in education levels as well as health status are as important as the improvement in per-capita income. Therefore, education and health should be given the same weight as per-capita income in the welfare analysis. One of the strongest features of Sen’s capability approach is that it is flexible and does not restrict itself to identified concepts. The inclusion and exclusion of the concepts, as well as their weights, are left to the judgment of the researchers. However, critics of the capability approach consider its flexibility as a degree of incompleteness (i.e. it lacks the comprehensive identification of capabilities) (Clark, 2005). However, despite the limitations of the approach, it provides significant value additions to the traditional view on welfare economics.

In addition to the conceptualization of human development approach, the measurement of human development has also been discussed for decades. The traditional measures of well-being, such as GDP or GNP per-capita, have been criticized by many studies in the past, because as growth accounting generally deals with monetary values, it ignores the distribution of income and it equates economic “goods” with economic “bads” (Hicks and Streeten, 1979). As an alternative, the so-called Human Development Index (HDI) was first introduced in the Human Development Report (HDR) of the United Nations Development Programme in 1990 (United Nations Development Programme, 1990). The HDR 1990 introduced the terminology of “Human Development” as a measure of well-being and presented a wide range of country-level indicators to assess the status of well-being across the world. The HDI index aims to quantify the capabilities concept of Sen by highlighting the roles of education and health, in addition to per-capita income, in the measurement of well-being with an idea that better health and education implies better capabilities. In the measurement of HDI, normalized life expectancy is used as a proxy for health status, school enrollment is used as a proxy for education level and per-capita income is used as a proxy for purchasing power.

Similar to its predecessors, the HDI has also been criticized by many scholars on various grounds. One of the most common criticisms on HDI is its simplicity, which makes it vulnerable to measurement errors and makes it incomparable across countries (Stanton, 2007). However, the simplicity of HDI makes it much more useful as compared to analyzing dozens of indicators in policy making (Streeten, 1994). Moreover, as a simple indicator, HDI is more attractive for the public than any other measure, as it allows a common person to compare relative well-being of a country without a need to understand complex analytical tools.

Diversification and human development

The relationship between diversification and human development can be formulated broadly under the endogenous growth theory, especially under the innovation-based theories, where economies are shown to transform and grow with the growth in a variety of products. The conceptualization of diversity or specialization can be traced back to Adam Smith (1776) who proposed division of labor based on their respective abilities for better efficiency and growth. From a general perspective, division of labor may sound like a policy that would lead to specialization; however, this is not necessarily true. The structural hierarchy of an industrial classification allows for the coexistence of diversity and specialization at the same time at different levels of industrial classification. Division of labor has important implications for efficiency, learning-by-doing, improvement in skills and gains in aggregate productivity, where the resulting overall structure of the economy may have a high level of diversification or specialization. With respect to the change in overall industrial structure, innovation-based endogenous growth theory also presents the concept of creative destruction, where new sectors result in the obsolescence of the older ones as a result of technological progress and structural change (Schumpeter, 1912). Whether the transformations and structural changes result in specialization or diversification is a complex question; however, technological progress and diversification are generally identified as the key drivers of sustainable and rapid growth (Saviotti, 1996).

Given the distinct nature of related and unrelated types of variety, many studies have analyzed the relationship between (un)related variety and economic growth at mostly a regional level. However, it is unclear whether diversification is also beneficial for social welfare. The aspect of social well-being is generally ignored in the growth-enhancing and development-related policies with the assumption that an increase in per-capita income implies social well-being (Hartmann and Pyka, 2013). We know from the recent literature that innovativeness has positive and significant relationship with income inequality (Aghion, 2002; Aghion et al., 2015), which is detrimental to social welfare. Since diversification, especially related variety, appears to have a strong relationship with innovation, it may have a negative effect on social welfare due to the relationship between innovation and income inequality. In the case of unrelated variety, if the introduction of new sectors results in the obsolescence of existing skills and resources, then the resulting unemployment would have a negative effect on social welfare, at least in the short run. However, since competition breeds innovation and vice versa, innovation may also create economic opportunities and foster innovation, which may increase employment opportunities, resulting in social welfare gains. Diversification also facilitaties the distribution of power and reduces social inequality.¹ The heterogeneous effects of diversification on human development raise questions regarding the impact of economic diversification on human development.

Diversification strategy requires technological and productive capabilities to produce a wide range of products and services (Hidalgo and Hausmann, 2009; Hidalgo et al., 2007). Human capital is one of the most important capabilities that shapes the direction of economic development. The use of knowledge through external sources depends on the level of absorptive capacity, without which new knowledge (especially the tacit component) cannot be decoded (Cohen and Levinthal, 1990). Higher levels of human capital would allow the economy to not only increase diversification but also reap the benefits of diversification in terms of employment, entrepreneurship, and innovation. In an event of significant structural change, higher level and quality of human capital would allow the economy to smoothly move from one development path to another, resulting in better levels of well-being.

