Report No. 42269-CO Colombia Inputs for Sub-Regional Competitiveness Policies June 12, 2008 Poverty Reduction and Economic Management Unit Latin America and the Caribbean Region Document of the World Bank ACRONYMS AND ABBREVIATION A I Domestic Agenda Agenda Intema Bancoldex Foreign Trade Bank of Colombia Banco de Comercio Exterior de Colombia CAEs Centers for EnterpriseAssistance Centros de Atenci6n Empresarial CEER Center for Regional Economic Studies Centro de Estudios Econ6micosRegionales CNC National Competitive Commission Comisibn Nacionalparala Competitividad Coinvertir Invest inColombia Corporation Corporacidn Invertir en Colombia Colciencias ColombianInstitutefor the Development of Instituto Colombiano parael Desarrollo de la Science and Technology Ciencia y la Tecnologia Confecharas Confederationof Chambers of Commerce Confederaci6n Colombiana de Charas de Comercio CORFO Chile's Economic Development Agency Corporacibn de Foment0de la Producci6n DANE National Statistics Department DepartamentoAdministrativo Nacional de Estadistica FDI Foreign Direct Investment Inversi6nExtranjera Directa FTA Free-TradeAgreement Acuerdo de Libre Comercio GDP Gross Domestic Product Producto Interno Bruto GFP Global FacilitationPartnershipfor Transportation Asociaci6n Mundial para la Facilitacibndel andTrade Transportey el Comercio GoC Governmentof Colombia Gobiemo de Colombia HHl Herfindahl-Hirschman index indice Herfindahl-Hirschman ICA Investment Climate Assessment Evaluaci6ndel Clima de Inversiones ICs Investment Climate Survey Investigacibn del Clima de Inversiones ICT Information and CommunicationsTechnology Tecnologia de Informaci6ny Comunicaci6n IDA Industrial DevelopmentAuthority Autoridad parael Desmollo Industrial IO Input-output Insumo-Product0 ISIC International StandardIndustrialClassification Clasificaci6n IndustrialIntemacional Uniforme LAC Latin America America Latina LALC Latin America Logistics Center Centro para Logisticade America Latina LPI Logistics PerceptionIndex fndice de Percepci6nLogistica MPM Multiplier Product Matrix Matriz Multiplicadora del Producto MSME Micro, Small and Medium Enterprises Micro, PequeAas y MedianasEmpresas NPCC National Productivity and Competitiveness Asociaci6nNacionalpara la Productividad y Committee Competitividad NAFTA NorthAmerican Free Trade Agreement Acuerdo de Libre Comercio de America del Norte PISA Programmefor International Student Assessment Programade Evaluaci6n de Estudiantes Extranjeros ProChile Chilean Trade Commission Direcci6nde Promoci6n de Exportaciones PROMPEX Peruvian Export Promotion Agency Comisi6n para la Promoci6n de Exportaciones R&D ResearchandDevelopment Investigacibny Desmollo RED1 Recent Economic DevelopmentsinInfrastructure Desarrollos Econ6micosRecientes en Infraestructura RTDC Researchand Technological Development Centers Centros de Investigaci6n y Desmollo Tecnol6gico SABER National Evaluation System for the Quality of SistemaNacionalde Evaluaci6n de la Calidad de la Education Educaci6n SENA National Learning Service Servicio Nacional de Aprendizaje SME Small and Medium Enterprises PequeAasy Medianas Empresas TFP Total FactorProductivity ProductividadTotal de 10s Factores TIMSS Trends inInternational Mathematicsand Science Study Tendencias enel Estudio de Matemhticay Ciencias USAID United States Agency for International Agencia de 10s Estados Unidosparael Desmollo Development Intemacional WEF World Economic Forum Forum Econ6mico Global ACKNOWLEDGEMENTS The report "Colombia: Inputs for Sub-Regional Competitiveness Policies" was prepared by a team led by Jose Luis Guasch and Jose Guilherme Reis, based on contributions from a team of Bank staff and consultants. Tito Yepes acted as a third team leader, providing inputs for all chapters and leading contacts with the authorities and the organization of the workshops held in Colombia as part of the preparation of this report. Mariam Dayoub provided comments and assistance for the report and co-wrote Chapter 4 with Jose Guilherme Reis. Jose Barber0 preparedChapter 5 and Erik Bloom provided a revision of Chapter 6. Juan Carlos Junca, Pooja Kacker and Darwin Marcel0 contributed in the preparation of Chapters 2 and 3, and Adam Calderon helpedwith the editing ofthe report. Victoria Alexeeva provided assistance at the early stages of the project and Patricia Melo assistedwith the production of the report. Peer reviewers Mauricio Carrizosa, Maria Emilia Freire, Tomas Serebrisky and Sahid Yusuf provided valuable comments and suggestions. The report also benefited from helpful inputs from Makhtar Diop, Lily Chu, David Rosenblatt, Anna Wellenstein, Miguel Lopez-Bakovic, Jonas Frank, Cristian Torres, Juan Carlos Mendoza and Mark Hagerstrom. The team acknowledges important contributions received from government officials of the Departamento Nacional de Planeacidn, especially Orlando Gracia, Director de la Division Empresarial, Rodrigo Moreira, Subdirector de Politica Industrial y Comercial, and Rafael Lopez, advisor, and of the Ministerio de Comercio, especially Sergio Diazgranados, Vice- Ministro, Santiago Pinzh, Director de Productividad y Competitividad, and Luz Deicy Florez, advisor. The team also benefited from insights provided by participants in two meetings organized in Barranquilla and Bogota in February and May 2007, respectively, as well as in Bogota, Popayan, Manizales, and Bucaramanga in May 2008. The team also thanks the institutions and individuals that helped to organize those meetings, especially Banco de la Republica, Universidad del Norte, Jaime Bonet, Adolfo Meisel, Karina Ricaute and the public authorities of Barranquilla. The team appreciates the support provided by Alejandro Vivas and Lilia Salgado from PontiJicia Universidad Javeriana and Carlos Villate from IDC Colombia, and