1. INTRODUCTION

Pro-poor economic growth is crucial for poverty reduction through decent and productive employment, including the reduction of informality. In the last decades, several countries have experienced jobless growth; this is, positive economic rates of growth accompanied by an insufficient increase in employment. The output- employment elasticity, labor intensity of growth or employment elasticity (EE) denotes the rate at which employment grows when output increases by one percent.

Like other labor market indicators, the EE does not have a unique interpretation, which could be misleading for policymakers. Since this elasticity is represented as the ratio between the change in employment and the change in output, a positive (negative) value is not always a good (bad) signal. High, positive or increasing values can indicate that more jobs are created per unit of output growth, but also could mean that labor productivity is falling and therefore working poor are increasing. Low, but positive or decreasing values could indicate that fewer jobs are created per unit of economic growth, and therefore unemployment is rising or could be the result of increasing productivity and falling unemployment. Negative elasticities arise when either employment or economic growth is negative, which is always a negative sign. However, this elasticity could end up being positive if both employment and output fall simultaneously³.

It is not just the magnitude of the effect that matters. It is also necessary to address the question about in what sectors are these elasticities observed and how they relate to the quality of jobs, because promoting employment-intensive growth may not necessarily imply poverty reduction⁴. Raising the EE or improving the quality of the available jobs (which is expected to raise the income these jobs generate) is a dilemma policymakers face for pro-poor growth purposes. Another dilemma is whether these policies should promote growth in the sectors where the majority of workers are poor, to enhance employment opportunities for them, or to promote growth in other sectors where poor could move in (ILO, 2003). The idea behind this, that concerns to this research, is that promoting labor-intensive growth in some key sectors might be better for poverty reduction than raising the total impact of growth (sectoral impact of growth), by dragging poor workers from low to higher- earning sectors of the economy, i.e. intersectoral shifts.

Therefore, it is not just the EE by itself that matters, but also its sectoral pattern and the productivity behavior in those sectors. According to Gutierrez, Orecchia, Paci, & Serneels (2007), high EE in the secondary sector reduces poverty, while high EE in the primary sector increases it so that pro-poor policies should promote more jobs in the first and more productivity in the second. Since productivity-intensive growth in agriculture has a significant effect on poverty, Manning & Purnagunawan (2013) call for policies targeting the EE in some specific sectors and subsectors. Particularly, they consider that a negative employment elasticity in agriculture is a positive sign, indicative of an inter-sectoral shift away from low productivity jobs.

As said before, EE is highly associated with productivity factors which, in turns, determine the capacity of an economy to compete in international markets. Like many other Latin American countries, Colombia has struggled to improve its productivity and become more competitive. According to the Global Competitiveness Report 2016-2017 (WEF, 2016), Colombia is the fifth more competitive country in Latin America, occupying the position 61⁵ out of 138, after Chile (35), Panamá (50), Costa Rica (52) and Mexico (57). This is a good indicator, especially if we consider all the variables taken into account to compute the Global Competitiveness Index (GCI), which includes institutional, political and economic factors⁶. This result classifies Colombia as an efficiency-driven economy (stage two of the GCI)⁷. Particularly, the country exhibits its main strength in the efficiency sub-index (position 48) and innovation (63), but perform worse in the basic requirements sub- index (85). However, Colombia is a heterogeneous country so that that index as a whole does not reflect the existing deep inequalities across regions. This is why since 2008 the Private Competitiveness Council (CPC, by its name in Spanish) estimates a Departmental Competitiveness Index (DCI), similar to the GDI, for Bogotá and 25 of the 32 departments in the country⁸.

The purpose of this paper is twofold. First, we estimate the general and sectoral EE (agriculture, manufacturing and services sectors) for the country as a whole as well as for 24 departments, including Bogotá D.C., for which information is available. Second, we use Principal Component Analysis to relate these EE to macroeconomic variables and competitive indicators at the departmental level and use cluster analysis to classify departments according to their elasticities and competitiveness. The results show an EE above one for the country, and a promising behavior at the different sectors: Greater than one in services and manufacturing sectors, and high but less than one in agriculture, behavior that could suggest workers are moving from low- productive to high-productive jobs. Unfortunately, it says nothing about the quality of the jobs, especially if we consider the high level of informality that averaged 56.3% in 2016.