From the discussion in previous sections it becomes clear that different types of variety – related, unrelated, or overall – have different effects on the economy. Therefore, we hypothesize that all three types of diversification have a positive relationship with human development. We also hypothesize that human capital moderates the relationship between diversification and human development for all three types of variety.

Export composition in South Asian countries

Before moving to the empirical analysis of the main research questions, this section analyses the trends and composition of exports in six South Asian countries in the sample. The aim of the section is to paint a rough picture of the sectoral composition of exports in the sample countries.

High reliance on primary products has made countries suffer due to unpredicted climate changes and natural disasters. Historically, countries that managed to increase their share of manufacturing exports in the total exports grew at a rapid pace. In the context of South Asia, data from World Development Indicators (WDI) show that exports of goods and services as a percentage of GDP have increased from 12.1% in 1995 to 22.72% in 2011, which is similar to the aggregate world trend where exports as a percentage of GDP increased from 21.57% in 1995 to 29.69% in 2011. Trends for individual countries in South Asia, however, show mixed trends over time, where exports as a percentage of GDP have increased for India, Bangladesh, and Maldives, while it has decreased for Pakistan, Sri Lanka, and Nepal. Bangladesh and India have particularly shown rapid increase in exports over the years, where the exports as a percentage of GDP almost doubled for both countries between 1995 and 2011. The Maldives have unusually high levels of exports per GDP standing at 105.76% in 2011, which is three times higher than the world export to GDP ratio and almost five times higher than South Asian aggregate. While exports to GDP ratio for Pakistan decreased by approximately 4 percentage points, it reduced to half for Nepal and Sri Lanka declining from 24.97% in 1995 to 8.9% in 2011 for Nepal and 35.59% in 1995 to 19.55% in 2010. The mixed trends over the years may imply that South Asian countries are competing for the same markets, where each country bids for the international market share for the similar products with competitive prices and quality as well as variety, resulting in an increase in exports for one country, while a decrease for the other. It could also mean that the countries differ significantly from each other in terms of their exports portfolio, where a decrease in demand for textiles, for example, would only affect some countries and not the others, resulting in a rapid decline in exports for the former and unaffected exports growth for the latter. Later in this section it will be shown that most South Asian economies are heavily reliant on few commodity sectors, which is most likely the reason for the noticeable decline in exports over the years for these countries.

In order to estimate the reliance of South Asian economies on primary products, share of manufactured exports in total merchandise exports is analyzed. The share of manufactured goods in merchandize exports for South Asian countries as well as the world declined from around 75% in 1995 to 66% in 2011, showing the overall decrease in the share of manufactured products in the total exports all over the world on average. In South Asia, only Bangladesh showed an increase in manufactured exports as a percentage of GDP from 85% in 1995 to 92% in 2011 due to the boom in manufactured textile industry in Bangladesh during the last decade. The Maldives showed a rapid decline from 25% manufactured products in total merchandise exports to 0.1% in 2011 due to significantly high levels of fish-related exports in Maldives in the recent years (as it will be shown later in this section). All the other four countries showed on an average 10% decline between 1995 and 2011. Another indicator to measure the reliance on primary products is the share of agricultural raw materials in total merchandise exports. Data show that India, Nepal, and Sri Lanka have increased their raw agricultural exports while for Pakistan, Maldives, and Bangladesh it has decreased over time. The most significant increase is noted for Nepal, where the share increased from 1.11% to 3.24%, while the most significant decrease is noted for Maldives, where it declined from 0.73% to 0.0004%.

In order to delve deeper into the exports statistics for South Asian countries, Table 1 compares the top 3 two-digit sectoral shares of exports for each country in 1995 and 2013. The first observation from the table is that the total number of exports include a high share of textile in South Asian countries except for the Maldives. It is also evident that large share of exports comprise primary goods such as fish, textile yarn, minerals, cocoa, vegetable and fruits, while shares of technologically intensive manufacturing products are negligible. In terms of the dominance of sectoral shares, Table 1 shows that both Bangladesh and Maldives are highly reliant on a single two-digit industry for their exports. In particular, 66% of the exports of Bangladesh comprised articles of apparels and related accessories (SITC code 84) in 1995, which increased to 80% in 2013. Similarly, the exports of Maldives are concentrated in fish and related products (SITC code 03), for which sectoral share increased from 66% in 1995 to 86% in 2013. The numbers point toward a high degree of specialization in few industries, which implies that countries have a comparative advantage in the said products and also some degree of path dependence. However, a high reliance on a few sectors also makes them prone to sector-specific shocks, which could significantly affect the export performance of these countries. A quick comparison between the top 3 sectoral shares in 1995 and 2013 shows that South Asian countries have more or less the same sectors as their top 3 exporters, which hints toward the lack of structural transformation in the last 19 years. India appears to be the best performer in terms of sectoral diversification of exports, with the highest exporting sector, Non-metallic Manufactures (SITC code 66), taking only 16.4% of the total exports in 1995. In 2013, Petroleum exports (SITC code 33) overtook Non-metallic Manufactures as the highest exporting sector, with the highest industry share standing at 21%. A high level of export diversification in India implies that if one industry is affected by external shocks, it will only affect a small share of exports in India as other sectors are less likely to be affected by sector-specific shock.