from the Comisiones Regionales de Competitividad in Popayan, Manizales, and Bucaramanga. The econometric analysis of the investment climate survey was based on a background paper Escribano, A,, J.L. Guasch and J. Pena (2007) "Colombia: Investment Climate Assessment on Productivity and Allocative Efficiency: Analysis based on Firm-Level Data from 2005," The findings and views expressed here are exclusively those of the World Bank and do not represent the views of the government of Colombia. Colombia: Inputsfor Sub-Regional CompetitivenessPolicies Table of Contents Executive Summary ....................................................................................................................................................... i Chapter 1-Introduction ................................................................................................................................................ 1 Chapter 2 -Policies for Sub-RegionalCompetitivenessinColombia 6 2.1. Competitiveness .................................................................................................................................. .......................................................................... 6 2.2. Policy Instruments for Competitiveness at the Sub-Regional Level and Some Lessons from the International Experience .................................................................................................................................. 7 2.3. Colombia's Institutions for Sub-Regional Competitiveness ................................................................... 16 Chapter 3 -Competitivenessofthe Sub-regionsof Colombia ............................................................................ 19 3.1. Productivity and Competitivenessat the National Level 3.2. Sub-RegionalEconomic Structure ...................................................................................................... ........................................................................ 19 -21 3.3. Economic ConvergenceamongSub-Regions ....................................................................................... 24 3.4. Productivity and Location of Manufacturing Firms ............................................................................... 27 Chapter 4 -InvestmentClimate inselected regions of Colombia ....................................................................... 32 4.1. Introduction ...................................................................................................................................... 32 4.2 Regional Benchmarking ..................................................................................................................... 33 4.3 A Closer Look at Technology andInnovation 38 4.4 Productivity Determinants .................................................................................................................. ...................................................................................... 39 Chapter 5 -Productive Infrastructureand Sub-RegionalCompetitiveness inColombia ........................................ 50 5.1. Introduction ...................................................................................................................................... 50 5.2. The GeneralConsensus of Low Inter-Sector and Interregional Links inColombia's Economy .................51 5.3. The Freight Logistics System Responseto the SpatialInteractionDemands ............................................ 55 5.4. Perspectivesfrom aMicro Approach 5.5. ConclusionsandRecommendations .................................................................................................... 63 ................................................................................................... 60 Chapter 6 -Human Capital andInnovation: National and Sub-RegionalAnalysis 6.1. HumanCapital Endowment ................................................................................................................ 65 ............................................... 65 6.2. Innovation and Technological Development ........................................................................................ 72 6.3, Conclusionsand Recommendations .................................................................................................... 75 Chapter 7 - Conclusions ................................................................................................................................. 79 Annex: Econometricsusing InvestmentClimate Data ....................................................................................... 89 List of Tables Table 1 1:Regional developmentpolicy alternatives.................................................................................................. Table 1 2: GDP per capita, annual real GDP growth, GDP composition and educational attainmentby sub-region .. ..-24 Table 3 1:Total factor productivity inLatin America, selected countries ................................................................ 20 Table 3 2: Colombia: Distributionof manufacturing f m s by sub-regionand sectors, 2002 (percent) ..................... Table 3 3: Two-stage Probit: Determinantsofthe probability of locating ina sub-region vs any other sub-region.3 1 ... 23 . Table 4. 1: Colombia ICs: Sample composition by region and industry, 2006 .......................................................... Table 4. 2: Marginaleffects on innovation inputs and outputs inColombia............................................................... 33 40 Table 4. 3: IC elasticitiesand semi-elasticitieswith respectto productivity ............................................................... 42 Table 5. 1:Inter-industry linkagesfrom the 2004 input-output matrix....................................................................... 