Competitiveness appears to be negatively associated with informality and unemployment but positively related to high elasticities in agriculture, sector that represents the majority of the non-oil Colombian exports. High elasticities in manufacturing and services are accompanied by higher levels of economic growth. However, according to these results, economic growth does not necessarily mean to be competitive, and viceversa. The majority of the departments show negative elasticities in agriculture, while all elasticities in services are positive. The cluster analysis generates four groups, with the most competitive departments showing positive elasticities, larger in services. Even though there is no clear indication of how competitiveness and elasticities relate to each other, the clusters based on competitiveness, on elasticities, or both show similar results.

2. THEORETICAL FRAMEWORK

The EE represents the ratio between the relative change in employment (.) and the relative change in output (Y):

(1)

which can be obtained by either direct application of the above expression or through econometrics models. The simplest model is given by

(2)

or

(3)

where è. and y˙_t represent rates of growth, Ln_Et and Ln_Yt stay for the logarithm of employment and output, and β₂ is the estimated EE. Seyfried (2014) modifies equation (2) to consider the persistence of employment growth for several European countries during the period 1999-2012, using impulse-response analysis to determine the appropriate lag-structure based on the Akaike information criterion:

(4)

Gutierrez et al. (2007), on the other hand, measures the effect of changes in per capita value added on employment by sector, instead of economic growth, considering a sample of 37 developing countries for the period 1980- 2004 based on the following expression:

(5)

where y= Y/A represents the value added per working person, with Yindicating the total value added, Aas the total population working age, and y¯_i as the share of growth in yin sector i attributable to economic growth; a¯ corresponds to the marginal contribution of the inverse of the dependency ratio (A/N) to the change in per capita income, with . indicating total population; e_i.̅ represents the share of total growth attributable to growth in the employment rate, with e=E/A.

Besso (2010) analyzes the impact of some macroeconomic variables on EE in Cameroon for the period 1994-2003. To do that, he estimates equation (2) considering the production of primary, secondary and tertiary sectors, and its share (h) in the total output, as follows:

(6)

Then, he uses the estimated elasticities to fit a model in which labor force, growth rate of rural and urban population, final consumption expenditure of household, household income, rate of growth of financial capital and productivity of labor are used as explanatory variables. He also estimates the EE separately for public and private (formal and informal) sectors, gender and age group. His results indicate that the effect of all these variables is weak.

It is well known that in developed economies the primary sector employs a relatively small proportion of workers with a declining and even negative labor intensity but increasing productivity⁹. In fact, countries that have proven to substantially reduce poverty showed a significant increase in workers’ productivity that allows them an inter-sectoral shift in their transition out of agriculture. Ernst & Berg (2009) indicate that alleviating poverty in the agriculture sector requires incentives such as fair prices, access to services and markets, and land rights to encourage investments in technology. The higher the productivity in this sector, the higher the opportunity for workers to move to more productive and better-paid jobs in manufacturing and services, with a substantial reduction in poverty as proven by the experience in Indonesia, China, Vietnam and Uganda in the last decades of the XX century. In fact, the poverty reduction in East Asia was mostly due to high EE in secondary and tertiary sectors (Khan, 2007). This does not imply abandoning agricultural activities to saturate the labor market in the other sectors, but making the primary sector productive enough to support the development of the others to become competitive.

Papers such as Nuñez & Espinosa (2005) and Sarmiento, González, Alonso, Angulo, & Espinoza (2005) analyze the economic evolution of Colombia across years and conclude that the observed growth was not in favor of the poor due to the lack of policies aimed at the creation of quality employment and to the improvement in the productivity level of each sector. For several years, the CPC has been working on the design of proposals that guide public policies around a structural change in the country that allows improving its level of competitiveness and reduce poverty. These proposals revolve around the sophistication of the Colombian productive system, and the solution of some structural problems that prevent such sophistication.

The Colombian National Development Plan 2014-2018 (República de Colombia, 2015) proposes a modern industrial policy based on achieving a sustainable economic growth by improving the productivity in such a sector and the generation of export products with high value added. According to this Plan, the government grants a very important role to the industrial sector as a source of employment and economic growth, for which it is necessary the introduction of more sophisticated and diversified production processes, the investments in research and development by both, the public and private sector, and the generation of opportunities for workers to move to better paid jobs.