Table 1.

Top 3 two-digit industries in total exports.

	Share of top 3 exports in total exports
	1995		2013
Bangladesh	Apparel and accessories (84)	65.6%	Apparel and accessories (84)	79.7%
	Textile yarn (65)	10.3%	Textile yarn (65)	6.8%
	Fish (03)	8.1%	Fish (03)	3.1%
India	Non-metallic manufactures (66)	16.4%	Petroleum (33)	20.8%
	Textile yarn (65)	14.0%	Non-metallic manufactures (66)	10.0%
	Apparel and accessories (84)	13.2%	Textile yarn (65)	5.7%
Maldives	Fish (03)	66.4%	Fish (03)	86.0%
	Apparel and accessories (84)	27.6%	Natural Gas (34)	5.1%
	Animal Feed (08)	2.7%	Metal scrap (28)	1.0%
Nepal	Textile yarn (65)	50.0%	Textile yarn (65)	31.5%
	Apparel and accessories (84)	33.6%	Apparel and accessories (84)	12.3%
	Leather (61)	3.5%	Iron and steel (67)	11.8%
Pakistan	Textile yarn (65)	52.3%	Textile yarn (65)	37.2%
	Apparel and accessories (84)	19.8%	Apparel and accessories (84)	18.1%
	Cereals (04)	5.9%	Cereals (04)	9.8%
Sri Lanka	Apparel and accessories (84)	48.6%	Apparel and accessories (84)	45.1%
	Coffee (07)	11.8%	Coffee (07)	18.8%
	Non-metallic manufactures (66)	7.7%	Rubber (62)	6.9%

Note: two-digit SITC codes in parenthesis.

In the context of Pakistan, the share of textiles in exports in 1995 stood at 72%, which decreased to 55% in 2013. The composition of exports in Pakistan implies that, even though there is some degree of diversification present in the export portfolio, the whole export sector is very much reliant on textile-related products. Any textile-specific shock is likely to significantly slow down the export performance of Pakistan, unless shares of exports in other sectors are increased. Nepal has significantly reduced its reliance on textile-related products from 84% in 1995 to 44% in 2013, showing a sign of structural transformation, while for Sri Lanka apparels and accessories and coffee jointly represented approximately 60% of total exports in both 1995 and 2013, showing some level of path dependency.

The set of figures under Figure 1 show the evolution of variety at different levels of industrial classification. The first figure in Figure 1 shows the trend of diversification at a three-digit industrial classification (overall variety), the second sub-figure shows the trend of variety at a three-digit level within a two-digit level (related variety) and the third sub-figure shows variety at a two-digit level (unrelated variety). Since we have data for 255 three-digit industries, the range of the OV index can therefore be defined as from 0 to log₂(255) = 8. Similarly, RV ranges from 0 to log₂(255)-log₂(64) = 7.99-6 =1.99 and UV ranges from 0 to log₂(64) = 6. Trends of overall variety show that India had highest overall variety in 1996, with marginal decline in 2011. The highest relative increase in overall variety during the period of analysis is observed in Nepal, which took it from 4^th place to 1st among South Asian countries in the sample. Pakistan also experienced an increase in overall variety over time; however, its relative position among South Asian countries remained the same.

Figure 1.

Trends of overall, related and unrelated variety in South Asia.

In terms of related variety, Pakistan had the highest related variety in 1996, however, it decreased over time. Only Bangladesh and Nepal showed an increase in related variety over the course of 16 years. Nepal increased its relative position from the lowest to the third highest related variety in South Asia. In the case of the Maldives, the related variety has rapidly declined over time. The stats for the Maldives also hint toward the relative specialization within the fishing industry, as fishing stands at 86% of the total exports in 2013.