53 Table 5 2: Qualitative analysis of the competitive and complementaryrelationships amongregions ....................... Table 5 3: Backward and forward linkage oriented sectors, based on standardizedpure linkages indices .. ................53 54 Table 5.5: Logistics PerceptionIndex for Colombia and selected countries .............................................................. 60 Table 5. 6: Trade facilitation indicators for Colombia and other selected countries................................................... 60 Table 6 : International Comparisons 66 Table 6 2: International Innovation PerformanceRanking. 2002-2006 .. 1 ........................................................................................................................ ..................................................................... 73 Table 6. 3: Capacitiesinscience. technology and innovation by Department ............................................................ 78 ListofFigures Figure 1. 1: Sub-regional competitivenessinColombia ................................................................................................ v Figure 1. 2: Colombia: Geographicaldistribution of GDP vi Figure 1 .3: ........................................................................................... Average distance fkom principal industrial production to main ports ....................................................... vi Figure 2 . 1: Stages ofpublic policies for sub-regionalcompetitiveness ..................................................................... 13 ................................................ Figure 3 2: TFP: RegionaVgroupcomparisons, 1970-2005....................................................................................... Figure 3 1:TFP change inselected Latin American countries. 2001-2005 (percent) 21 Figure 3 3: Economic structure of Colombia's sub-regions,2005 (percent) .............................................................. 21 22 Figure 3 4: Concentrationof Manufacturing Firms.................................................................................................... .... Figure 3. 5: Colombia: Distributionof services by sub-region, 2005 (percent) .......................................................... 23 23 Figure 3 6: Proportion of exporting firms, 1993 and 2002 (percent) 24 Figure 3 7: Colombia: Sub-regionalincomeper capitarelative to Bogota's (percent) 25 Figure 3 8: Average incomeper capita growth by sub-regionsor group of sub-regions 25 Figure 3. 9: Sub-regionalcompetitivenessinColombia.............................................................................................. ... .......................................................................... .............................................. ............................................ Figure 3 10:TFP inlevels (2002) and growth (1994-2002) by region....................................................................... . 26 29 Figure 3. 11: Intensityinthe use offactors, 2002 ....................................................................................................... 29 Figure 4 : City-region quality of electricity services . 1 ................................................................................................ 34 Figure 4. 2: City-region: Share of manufacturing firms with ownershipofpower generator (percent)...................... 34 Figure 4. 3: City-region: Share of f m s rating electricity and transport as major or severe obstacles to their growth (percent) 34 Figure 4 4: City-region: Extent of markets (percent ) 35 Figure 4 5: City-region: Average share of sales to other regionsby manufacturing f m s (percent) ......................... Figure 4 6: City-region: Extent of markets of manufacturing f m s (percent)............................................................ 35 35 Figure 4 7: City-region: Average educationalattainment of workforce (percent ) 36 Figure 4 8: City-region: Share of f m s usingIT ........................................................................................................................................................................... ................................................................................................ ........................................................................................................ ..................................................... 36 Figure 4.9: City-region: Average share of managerialtime spent dealingwith government regulations,2006 (percent) Figure 4 10: City-region: Share of f m s agreeingthat government officials' interpretationsof laws and regulations are consistent(percent) ................................................................................................................................................ . ...................................................................................................................................................................... 37 Figure 4 11:City-region: Average share of sales and labor declaredfor tax purposes .............................................. 37 37 Figure 4 12: City-region: Tax inspections,2006........................................................................................................ .. ................................................................................................. 37 Figure 4 4: Colombia: innovation inputs and outputs by region (percent) ............................................................... Figure 4. 3: City-region: Judicial decisions (weeks) 38 38 Figure 4 5: Relative IC effects on aggregateproductivity (decompositionin logs) 43 Figure 4 6: O&P decompositioninieveis of aggregateproductivity (restrictedSolow residual) 44 Figure 4 7: Relative IC effects by group of variableson aggregateproductivity. averageproductivity and ... . ................................................. ............................ efficiencj by region . .................................................................................................................................................... 45 Figure 4. 19:Medellin:Relative IC effects on the Olley andPakes decompositioninlogs........................................ Figure 4 8: Bogota: Relative IC effects onthe Olley andPakesdecompositioninlogs .......................................... 47 47 Figure 4 Figure 4 21: Barranquilla: Relative IC effects on the Olley andPakes decompositioninlogs.................................. ..20: Cali: Relative IC effects on the Olley and Pakes decompositioninlogs ............................................... 48 49 Figure 5 . 1: Flows oftrucks on Colombian roads ....................................................................................................... 57 Figure 6 : Years ofEducation of labor force, by department . 1 ................................................................................... 70 Figure 6.2: Percentageof work force with post-secondaryeducation. by sub-region................................................ Figure 6. 3: CurrentEnrollment Rates inHigher Education....................................................................................... 70 71 Figure 6 4: Schooling vs.Income per capita. by department 72 Figure 6 5: Concentrationof Researchers by Department, 2004 74 Figure 6 6: Per capita investmentininformation andtechnology, 2006 ... ...................................................................................... ................................................................................ .................................................................... 75 Figure 6. 7: I S 0 9000:Normalized manufacturingvalue-added, 2003....................................................................... 75 Figure 7. 1:Four questionsfor sub-regionalcompetitivenessstrategies ..................................................................... 80 ListofBoxes Box 2. 1: The IrishExperience.................................................................................................................................... 10 Box 2 2: Supply Chain StrengtheninginPeru ............................................................................................................ 14 Box 2 3: Chile's Wine Industry andthe Coordination amongActors ........................................................................ .. 15 Box 4. 1: Olley andPakesDecomposition.................................................................................................................. 43 Box 5 59 Box 5 2: MeasuringLogistics Costs .. 1: The Trucking Industry inColombia ........................................................................................................................... ............................................................................................................ 61 Box 6 : Health and Competitiveness . 1 ........................................................................................................................ 68 EXECUTIVESUMMARY 1. In recent years, the Government of Colombia (GoC) has placed a high priority on competitiveness. Increasing globalization trends and Colombia's decision to increase trade integration, with the negotiation o f a free trade agreement (FTA) with the US, has led the government to focus on a complementary agenda to boost competitiveness in order to reap the benefits o f increased trade integration. A bottom-up process o f consultation, known as the Domestic Agenda (Agenda Interna, AI), was launched with the aims o f identifying key constraints at the local and sector levels and developing a set o f competitiveness-oriented measures. 2. Among the complex questions raisedwhen implementingthis set of policy actions, one key issue is how to address the sub-regional dimension of competitiveness. Going beyond the general statement o f improving competitiveness and implementing a complementary agenda for trade integration, one o f the dilemmas faced by policy makers i s where and how to direct policy efforts. In other words, competitiveness policies have a spatial dimension, in the sense that most decisions about where to direct public spending and incentives involve a spatial choice. Therefore, should rates o f return be the sole criteria to govern decisions o f resource application? Or should lagging regions receive the focus instead? Or a combination o f both? 3. This study aims at helpingthe GoC to fine-tune the mix of policies and actions to assist its regionsin meetingdevelopmentchallenges and graspingopportunitiesfrom trade liberalization. Defining competitiveness as policies and actions to increase total factor productivity (TFP), this report seeks to provide inputs for the establishment o f a strategy for sub- regional competitiveness and growth through both examining the recent literature on the main instrumentsdirected towards these objectives and evaluating sub-regional endowments, capacity, productive structure and the determinants o f productivity levels in selected regions o f Colombia. Three key areas for competitiveness are further explored: (a) overall investment climate; (b) infrastructure and logistics; and (c) humancapital and innovation. 