So far some improvement in the productivity of all sectors has been achieved, especially in the services one. However, a report of the CPC (2014) shows that Colombia is very far from this goal, with an inefficient allocation of resources and high heterogeneity in labor productivity across regions. Production factors should be allowed to migrate among sectors in order to improve the worker´s opportunities to have access to higher quality jobs and, therefore, raise their living conditions. According to the report, by the year 2012, the agricultural was the sector with the lowest level of productivity but employed nearly 19% of all workers, almost the same proportion as the much more productive and dynamic industrial sector. The remaining 60% of the workers were occupied in activities at the services sector, especially in the area of commerce, restaurants, and hotels, of relatively low productivity. Which sectors should then be prioritized? Although the document recognizes the difficulty of considering whether one sector is «better» than another, it is true that some activities trigger greater externalities or serve as a bridge towards productive diversification (Hausmann & Klinger, 2008). However, not all studies point to the same direction. Porter (1998), for example, states that it is not the sector that matters, but its productivity. In spite of it, the study of the CPC emphasizes the need for defining priority macro-sector, according to the potentials and limitations of each region or department, as well as the identification and development of productive clusters, such as tourism (services) and the manufacture of products based on the comparative advantage and resources of each region (dairy products, chocolates, cosmetics, energy, etc.), something similar to that public policies proposed in similar countries, such as Malaysia.

Even more, the OECD (2017) supports the need to facilitate the reallocation of labor and capital to the most competitive sectors in Colombia, in order to take advantage of its potential for inclusive growth, not forcing workers to remain in low-paying jobs. The report recognizes that public investment is essential to reduce the gaps in infrastructure that will increase the competitiveness of all sectors and reduce regional inequalities. In this sense, the study points out that Colombia should take advantage of the identification and development of value chains or clusters, some of which has been identified by the CPC. To do this, it is necessary a better understanding of the pattern of behavior of the employment generated per unit of economic growth, especially across regions.

As said previously, a competitive pattern of growth should show differences in the values and directions of the EE across sectors, declining and even becoming negative in agriculture, while increasing in manufacturing and services as these two last sectors absorb labor force fast enough as to exceed its growth. On the contrary, high labor intensity in the primary sector indicates that agriculture is acting as a refuge for displaced workers (Ernst & Berg, 2009) that absorbs labor force during a recession and expels it to another sector in periods of expansion (Manning & Purnagunawan, 2013). According to these studies, poverty reduction does not imply increasing employment intensity of growth in this sector but its productivity, to facilitate the workers’ transition to secondary and tertiary more productive jobs in sectors with higher elasticities of employment. Therefore, policies and programs should be targeted to increase employment in the secondary and tertiary sectors as well as increase productivity in the primary sector. This seems to contradict the conclusion reached in 2008 by the Committee of Experts on the Application of Conventions and Recommendations that highlights the importance of rural employment in poverty reduction and pointed out as a policy option identify and adopt measures to increase labor intensity as well as employment promotion in rural areas (De Luca, Fernando, Crunel & Smith, 2012). However, this kind of promotion must be oriented to the nonagricultural sector. Of course, the weight of the primary sector in generating jobs may affect the results¹⁰.

3. LITERATURE REVIEW

Several studies have attempted to estimate the EE for different countries. The results vary among regions and periods. Piacentini & Pini (2000) find negative elasticities for Italy, Germany, United Kingdom and Sweden during the 90s, while Seyfried (2014) show positive values for Portugal, Ireland, Italy, Greece, and Spain in more recent years. According to Padalino & Vivarelli (1997), the EE averages 0.5 in the US and Canada, while Choi (2007) observes declining values of the EE in Korea, going from 0.49 in the period 1971-1980 to 0.38 in 2000-2005. The relationship between employment and economic growth may differ across sectors and regions within a country. Perugini (2009) illustrates these inequalities in Italy, using regional data for period 1970-2004. However, not only regional and sectoral differences are important. It is necessary to identify the labor-intensive activities within each sector that should be targeted. Sassi & Goaied (2016) use panel data information for 15 industries to estimate the long-run elasticities and identify the most labor-intensive industries (services and export manufacturing industries).