Comparing relative positions in terms of unrelated variety, we can see that apart from Bangladesh, unrelated variety has increased in the other five South Asian countries. India had the highest level of unrelated variety in 1996 and maintained its position 16 years later as well. As we saw in Table 1, the composition of exports in India is quite diversified, with many sectors sharing a small proportion of overall export share. A high unrelated variety shows that Indian exports are much more diversified than its counterparts in South Asia, which gives the Indian economy, at least in terms of international trade, a significant resistance to economic shocks. Pakistan has also shown an increase in unrelated variety over the years, which suggests improvement in the diversification of exports over time.

Data and methodology

The empirical analysis of the paper is based on data for six South Asian countries, namely India, Pakistan, Nepal, Bangladesh, Sri Lanka, and the Maldives while Afghanistan and Bhutan were excluded due to data unavailability. Choice of countries is made based on their relatively similar growth trajectories and similar economic and geographic conditions. Since India, Pakistan, Sri Lanka, and Bangladesh (former East Pakistan) gained their independence from British rule between 1947 and 1948, the development patterns and differences in economic structures after approximately 60 years of independence for these countries makes it an important question to be explored. The period covered in the empirical analysis is from 1996 to 2011.

Key variables of interest in this study are measures of diversification. In literature, these variables are commonly measured by entropy indices. Entropy indices, in our context, measure the degree to which exports of a country are distributed over different sectors. If all sectors contain an equal share, variety takes the maximum value and when only one sector takes all the share, then variety is zero. Related variety is measured at the three-digit level within the two-digit industrial classification, while unrelated variety is measured at the two-digit industrial classification. Related variety is used as a proxy for Jacobs’ externalities, while unrelated variety can be interpreted as a risk-spreading strategy or portfolio effect, which reduces the impact of demand shocks, both national and international, in one sector. Related variety and unrelated variety were calculated based on methodology used by Frenken et al. (2007). Descriptive statistics and correlation matrix are provided in appendix.

Econometric model

The econometric model used in this study is inspired from the one used by Suri et al. (2011) (henceforth SU). The generalized functional form of their model represents human development (Infant Mortality Shortfall Reduction) as a function of per-capita GDP (GDPpc), per-capita GDP growth (GDPpc_Growth), education attainment (Edu), average public expenditure on health (Health.Exp), average public expenditure on education (Edu.Exp), income inequality (Gini), poverty headcount (Pov), and dummies for the country groups (Dummy). Formally the functional form can be written as:

I M S R = f (G D P p c, G D P p c_G r o w t h, E d u, H e a l t h . E x p, E d u . E x p, G i n i, P o v, D u m m y)

A noticeable feature of SU model is the use of Infant Mortality Shortfall Reduction (IMSR)² as a proxy for human development instead of the traditional Human Development Index (HDI). A common critique on HDI, especially when it is used in a regression framework, is that it includes a component of per-capita income. Since the human development approach of Amartya Sen argues that the production side of the economy does not necessarily include the welfare aspects of the economy, the proposed alternative in the form of HDI index includes a component of wealth in the form of per-capita income in addition to the health- and education-related indicators in order to capture the capabilities of the economic agents. However, the inclusion of per-capita income in the construction of HDI limits the ability of the index to extract welfare effects that are not necessarily arising from the production side of the economy. This is so because in context of regression analysis, human capital, and per-capita income are likely to have a positive relationship with human development and the resulting positive coefficient might be driven by per-capita income instead of human capital. Therefore, the interpretation of the positive sign would still be in context of the production side of the economy instead of capabilities. Alternatively, the IMSR index, proposed by SU, aims to rectify the caveat of HDI by using a relative measure of welfare, which is based on infant mortality. IMSR calculates the differences in infant mortality rates between country ‘i’ and a best performing country in the world (the United States in this case). Formally:

I M S R_{i} = \frac{I M_{U S} - I M_{i}}{I M_{U S}}

where IM is infant mortality and subscripts ‘i’ and ‘US’ represent the country in question (i) and the best performing country (US). The IMSR index is a relative measure of human development, which is purely based on a health indicator with an assumption that better health status is likely to reflect better capabilities. We assume that having a higher income and a better education (the other two components of the traditional HDI) would be reflected in the improvement of IMSR: as improvement in income increases, the ability to pay for healthcare also increases, which leads educated parents to become better aware of the health needs of their children, which, in turn, reduces the infant mortality rate. A noticeable feature of this index (and the model of SU in general) is that there is no time dimension involved in the analysis (i.e. estimations in their study are based on cross-sectional data). Cross-sectional regressions for economic development suffer from two important caveats: first, estimated relationships cannot be interpreted in terms of causality, and second the theory of convergence requires yearly data to observe changes in key variables, which is not possible in a cross-sectional setting. Taking note of these caveats, we modified the SU model by transforming it into a panel-data model as shown by equation 1 below. Equation 2 is then the modified version of equation 1 with an additional diversification index, where Div can take overall variety (OV), related variety (RV), or unrelated variety (UV). We expect RV to have a positive and significant relationship with human development because increasing variety in related products would increase employment opportunities for the existing labor force. On the contrary, the relationship between UV and human development can be positive or negative depending on the context. Increase in unrelated variety would require capabilities that are not available locally, and it may take time to develop such skills. Therefore, in the short run, UV might have a negative relationship with human development, but in the long run it is expected to have a positive relationship with human development.