4. The five mainmessagesof this reportare: There are signijkant differences in terms of productivity determinants across Colombian regions, suggesting that a combination of instruments-policies and actions, tailored to respond to specific needs and conditions of each sub-region are the adequate response to the challenges of sub-regional competitiveness in Colombia An econometric exercise based on the Investment Climate Survey (ICs) showed that there are significant differences in terms of productivity determinants, even restricted to a comparison among four o f the leading regiondcities o f the country. These results coincide with the international literature that suggests that the most successful experiences are based on a well crafted package o f interventions: a combination o f instruments-policies and actions, tailored to respond to specific needs and conditions o f each sub-region and leveraging on their individual endowments and capabilities and productive structure. i e Colombia faces low total factor productivity growth, particularly since the 1990s. Investment in infrastructure, measures to reduce informality, corruption and crime and improvements in technology and skills are the main elements of the agenda. An econometric analysis o f the determinants o f productivity at an aggregate level for the country shows that variables related to red tape, corruption, crime and infrastructure present the higher relative impacts on productivity. The impact o f variables linked to quality, innovation and labor seems a bit lower, but the results have the expected sign and are thus important determinants of productivity as well. e The design of competitiveness policies for Colombia's sub-regions has to take into account both the observed income or productivity differentials and the expected impact by the change of relative prices associated with the ratification of a FTA with the United States. Evidence that trade flows has an impact on firms' locations in Colombia was found; and information collected in a value-chain exercise for fruit-horticulture and glass-ceramics suggests that important changes may be taking place interms o f the location o f productive activities inthe country. The government has to carefully observe these changes when designing its sub-regional competitiveness strategy. e There is evidence that the economy is currently characterized by weak inter-industry and interregional economic linkages and unexploited and mist-matched technological capabilities and with most forward and backward links concentrated in the richest regions. Transport logistics i s clearly one o f the major factors constraining spatial interaction. Public policies may help to facilitate these new potential flows. They may include enhancing infrastructure networks, but the analysis developed inthis report confirms previous findings that special attention should be placed on: the trucking industry performance, ports organization and management, and the inspectionprocess at gateways. 0 The strengthening of the higher educatiodhuman capital and the innovation systems should be at the core of sub-regional competitiveness policies. Strengthening the relationship between universities and the private sector, exploiting technological capabilities, matching the needs o f sub-regional productive structures and so on could help ensure that scarce resources for technological development, innovation and broadly research and development (R&D) and education respond to the needs o f the productive sector, given the evidence o f a disconnect between investmentinR&D,technological capacities and GDP per capita. There are deficiencies in the quality of education as measured by international reading literacy tests. The use and investment in information and communications technologies seem also very weak compared to LAC countries not to mention that of the fast growing economies. Strengthening the Colombian National Innovation Systemshould be a key priority. 5. A word of caution is required regarding the impact of violence on the location of economic activities. Despite the fact that the overall objective o f the study relates to all Colombia's sub-regions, higher private sector involvement in some sub-regions may be difficult to achieve under lack o f good security conditions. However it is difficult to assess to what extent relatively low frequency variables, like homicides rates, may affect the location decisions o f firms in specific business districts. ii 6. It is crucial, though, to recognize the value of improved security conditions on the investmentclimate. For instance, in Chapter 5 a survey of entrepreneurs ofthe main four cities o f Colombia shows a strong negative effect o f losses associated with crime (thefts, burglary, etc) on productivity. Yet other variables like red tape and corruption have also substantial negative effects inthe same model. 7. On the other hand, the assumption of security as a prerequisite for investments, locationof firms and competitiveness - althoughvalid in general may need to be analyzed - differently in a complex security context such as Colombia. Operating in a secure environment may be a business objective and not just an expected previous condition. There are examples o f industries that became successful, among other factors, by dealing directly with questions o f poverty and inequality in the surrounding environment as part o f their business logic. They show that a particular way o f doing business may provide better security than traditional security measures used by other firms and businesses. 