Manning & Purnagunawan (2013) analyze the situation of Indonesia considering agricultural, non-agricultural and total GDP. Their findings show negative values (-1.7) in the former sector and a positive elasticity for nonagricultural activity (3.4), with a total average of 1.4. Mazumdar (2003) analyzes the manufacturing sector to estimate the effect of economic growth on employment in Latin America and the Caribbean, East Asia, and the OECD countries, finding negative manufacturing EE in the first region. Kapsos (2005) is probably the largest study in the field for the period 1991-2003, with 160 countries including Colombia. The study is conducted at the general level but also by gender and sector (agriculture, industry, and services) in order to detect structural changes, considering both value added and GDP. He estimates the EE to be about 0.23, 0.18, 0.45 and -0.42 for North America, East Asia, Latin America and the Caribbean region, respectively. The highest values are observed in Africa and the Middle East. His analysis is also conducted by country in three different periods (1971-1980, 1981- 1992 and 1993-2003). The declining trend in the responsiveness of employment to changes in GDP observed by Kapsos is confirmed by Navarro (2009), based on data from 15 Latin- American countries, including Colombia, over the period 1980-2008.

Hvozdíková & Morvay (2015) estimate the EE in Slovakia and compares it with other economies in the region during the period 1997-2010. According to them, the observed low¹¹ values are common in developing economies, trying to overcome the technological and productivity gap. This contradicts some statistics shown by the International Labour Organization (ILO) according to which Europe and East Asia exhibit elasticities as low as 0.2 and 0.3 while Latin America has values above 0.5. Pagés, Pierre, & Scarpetta (2009) evaluate the factors affecting the countries’ ability to create more jobs in spite of experiencing economic growth, considering the case of Latin America and the Caribbean for the period 1990-2006. In general, they conclude that the low productivity levels must be blamed for the modest economic growth observed by most of them: by 2006, the average productivity in the region was 21 percent of that of the United States. Only a few countries, such as Brazil, Mexico, Panamá, Peru, and others experienced simultaneously an increase in employment and in productivity; Colombia, on the other hand, along with Argentina, Chile, Jamaica, and Uruguay suffered jobless growth. The ability of the labor market to react to changes in the output in the region was also studied by González (1999) based on the Okun’s Law applied to thirteen Latin American countries for the period 1960-1995. He finds out that the labor market in the region reacts adjusting real wages more than they do in the United States, while employment reacts less to changes in GDP¹².

For the case of Colombia, the EE estimated by Kapsos (2005) for the three periods considered are 0.63, 0.42 and 0.94. The EE is about 1.14 in the agricultural sector, -0.29 for manufacturing activities and 0.93 in the services sector. Moreover, the elasticity shows a decreasing pattern for women (moving from 0.82 to 0.35) and increasing for men (from 0.52 to 1.33). Also for Latin America, the ILO (2010) finds evidence of a shift away from employment in the agricultural towards the services sector. Another study involving Colombia as part of a larger cross-country dataset estimates the EE for the country to be about 0.98 for years 1991-2011 (Ben Slimane, 2015). Angel (2004) analyzes the effect of an increase in minimum wage on EE, while Argüello, Jiménez, Torres, & Gasca (2016) measure the impact of a sustained oil export boom in the labor dynamic, including the EE in that sector. To our knowledge, there does not exist a study that attempts to estimate EE in Colombia at the departmental and sectoral level, and associates them to relevant economic variables.

4. METHODOLOGY

To estimate the general and sectoral EE for Colombia, the study uses information from the National Administrative Department of Statistics (DANE in Spanish) and the Bank of the Republic (BR), as well as the World Bank indexes (WDI) and the DCI from the Consejo Privado de Competivividad (CPC). Nationwide EE are estimated for the period 1990-2016 through ordinary least squares applied to equation (3), including lagged values of employment to account for its persistence effect. Sectoral elasticities are also estimated based on equation (6) just to corroborate the results obtained from (3). As for the departmental EE, these are obtained based also on equation (3) limited to the regions for which information is available (24 out of 32 Departments including Bogotá D.C.) At the general level, for these last elasticities, we use data for period 2001-2016, while sectoral values are based on the longest possible period of time available, varying from 2001-2016 to 2007-2016.

Like Besso (2010), Ben Slimane (2015) and others, we assess the factors associated with the EE at the department level. However, unlike them, we do not run regression models but use Principal Component Analysis (PCA) considering macroeconomic variables and DCI information for the year 2016. PCA is an exploratory multivariate analysis technique used to uncover the internal structure of the relationship in a large set of variables, grouping them into new variables called components, based on the correlation among them and their dispersion. Particularly, PCA explains the variance-covariance structure of the data through linear combinations of quantitative variables, in order to reduce the data dimensionality and to facilitate its interpretation. The analysis provides a graphical representation of such positive (variables in the same side) or negative (variables in opposite sides) associations, with the axes formed by the components on which individuals can be plotted.