\begin{array}{l} l o g I M S R_{i, t} = β_{0} + β_{1} l o g {(G D P p c)}_{i, t - 1} + β_{2} Δ l o g {(G D P p c)}_{i, t} + β_{3} l o g H C_{i, t - 1} + \\ β_{4} l o g {(S S E x p)}_{i, t - 1} + + β_{5} l o g G i n i_{i, t - 1} + β_{6} l o g P o v_{i, t - 1} + δ_{i} + γ_{t} + ε_{i t} \end{array}

(1)

\begin{array}{l} l o g I M S R_{i, t} = β_{0} + β_{1} l o g {(G D P p c)}_{i, t - 1} + β_{2} Δ l o g {(G D P p c)}_{i, t} + β_{3} l o g H C_{i, t - 1} + \\ β_{4} l o g {(S S E x p)}_{i, t - 1} + + β_{5} l o g G i n i_{i, t - 1} + β_{6} l o g P o v_{i, t - 1} + β_{7} l o g D i v_{i, t - 1} + δ_{i} + γ_{t} + ε_{i t} \end{array}

(2)

where SSEXP represent social sector expenditures calculated as a sum of public health- and education-related expenditures as percentages of GDP. Health and education sector expenditures are not included separately in the model, as they are highly correlated. The symbols $δ_{i}, γ_{t} and ε_{i t}$ represent country dummies, time dummies, and error term, respectively. All variables are taken in their natural logarithms and in order to reduce the risk of endogeneity bias, all variables are lagged by one year. Since data for the Gini coefficient and poverty headcount is not available for the full sample period of 1996–2011, we excluded these variables from the model.

In a subsequent modification of Model 2, an interaction of diversification indices with human capital is included in Model 3 in order to capture the effect of absorptive capacity as proposed by Cohen and Levinthal (1990). A positive and significant coefficient for $β_{8}$ will show that better human capital moderates the relationship between diversification and human development. In other words, positive interaction would mean that when a country has higher human capital, it is better able to transform the benefits of diversification for the betterment of human development by making use of the opportunities provided by diversification strategies.

\begin{array}{l} l o g I M S R_{i, t} = β_{0} + β_{1} l o g {(G D P p c)}_{i, t - 1} + β_{2} Δ l o g {(G D P p c)}_{i, t} + β_{3} l o g H C_{i, t - 1} + \\ β_{4} l o g {(S S E x p)}_{i, t - 1} + + β_{5} l o g G i n i_{i, t - 1} + β_{6} l o g P o v_{i, t - 1} + β_{7} l o g D i v_{i, t - 1} + β_{8} l o g D i v_{i, t - 1} * \\ l o g H C_{i, t - 1} + δ_{i} + γ_{t} + ε_{i t} \end{array}

(3)

Results and discussion

Estimation results of the econometric model presented above are presented in Table 2. Based on the results of the Hausman test, we used a fixed-effect estimation method. Model I presents the estimation results without time-fixed effects. Results show that human capital, as expected, has a positive and significant relationship with human development. The relationship between human capital and human development stays positive and significant for the rest of the six specifications as well showing that the result is robust to the changes in model specifications.

Table 2.

Fixed effects regressions with and without time fixed effects: dependent variable IMSF.