8. Overallthe investmentclimate in Colombia is beinginfluencedpositively by recent trends in security conditions. The effects on specific sub-regions from national improvements in security are difficult to identify, as many other local attributes enter into action to attract specific firms. In certain sub-regions o f Colombia the package for improved competitiveness should include security along with other variables like infrastructure, human capital and financing, among others. Policiesto ImproveCompetitiveness 9. Policies to increase competitiveness at the sub-regional level are designed to promote long-runeconomic growth through productivityincreases and should be tailored to sub-regionalconditions. They are a mix o f equity/social and productive/growth focus, with the weights depending on sub-regional conditions. Policies focusing on growth include several instruments from direct intervention (e.g., credit subsidies, matching grants and tax incentives to f i r m s already in or locating in areas where the government aims at boosting economic growth) to more passive policies working on the investment climate o f a sub-region, which encompass actions such as the provision o f infrastructure and/or labor training and regional marketinghranding efforts. 10. Institution building plays a decisive role in crafting sub-regional policies for competitiveness, as the identification of political economic forces and economic opportunities will always be imperfect. There are inherent difficulties for the central government or planner when trying to identify what are the key opportunities at the sub-regional level. A holistic approach should be adopted in solving the puzzle o f matching opportunities to sub-regional capacities. It has to take into account the local political economy and its capacities as it was done inColombia with the Agenda Internu (AI). Inaddition, it should check the realism o f opportunities being considered by selecting a few opportunities that are consistent with local capacities similar to the schemes adopted in Chile and Ireland, where specific sectors were 'A general caveat is inorder for this task on Colombia. As i s often the case when trying to do sub-regionalanalysis for a given country, data availability at the sub-regionallevel limits and shapes the extentand rigor ofthe particularly quantitative analysis. ... 111 selected to be emphasized by the government's actions. Colombian institutions have evolved creating an environment that represents one o f the strongest opportunities for Colombia's competitiveness and the recently created System for National Competitiveness has consolidated a very modern set o f transversal policies with the ability to respond to local and sector needs. 11. Sub-nationalgovernmentshave an importantrole in the preparationand execution of competitiveness policiesin Colombia. Better macroeconomic conditions and consequent by a greater focus on micro determinants o f growth have led, throughout the world, to an increased involvement o f sub-national governments in economic development policies. Local and regional governments are often much closer to the conditions that affect companies most, and they control many of the investments that are needed to upgrade business environments. In Colombia sub- national governments are responsible for a significant part o f the expenditures related to competitiveness, like education and transport, and may increase their participation with the expected increase intransfers. As the efficiency in spending at the sub-national level shows high heterogeneity (World Bank, 2007b), it seems important to adopt measures to improve the quality o f spending. Suggestions presented in a recent World Bank report (World Bank, 2007b), like a clearer assignment o f responsibilities in the area o f transport, work on a more complete decentralization in education and expanding un-earmarked resources for departments are key to provide conditions for more effective role o f sub-national governments in the competitiveness agenda. 12. Growth strategies are, to large extent, experimental exercises that require the implementation,monitoring and evaluation of systems; retaining flexibility in the use of instruments is decisive. In lagging regions, it is necessary to strengthen coordination mechanisms, develop capacity building for monitoring and evaluation, as public policy depends critically on the participation o f local institutions and their interaction with the central government. Sub-RegionalTrendsinProductivityand itsDeterminants 13. The average TFP for the manufacturing sector does not present significant variations across industrialagglomerationsin Colombia. However, between 1994 and 2002, TFP has grown faster in Bogota, Cartagena, Medellin and smaller urban agglomerations than in the remaining citieshub-regions. 14. Colombian sub-regions differ significantly in terms of competitiveness, ranging from those that can and are competing in international markets to those that require specialattentionfrom the publicsector. Using indicators o f economic and social development (e.g., GDP level and growth, and export orientation) and knowledge and innovation indicators, the sub regions inthe country can be grouped into four different categories. The first two groups comprise sub-regions that can compete internationally. The difference between the two is the degree o f local specializatioddiversification: Antioquia, BogotKundinamarca and Valle already show a degree o f economic diversification, while Bolivar, Boyaca and the coffee region are highly specialized sub-regions. Inaddition, Atlhtico and Santander are inbetween these two groups. The third group, called "poles for local development," is formed by 8 sub-regions, which present an intermediate level of economic development and knowledge indicators but lack iv important instruments to compete and grow. The last group lacks most o f the instruments for growth and competitiveness, requiringspecial attention (see Chapter 6). 