Due to limited information, we apply a two-step PCA: first, all social, economic and competitiveness variables are used to construct an economic framework; this framework is then used to analyze the estimated elasticities. Only components whose eigenvalues are greater than or equal to 1 are retained. By doing so we can take advantage of all variables available, in spite of their limited number of observations. Following Ben Slimane (2015), the variables included in the first stage of the analysis are real value added (in constant - 2010- dollars); real imports and exports (in constant -2010- dollars), to consider the role of international trade; inflation rate (%) that introduces noise and uncertainty into the economy, and social and labor market indicators such as working age population (%), unemployment rate (%), underemployment (%), and labor force rate (%), to which we also add the informality rate (%) and the proportion of poor households according to the index of unsatisfied basic needs (%). To account for the effect of the guerrilla warfare, we used the proportion of displaced from each department. Finally, the DCI is included, as well as its basic requirements, efficiency, and innovations and sophistication sub-indexes. The variables included in the second stage of the PCA are the components previously obtained as well as the estimated general and sectoral elasticities for the year 2016. Biplot is graphically used to represent these results.

Finally, we group departments according to their elasticities and levels of competitiveness, being the complete linkage hierarchical cluster analysis the one that yielded the most comprehensive results. This is an iterative method in which, starting from as many clusters as available observations, we sequentially combine them based on the shortest distance between elements in two different groups that are as far away as possible from each other. The results are represented by a dendrogram (Figure Nº 1).

5. RESULTS AND DISCUSSION

As stated before, high and positive EE could indicate that more jobs are created per unit of output. For the case of Colombia, the estimated value of the employment intensity of growth is 1.03, larger than the value estimated by Kapsos (2005) and Ben Slimane (2015), and above the 0.5 estimated by ILO for the region (see Table Nª 1).

A pro-poor pattern of growth is expected to show larger EEs in the manufacturing and services sector, and low or even negative numbers at the agricultural level. The results show that sectoral elasticities match this behavior, with the largest value in manufacturing (2.39) and the lowest but still high in agriculture (0.85)¹³, which differ from the estimates obtained by Kapsos (2005) for a different period.

Dragging people out from low- productivity jobs to more productive sectors helped some Asian countries to substantially reduce poverty and become more competitive. In Colombia, the employment in agricultural sector shows a sharp reduction, accompanied by an increasing participation of the tertiary sector in the labor market, as well as a sustained growth of the value added in that same sector (see Figures Nº 1a and Nº 1b). This decreasing employment in agriculture comes with a raising productivity in such a sector, moving from 4,174 constant (1980) dollars to USD 4,250 in 2000 and USD 6,536 in 2016.

The results at regional level highlight the heterogeneity within the country, but somehow support conclusions for the country as a whole. The majority of the departments shows negative EE in the agricultural sector, suggesting individuals are escaping for poor jobs in this sector or from the violence in rural areas. In many cases, these negative elasticities in the primary sector come along with positive and large elasticities in manufacturing and/or services. Notice also that the number of departments with negative elasticities declines as we move from the agricultural to the tertiary sector, with no negative elasticities in the last one. The richest and usually more competitive departments (Bogotá D.C., Antioquia, Valle del Cauca, Santander, Cundinamarca, and Meta) do not necessarily show the largest EE, in the same way that the poorest ones do not exhibit a clear pattern of behavior at this regard. However, high elasticities usually accompany high levels of innovations and sophistication. Chocó, which can be considered the poorest department in Colombia, shows the largest negative global elasticity.

The first stage of the unrotated PCA on macroeconomic variables generates four factors or components that can be described as follows. The first is a component of competitiveness explains 57.99% of the total variability of the variables and is named COMPETE, since added value, DCI and its sub-indexes are strongly opposed to poverty and informality, meaning that high levels of competitiveness are associated to high levels of added value and low levels of these two problems. The second component explains 12.39% of the total variability and opposes added value and international trade variables to labor market indicators such as unemployment rate and labor force, reason why it can be considered a component of economic growth and is given the name of GROWTH. The third is a Phillips curve component that explains 9.59% and opposes inflation to unemployment, therefore receiving that same name. Finally, the last component has an explanatory capacity of 7.07% and opposes informality, inadequate employment, and the displaced to exports, so that it reflects the capacity of the economy to absorb the labor force, and is given is named INFORMAL(see Table Nª 2).