	I	II	III	IV	V	VI	VII
logGDPpc_(t-1)	-0.08 (0.11)	0.09 (0.13)	0.10 (0.12)	0.38*** (0.08)	0.38*** (0.09)	0.37**	0.42*** (0.09)
logGDPpc_(t-1)	-0.08 (0.11)	0.09 (0.13)	0.10 (0.12)	0.38*** (0.08)	0.38*** (0.09)	(0.10)	0.42*** (0.09)
∆logGDPpc_(t-1)	0.11 (0.16)	0.28 (0.14)	0.26 (0.13)	0.35** (0.11)	0.35** (0.10)	0.34** (0.12)	0.28** (0.10)
logHC_(t-1)	3.90*** (0.46)	4.81** (1.38)	4.92** (1.34)	5.54*** (0.96)	5.58*** (1.01)	5.60*** (0.86)	5.25*** (0.91)
logSSExp_(t-1)	−0.23** (0.08)	−0.30 (0.17)	−0.27 (0.19)	−0.16 (0.16)	−0.18 (0.17)	−0.15 (0.17)	−0.22 (0.15)
logGovtEff_(t-1)			0.11(0.18)	0.08(0.11)	0.11(0.12)	0.02(0.11)	0.03(0.08)
logOV_(t-1)				0.32*** (0.06)		0.35*** (0.05)
logUV_(t-1)					0.16** (0.06)		0.16** (0.04)
logRV_(t-1)					0.14** (0.06)		0.31*** (0.05)
logHC_(t-1)*logOV_(t-1)						0.33(0.44)
logHC_(t-1)*logUV_(t-1)							−0.32(0.26)
logHC_(t-1)*logRV_(t-1)							0.82** (0.25)
Country Dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year Dummies	No	Yes	Yes	Yes	Yes	Yes	Yes
Constant	2.55**	1.36	1.15	−1.06**	−1.04*	−0.90	−1.24**
	(0.93)	(0.80)	(0.78)	(0.38)	(0.41)	(0.46)	(0.41)
R-squared	0.861	0.925	0.926	0.945	0.945	0.945	0.952
Adjusted R-squared	0.854	0.905	0.905	0.928	0.928	0.928	0.935
Observations	90	90	90	90	90	90	90

Note. Standard errors in parenthesis. Standard errors are clustered by countries to account for heteroscedasticity and autocorrelation. All variables involved in interactions are mean-centered to account for the multicollinearity problem.

p <0.10, ** p <0.05, *** p <0.01.

The size of the coefficient for human capital is relatively high, which shows that a 1% increase in human capital would increase human development by approximately 5%. The positive and significant coefficient for human capital supports the expectation that a better-educated population is likely to have better levels of capabilities, which would result in an improvement of human development in the long run. As far as the income-related variables are concerned, both per-capita GDP and growth of per-capita GDP do not have significant coefficients contradictory to the expectations that a higher per-capita income would also imply better human development. This result supports the argument against the use of per-capita income for measurement of welfare, as per-capita income does not seem to have a direct positive relationship with human development. Moreover, the coefficient of social sector expenditures is found to be negative and significant, which is counter-intuitive. The significance of the coefficient for social sector expenditures disappears when time-fixed effects are included in Model II onwards. The insignificance of the social sector expenditures as a determinant of human development is also intriguing because one would expect social sector expenditures to improve the standard of living in a country. One could think of the ineffectiveness of the government to allocate the expenditures effectively as a reason behind an insignificant impact of social expenditures on human development. Keeping this in view, we introduced the government effectiveness indicator in Model III with an expectation that if the insignificance of social sector expenditures is due to lack of government effectiveness, then the coefficient should show the correct sign and significance once we control of government effectiveness. Government effectiveness is proxied by the index taken from World Governance Indicators of the World Bank. The World Bank defines government effectiveness as “Government Effectiveness captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies” (Kaufmann et al., 2011). The results, however, did not show support for our expectation in this regard. Social sector expenditures remained insignificant and government effectiveness also entered the model with an insignificant sign in Models III to VII.

In Model IV we introduced our first measure of diversification (i.e. Overall Variety (OV)). The result shows that OV positively and significantly affects human development in South Asia. This result shows support for our hypothesis that overall variety has positive relationship with human development. The elasticity of OV is such that a 1% increase in OV would increase human development by 0.32%. A noticeable difference in the results of Model III and Model IV is that the coefficient for growth of per-capita GDP became positive and significant, suggesting that Model III might have been suffering from omitted variable bias.

Similar to Model IV, when Related Variety (RV) and Unrelated Variety (UV) are included in the regression³ (Model V), both per-capita GDP and growth of per-capita GDP became significant. Both RV and UV appear with a positive and significant coefficient showing that diversification is beneficial for human development. The results show support for our hypothesis regarding the positive relationship between RV and UV with human development. Sizes of the coefficients for UV and RV are approximately half of the coefficient for OV in Model IV. They show that a 1% increase in RV and UV would increase human development by 0.14% and 0.16%, respectively. In terms of differences in the size of coefficients for RV and UV, the Wald test shows that both coefficients are statistically equivalent to each other (F-value: 0.04 with p-value: 0.86). The results presented here are different from existing literature on the relationship between (un)related variety and growth on two grounds: first, both UV and RV appears to have an impact on human development with similar time lags, and, second, both types of variety have a similar size effect on human development. The result supports the argument that findings for growth determinants may not be equally relevant for determinants of social welfare.