15. An econometric exercise shows that the impact of the trade liberalization process can be an important driver of firms' locationin Colombia and that the evolution of firms' location has to be carefully monitored by the central government. The design of competitiveness policies for sub-regions in Colombia has to take into account not only observed differentials in income or productivity but also expected movements put in place by the change o f relative prices associated with making progress on trade liberalization. The Commission on Competitiveness could systematically monitor and assess these trends in order to formulate its development strategies (see Chapter 3). Figure 1. 1: Sub-regionalcompetitiveness inColombia EconomicDevelopment Source: World BankAnalysis. 16. The econometric exercise using ICs data showed that, even restricted to a comparison among four of the leading citieshegions of the country, there are significant differencesin terms of productivitydeterminants.While there are clearly common elements, such as the impact o f informality on productivity, there are also specificities o f each region that necessarily lead to different emphases in terms o f policy recommendations. Infrastructure variables, for instance, are more strongly correlated with productivity inregions like Barranquilla and Medellin, while variables related to quality, innovation and technology are relatively more important inBogota. InfrastructureandLogistics 17. Colombiapresents a widely diversifieddistributionof economic structuresacross its sub-regions. Bogota, Cali and Medellin, located in the Andean region, are its major economic hubs, responding for 75 percent o f GDP (excluding miningactivities) (Figure 1.2). Incomparison to selected peers, goods and services in Colombia travel large average distances from the production areas to the country's main ports. For example, this indicator for Colombia i s about V three times those for Brazil and Chile, five times that for Malaysia, and six times those for China, Argentina and Korea (Figure 1.3) (World Bank, 2006). Figure 1. 2: Colombia:Geographicaldistribution Figure 1.3: Average distance from principalindustrial of GDP productionto mainports Source: World Bank (2005). 18. In terms of policy instruments, infrastructure and logistics can play a fundamental role. The existing analysis concurs in that most sectors are self-sufficient at a regional scale in Colombia, with little interregional interaction, and most forward and backward linkages are concentrated inside the richest regions; these results suggest that regional polarization may be perpetuated. The result o fthe extensive research carried on by Bonet concludes that interregional links are weak indeed in Colombia. One implication i s that growth, under this economic spatial pattern, will likely not helpto reduce the current regional imbalances, but could even reinforce it. Previous studies had reached similar conclusions (see Galvis Aponte, 2001; and Roca and Romero, 2007). 19. Some analysts perceive transportation costs, associated with weak infrastructure, as one of the leading causes of the low spatial interaction. Transport infrastructure and its associated services constitute the physical link between production and consumption centers, as well as key trade gateways. 20. The main weaknesses in the freight logistics system in Colombia are not restricted to hard infrastructure, but deal with regulations and government-managed processes. A recent Bank report showed that the freight logistics complex should be approached with a broader perspective, including transport infrastructure and services, business logistics development, andtrade facilitation processes. Three main problems were found to be particularly critical: (a) the trucking industry, (b) public use ports, and (c) inspections in international gateways (see World Bank, 2006). 21. A preliminary review of two productive chains (fruit-horticulture and glass- ceramics) suggests some new patterns of firm location in Colombia. Facing increased trade vi integration, firms are willing to expandtheir capacity to produce/distribute on the Atlantic coast, which may raise the spatial interaction through growing forward and backward linkages in the Caribbean region. The logistics problems pointed out by firms include infrastructure shortages and regulatory issues. An initiative to be considered by the government is the creation of a National Logistics Council to coordinate public and private actions at the national and sub- regional levels. HumanCapitalandInnovation 22. The strengtheningof the higher education and the innovationsystems should be at the core of sub-regional competitivenesspolicies. Investment in human capital has the most powerful (although long lagged) impact on income per capita. A region with a reduced number of trained people may face skill shortages and thus weaker horizontal and transversal linkages among firms, which are critical to move up the value and knowledge chain. In Colombia, the human capital and technological endowments are located very unevenly across sub-regions. Most of the human capital and technological endowments are concentrated in the three major productive centers, Bogota, Valle and Antioquia. There are significant differences across Departmentsinthree highly correlated factors: labor skills, educational attainment, and quality of education. 23. Strengtheningthe Colombia National InnovationSystem should be a key priority: more and better spending in supportingR&D and facilitating linkages among university andindustryshould beat the top of the nationalagenda. The relationship betweenlocation of technology centers and investment inR&D is weak, indicating, to some extent, the validity of the hypothesis of the lack of productive focus of R&D activities. Eventhough Cauca, Cundinamarca and Sucre have no technology centers, they are the three top sub-regions interms of investments in R&D. Strengtheningthe relationship betweenuniversities and the private sector could help secure that the available resources respondto the needs of the productive sector. 