Figure 1
Employment and real value added by sector
Authors, with data from WDI

The second stage of the PCA indicates a separation between elasticities in the agricultural sector (EAG) from those in the manufacturing (EMN) and services (ESR) sectors. The first of the three components obtained explains 34.28% of the total variability. According to this component, high elasticities in the secondary and the tertiary sectors imply economic growth but not necessarily competitiveness. The second component (23.84%) associates high elasticities in the agricultural sector with competitive capacity, but also with inflation, commonly observed among agricultural product. Once again, high elasticities at the agricultural sector and competitive capacity are opposed to high employment elasticity in manufacturing (see Table Nª 3).

The last component retained explains 14.6% of the total variability and clearly opposes the country’s capacity to provide formal and adequate jobs to competitiveness, indicating that informality and better working conditions are required for Colombia to become more competitive. The biplot tool allows us to graphically represent the departments along with these components. As expected, Bogotá, Antioquia, Valle del Cauca, Santander, and Cundinamarca appear at the right side of the plot as the most competitive departments, opposed to Chocó and Caquetá, the poorest in the country. High manufacturing employment elasticities are observed at departments with low agricultural elasticities (Cesar, Tolima, Cauca, Meta, among others), indicating a possible inter- sectoral shift of workers. In the same way, departments at the bottom of the plot show faster rates of growth in spite of their high informality levels, and higher elasticities in the tertiary sector (see Figure Nª 2).

Table 2

First-step PCA results: Economic framework

Authors, with data from DANE, BR, and UER

Figure 2
Biplot results
Authors, based on their own estimates and UER

The dendrogram suggests four groups at a distance of less than 5 (see Figure Nº 3)¹⁴. The first cluster includes Bogotá and Antioquia, regions that show high and positive elasticities, especially in the services sector¹⁵, as well as the highest levels of competitiveness, especially in terms of innovation and sophistication.

The second cluster includes the departments that follow in the DCI ranking, which are those with acceptable values in the basic requirements and innovation and sophistication sub-indexes, separated into two groups: regions with low elasticities in manufacturing but high elasticities in services (Atlántico, Cundinamarca, Valle del Cauca), and regions such as Caldas, Risaralda, Santander, and Boyacá with low elasticities in both sectors. The third cluster includes all the remaining departments but, Caquetá, La Guajira, and Chocó. The competitive levels in this group are relatively low and appear particularly weak in innovation and sophistication. Their elasticities do not show a clear pattern of behavior, ranging from all high and positive (Sucre) to all low or negative (Meta, Tolima, and Norte Santander). Finally, the last cluster includes the least competitive departments for which we observe extreme values of EE (the largest negative value in Chocó and the largest positive value in La Guajira), large negative values in the agricultural sector and very high values in manufacturing or services.

Figure 3
Dendrogram
Authors, based on their own estimates and UER

6. CONCLUSIONS

Employment elasticity is used to measure by how much employment increases per unit of economic growth. In spite of its weaknesses, it is necessary to analyze the behavior of such an index across sectors in an attempt to detect jobless growth and introduce policies to correct it. The experience observed in several countries indicate that promoting the creation of employment in some key sectors, may contribute to substantially reduce poverty.

Using information for the period 1990- 2016, the study estimates the EE for Colombia. The country shows a 1.03 elasticity as a whole, meaning that employment grows almost proportionally with the output. As for sectoral behavior, higher values are observed in manufacturing (2.39) and services (1.14) compared to agriculture (0.89), indicative of a potential inter-sectoral shift from poor to better jobs, especially in the light of the increasing added value in the former sectors, and declining in the last one. In fact, while the share of the agricultural sector in total employment has moved from 50% in 1980 to less than 17% in 2016, the real (2010) added value per worker in this sector went from US$ 4,174 to US$ 6,536 in that same period. Several studies suggest that declining or even negative elasticities in agriculture, along with high and positive elasticities in services and industry facilitate the transition of workers to more productive and better-paid jobs, a condition required for both economic growth and competitiveness, especially in the light of the negative association observed between this and informality.