We also hypothesize that diversification is expected to benefit the regions with better human capital. In order to test this hypothesis, interaction terms of diversification with human capital are included in models VI and VII. The results show partial support for the hypothesis, and this support suggests that the effect of RV on human development is stronger for the countries with higher levels of human capital. A noticeable feature here is that the coefficient of the main effect of RV increases substantially after the inclusion of the interaction term showing that human capital indeed has a sizeable impact on the relationship between RV and human development. The interactions of human capital with OV and UV, however, do not have significant coefficients. The results show that the effect of OV and UV do not depend on the level of human capital and that regions with higher levels of human capital benefit more from related variety. The result might be driven by the complementarity among industries and stock of human capital, especially in the context of RV and UV. Since RV requires similar type of capabilities, human capital for one industry might be useful to another related industry; hence, we may observe a positive moderating role of human capital in the case of RV. On the contrary, an increase in UV may require locally scarce capabilities; therefore, the local stock of human capital may not moderate the relationship between UV and human development.

Conclusion

In this paper we study and compare the trends of export diversification among South Asian countries. Diversification is differentiated between related, unrelated, and overall variety. Moreover, we evaluate whether (and which type of) diversification is beneficial for human development. The results point toward the importance of increasing economic opportunities, as well as increasing immunity to external shocks by increasing diversification. India has the highest value for unrelated variety among South Asian countries throughout our sample period, which points toward the stability of the Indian economy and the immunity to sector-specific shocks. On the contrary, exports of Bangladesh and the Maldives are heavily reliant on one two-digit sector, which means that if the respective sectors are affected by any internal or external negative shock, their whole export structure will be affected. In the case of Pakistan, we found that unrelated variety of exports has increased, while related variety has decreased over time. The increase in unrelated variety is a positive sign; however, more than half of the total exports are still reliant on the textile sector, which needs some attention.

Using data for six South Asian countries, we found that all three types of variety have a positive relationship with human development, implying that an increase in diversification increases economic opportunities and provides employment options, which leads to better standards of living. We further evaluated the relationship between diversification and human development by testing for the interaction effect of human capital in the relationship between diversification and human development. We found that the level of human capital significantly moderates the relationship between RV and human development. The interactions of human capital with OV and UV do not have significant coefficients showing that the relationship of OV and UV with human development do not differ for countries with different levels of human capital. The results imply that industrial policy should not only consider productivity and profitability but also consider human development, as economic diversification has a positive correlation with human development. It is also shown that human capital plays an important role in the relationship between diversification and human development, which suggests that policies related to human capital have an important role to play in the sectoral composition of the products produced by a country and also in the relationship between diversification and the human development.

The results contribute to the literature by showing that the dynamics of diversification may differ for growth and welfare-related policies. In contrast to the literature on growth effects of diversification, we show that both RV and UV have an effect on human development at the same time. We also show that the size of the impact is similar for both types. Further studies on the subject should test for the interaction effects of different types of diversification with other welfare-related indicators such as employment rates, income inequality, and poverty. It could be that diversification strategies have a stronger effect on human development through some channels and not all. It could also be that different types of variety may have different effects on human development through different channels as we saw in this study in the case of the interactions with human capital.

Footnotes

Appendix

Table 4.

Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
logIMSR	96	1.539	0.850	−0.735	2.443
logGDPpc	96	7.812	0.694	6.933	9.440
logHC	96	0.666	0.221	0.358	1.151
logSSExp	96	1.442	0.300	0.943	2.094
logGEff	96	0.757	0.170	0.447	1.227
logOV	96	−2.339	0.414	−3.060	−1.802
logUV	96	0.922	0.468	−0.244	1.548
logRV	96	0.410	0.275	−0.849	0.681

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Muhammad Ali

Notes

Author Biographies

Muhammad Ali is an Assistant Professor at the Department of Economics at School of Social Sciences and Humanities, National University of Sciences and Technology, Islamabad, Pakistan. His central research interest is role of human capital in diffusion of knowledge.

Manzoor Hussain Memon is a PhD candidate at Applied Economics Research Center, University of Karachi, Karachi, Pakistan. His central research area is vulnerability of households in rural communities of Pakistan.

References

Aghion

(2002) Schumpeterian growth theory and the dynamics of income inequality. Econometrica 70(3): 855–882.

Aghion

Howitt

(1992) A model of growth through creative destruction. Econometrica 60(2): 323–351.

Aghion

Akcigit

Bergeaud

et al. (2015) Innovation and Top Income Inequality. Working Paper No. 21247. National Bureau of Economic Research. Available at: http://www.nber.org/papers/w21247

Ali

Cantner

(2015) Economic diversification and human development in Europe. Presented at the European Meeting on Applied Evolutionary Economics (EMAEE) 2015, Maastricht, Netherlands.