The analysis is conducted also for 23 out of the 32 departments in Colombia in addition to Bogotá, but limited to the period 2001-2016 or 2007-2016, based on data availability. Their elasticities range varied from 1.23 (La Guajira) to -0.02 (Cauca) and -0.15 (Chocó). The majority (52.2%) show negative elasticities in agriculture, while for only 17.4% of them the largest positive elasticity is observed in this same sector. The number of departments with negative elasticities shrinks as we move from primary to tertiary sector, with only 25% of the departments with negative elasticities in manufacturing (the sector with the highest values), and no negative elasticities in the services.

Principal Component Analysis is used to uncover the association of the estimated elasticities with some socioeconomic variables. To overcome the lack of information, the analysis is performed in two steps. The first part of the analysis uses a set of macroeconomic, demographic, and competitive variables to build an economic framework in which to analyze the estimated elasticities. Four components explaining 87.04% of the total variability are obtained. The second part of the analysis combines these four factors along with the sectoral elasticities, yielding three factors that account for 72.72% of the total variability. According to them, competitiveness is associated with low levels of poverty and informality. In this scenario, relatively low but positive elasticities are observed in the manufacturing and services sectors. Another component suggests that the increasing valued added and active international trade come with low levels of unemployment, but at the cost of higher prices. In this case, higher elasticities in the secondary and tertiary sectors are especially associated with economic growth, while inflation is usually associated with high elasticities in agriculture.

The study also uses cluster analysis to classify departments based on their competitiveness and elasticities, which helps to better understand the relationship between these two factors. The results indicate that very highly competitive regions, especially in terms of innovation and sophistication, have positive and relatively higher elasticities in manufacturing and services. Only Bogotá and Antioquia are part of this cluster. Highly competitive departments in terms of basic conditions and innovations and sophistication, consistently show low elasticities in agriculture and manufacturing; this cluster includes Atlántico, Cundinamarca, Santander, Valle del Cauca Caldas and Risaralda. The least competitive regions, La Guajira and Chocó, have large negative elasticities in the agricultural sector along with extremely high elasticities in any of the others. As for the departments with average levels of competitiveness, no clear pattern of behavior of the elasticities is observed.

REFERENCIAS

Angel, D. F. (2004). A minimum wage increase can have an adverse distributional impact: The case of Colombia. Bogotá: Departamento Nacional de Planificación, Archivos de Economía, 247.

Argüello, R., Jiménez, D., Torres, E., & Gasca, M. (2016). Dutch disease, informality, and employment intensity in Colombia. Working Papers MPIA 2016-07, PEP-MPIA. Retrieved from https://jgea.org/resources/download/7656.pdf

Ben Slimane, S. (2015). The relationship between growth and employment intensity: Evidence for developing countries. Asian Economic and Financial Review, 5(4), 680-692.

Besso, C. R. (2010). Employment intensity of growth and its macroeconomics determinants. Munich Personal RePEc Archive-MPRA, working paper 35615. Retrieved from https://mpra.ub.uni-muenchen.de/35615/

Choi, C. K. (2007). The employment effect of economic growth: Identifying determinants of employment elasticity. South Korea: Chonbuck National University, Working paper (2017/10).

Consejo Privado de Competitividad, CPC. (2014). Políticas de desarrollo productivo para Colombia. Bogotá: CPC.

De Luca, L., Fernando, M., Crunel, E., & Smith, L. (2012). Unleashing the potential for rural development through decent work, 1970-2011. Geneva: Rural Employment and Decent Work Program, ILO.

Ernst, C., & Berg, J. M. (2009). The role of employment and labour markets in the fight against poverty. Promoting pro-poor growth: employment. Paris: OECD.

González, J. A. (1999). Labor market flexibility in thirteen Latin American countries and the United States: Revisiting and expanding Okun coefficients. Washington: Center for Research on Economic Development and Policy Reform, working paper 136. Retrieved from http://documents.worldbank.org/curated/en/864361468768724600/pdf/multi-page.pdf

Gutierrez, C., Orecchia, C., Paci, P., & Serneels, P. (2007). Does employment generation really matter for poverty reduction? Washington: World Bank, Poverty Reduction Group.

Hausmann, R., & Klinger, B. (2008). Achieving export-led growth in Colombia. Cambridge: Harvard University, Center for International Development.

Hvozdíková, V., & Morvay, K. (2015, August). Low employment intensity of growth and specifics of Slovak labour market. European Scientific Journal, special edition, 285-292.

International Labor Organization, ILO. (2003). Review of the Core Elements of the Global Employment Agenda. Geneva: ILO, Committee on Employment and Social Policy.

International Labor Organization, ILO. (2010). Global employment trends 2010. Geneva: ILO. Retrieved from http://www.oit.org/wcmsp5/groups/public/—ed_emp/—emp_elm/—trends/documents/publication/wcms_120471.pdf

Kapsos, S. (2005). The employment intensity of growth: Trends and macroeconomic determinants. Geneva: ILO, Employment strategy papers.

Khan, A. R. (2007). Employment and pro-poor growth. Brazil: Poverty in Focus, International Poverty Center. Retrieved from http://www.ipc-undp.org/pub/IPCPovertyInFocus10.pdf

Ludena, C. E. (2010). Agricultural productivity growth, efficiency change and technical progress in Latin America and the Caribbean. Washington: Inter-American Development Bank, working paper 186. Retrieved from http://www20.iadb.org/intal/catalogo/PE/2010/05828.pdf

Manning, C., & Purnagunawan, M. R. (2013). Using employment elasticities in the analysis of employment trends in Indonesia. Washington: USAID-SEADI. Retrieved from http://pdf.usaid.gov/pdf_docs/pbaaa147.pdf

Mazumdar, D. (2003). Trends in employment and the employment elasticity in manufacturing, 1971-1992: An international comparison. Cambridge Journal of Economics, 27(4), 563-582.

Mkhize, N. I. (2015). The sectoral employment intensity of growth in South Africa 2000-2012. Pretoria: University of South Africa, Department of Economics. Retrieved from http://uir.unisa.ac.za/bitstream/handle/10500/20131/thesis_mkhize_ni.pdf?sequence=1&isAllowed=y

Navarro, L. (2009). Employment dynamics and crises in Latin America. CEPAL Review, 99, 23-40.

Nuñez, J., & Espinosa, S. (2005). Pro-poor growth and pro-poor programs in Colombia. Bogotá: Universidad de Los Andes, Documentos CEDE 003379. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.612.8193&rep=rep1&type=pdf

Organization for Economic Cooperation and Development, OECD. (2017). Economic surveys: Colombia. Paris: OECD.

Padalino, S., & Vivarelli, M. (1997). Examining the employment intensity of economic growth in the G7 countries. International Labour Review, 136(2), 191-213.

Pagés, C., Pierre, G., & Scarpetta, S. (2009). Job creation in Latin America and the Caribbean: Recent trends and policy challenges. Washington, DC: The World Bank and Palgrave Macmillan. Retrieved from https://openknowledge.worldbank.org/bitstream/handle/10986/9386/9780821376232.pdf?sequence=1

Perugini, C. (2009). Employment intensity of growth in Italy. A note using regional data. Quaderni del Departimento di Economía, Finanza e Statistica, 9(1), 59-92. Retrieved from http://www.usc.es/economet/journals2/eers/eers913.pdf

Piacentini, P., & Pini, P. (2000). Growth and employment London. In Vivarelli, M. & Pianta, M. (Eds.), The employment impact of innovation: Evidence and policy (pp. 44-76). London: Routledge.

Porter, M. E. (1998). The competitive advantage: Creating and sustaining superior performance. New York: Free Press.

República de Colombia. Congreso de la República. (2014). Plan Nacional de Desarrollo 2014 -2018 «Todos por un nuevo país». Bogotá, Congreso de la República, Ley 200-C/2015.

Sarmiento, A., González, J. I., Alonso, C. E., Angulo, R., & Espinoza, F. (2005). Crecimiento pro-poor en Colombia 1996- 2004. Coyuntura Social, 33, 129-150.

Sassi, S., & Goaied, M. (2016). Long-term employment intensity of sectoral output growth in Tunisia. International Labor Review, 155(2), 253-263.

Seyfried, W. (2014). Examining the employment intensity of economic growth in the PIGGS. International Business and Economics Research Journal, 13(3), 593-598.

World Economic Forum, WEF. (2016). The global competitiveness report 2016-2017. Geneva: WEF. Retrieved from http://www3.weforum.org/docs/GCR2016-2017/05FullReport/TheGlobalCompetitivenessReport2016-2017_FINAL.pdf

Notes

Author notes