Boschma

(2009) Evolutionary economic geography and its implications for regional innovation policy. Papers in Evolutionary Economic Geography (PEEG) No. 0912. Utrecht University, Section of Economic Geography. Available at: http://econ.geo.uu.nl/boschma/OECD.pdf

Boschma

Frenken

(2015) Evolutionary Economic Geography. Papers in Evolutionary Economic Geography (PEEG) No. 1518. Utrecht University, Section of Economic Geography. Available at: http://econ.geo.uu.nl/peeg/peeg1518.pdf

Cainelli

Iacobucci

(2012) Agglomeration, related variety, and vertical integration. Economic Geography 88(3): 255–277.

Cantner

Meder

(2009) Regional effects on cooperative innovation activities and the related variety of regional knowledge bases. Available at: http://www.econstor.eu/handle/10419/32557

Castaldi

Frenken

Los

(2015) Related variety, unrelated variety and technological breakthroughs: an analysis of us state-level patenting. Regional Studies 49(5): 767–781.

10.

Clark

(2005) The capability approach: Its development, critiques and recent advances. Global Poverty Research Group ESRC Working Paper GPRG-WPS-032. Available at: http://economics.ouls.ox.ac.uk/14051/1/gprg-wps-032.pdf

11.

Cohen

Levinthal

(1990) Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly 35(1): 128.

12.

Desrochers

Leppälä

(2011) Opening up the ‘Jacobs Spillovers’ black box: Local diversity, creativity and the processes underlying new combinations. Journal of Economic Geography 11(5): 843–863.

13.

Frenken

Van Oort

Verburg

(2007) Related variety, unrelated variety and regional economic growth. Regional Studies 41(5): 685–697.

14.

Glaeser

Kallal

Scheinkman

et al. (1992) Growth in cities. Journal of Political Economy 100(6): 1126–1152.

15.

Grossman

Helpman

(1991) Trade, knowledge spillovers, and growth. European Economic Review 35(2–3): 517–526.

16.

Hartmann

Pyka

(2013) Innovation, economic diversification and human development FZID Discussion Papers (No. 65–2013). Available at: http://www.econstor.eu/handle/10419/69738

17.

Hicks

Streeten

(1979) Indicators of development: The search for a basic needs yardstick. World Development 7(6): 567–580.

18.

Hidalgo

Hausmann

(2009) The building blocks of economic complexity. Proceedings of the National Academy of Sciences 106(26): 10570–10575.

19.

Hidalgo

Klinger

Barabási

A-L

et al. (2007) The product space conditions the development of nations. Science 317(5837): 482–487.

20.

Imbs

Wacziarg

(2003) Stages of diversification. The American Economic Review 93(1): 63–86.

21.

Jacobs

(1969) The economy of cities. New York: Random House.

22.

Kaufmann

Kraay

Mastruzzi

(2011) The worldwide governance indicators: Methodology and analytical issues. Hague Journal on the Rule of Law 3(2): 220–246.

23.

Romer

(1990) Endogenous technological change. Journal of Political Economy 98(5): S71–S102.

24.

Saviotti

(1996) Technological evolution, variety and the economy. Cheltenham, UK and Brookfield, USA: Edward Elgar Publishing.

25.

Saviotti

Frenken

(2008) Export variety and the economic performance of countries. Journal of Evolutionary Economics 18(2): 201–218.

26.

Schumpeter

(1912) Theorie der wirtschaftlichen Entwicklung. Berlin: Duncker & Humboldt.

27.

Schumpeter

(1942) Capitalism, socialism and democracy. New York: Harper.

28.

Sen

(1983) Development: Which way now? The Economic Journal 93(372): 745–762.

29.

Smith

(1776) An inquiry into the nature and causes of the wealth of nations. Chicago: University of Chicago Press.

30.

Smith

Gibson

(1988) Industrial diversification in nonmetropolitan counties and its effect on economic stability. Western Journal of Agricultural Economics 13(2): 193–201.

31.

Solow

(1956) A contribution to the theory of economic growth. The Quarterly Journal of Economics 70(1): 65–94.

32.

Stanton

(2007) The human development index: A history. PERI Working Papers, (127). Available at: http://scholarworks.umass.edu/peri_workingpapers/85

33.

Streeten

(1994) Human development: Means and ends. The American Economic Review 84(2): 232–237.

34.

Suri

Boozer

Ranis

et al. (2011) Paths to success: The relationship between human development and economic growth. World Development 39(4): 506–522.

35.

Swan

(1956) Economic growth and capital accumulation. Economic Record 32(2): 334–361.

36.

United Nations Development Programme (1990) Human development report. New York: Oxford University Press.