24. Increasingand strengtheningfunding mechanisms and incentivesfor the National Innovation and Quality Systems across all areas and levels of government is also important.Centralized leadership, increasedand consolidated funding, and setting up high level incentive-driven programs-such as matching grants-are key components for an effective reform program. Demand support programs need to be complemented by strengthening the regional infrastructure that provides technological services vii CHAPTER 1 - INTRODUCTION 1.1 Inrecent years, the Government of Colombia has put competitiveness, defined as actions to increase total factor productivity, as a major priority for the country. The decision to increase trade integration, with the signing o f a free trade agreement with the United States, has led the government to focus on a complementary agenda to increase competitiveness inorder to reap the benefits of increased integration. A bottom-up process o f consultation, known as the Domestic Agenda (AgendaInternu, AI), was launched to identify the key constraints to competitiveness at the local and sector levels andto develop a set of competitiveness-oriented measures. 1.2 The Government o f Colombia (GoC) has already taken a number o f decisive steps to improve competitiveness. The following measures were adopted inthe last two years: First, the approval o f Law 962 inJuly 2005 was a significant achievement for reducing "red tape." The so- called Ley Anti-Trhites eliminated a significant number o f bureaucratic processes and prevents government agencies from creating new bureaucratic measures and from raising funds through fees charged through these procedures. The law also permits more documentation to be submitted electronically or by mail, reducing the need for in-person appearances, and the law rescinded the requirement that signatures be notarized in most bureaucratic procedures. Second, the creation of "one-stop shops" has helped streamline the processes involved in starting and operating businesses. The establishment o f the Centers for Enterprise Assistance (Centros de Atenci6n Empresarial, CAEs) in six major cities was carried out by the Colombian Confederation o f Chambers o f Commerce (Confecharas) and local governments. Third, the GoC has beenworking for more than 3 years with 17 different government agencies to simplify the procedures for import and export. Fourth, there were sustained efforts to increase financial depth, via a major reform ofthe financial system. The government reforms have aimedto deepen the domestic capital markets so that they can better provide financing and risk management tools to the private sector. Some o f these measures have been supported by the World Bank's Development Policy Loan series on business productivity and efficiency. 1.3 Despite these considerable improvements, there i s still a major pending agenda for improving the investment climate and overall competitiveness o f Colombia. The country ranks in intermediate positions in most international standards. The 2007 Doing Business Report ranks Colombia 79th out o f 175 countries for overall ease of doing business. Among the major Latin American economies, Colombia i s behind Chile (28th), Mexico (43rd), and Peru (65th), but it i s ahead of Argentina (101st), Brazil (121st) and Venezuela (164th). Despite a relatively good standing by L A C standards, Colombia needs to improve its infrastructure, particularly for productive and trade facilitation uses, as was illustrated in the Bank's Colombia Recent Economic Developments in Infrastructure (REDI) (World Bank, 2005a) and the Banks's logistics study (World Bank, 2006). Likewise, the innovation, knowledge and technological development and transfer agenda i s insufficient for Colombia's needs. 1.4 Among the complex questions raised when implementingthis set of policy actions, one key issue is how to address the sub-regional dimension o f competitiveness. Regional development policies have experimented with a variety o f initiatives that can be differentiated 1 according to their emphasis on growth, efficiency and equity (World Bank, 2005). The objectives o f regional policies usually represent a certain combination of interpersonal equity, economic efficiency and regional growth. A common misconception i s that regional equality should be the single goal o f a regional policy. Many studies that compare the performance o f regional policies focus only on the extent o f reductions in regional inequality, as well as on the so-called rates o f regional convergence. Yet regional equality, ifpursued at the expense o f other needs, may result inlower overall welfare (World Bank, 2005). Table 1. 1: Regionaldevelopment policy alternatives Social Distress and Poverty Factor Mobility StrategicGovernment Investment Alleviation Programs Programs Programs Short-TermTax Relief Labormarket policies Investmentininfrastructure FiscalSupport to Lagging Housingmarket reform Innovationandtechnology Regions Landallocationpolicies Improvingthe business environment SubsidizingWages to Offsetthe Effectso fWage Rigidity Reformingfinancialmarkets Fostering `growth poles' SocialPolicy Unemployment andtraining support Centralizedindustry oversight programs Incentives for special economic zones Migrationassistance programs Source: Adapted from World Bank(2005). 1.5 Inthis study we focus on competitiveness policies at the sub-regional level, a subset of regional development policies adopted throughout the world. The policies we consider here are those presented in the last column o f table 1, and can be summarized into the following categories: (a) investments in infrastructure, (b) policies to encourage innovation and technology adoption, and (c) policies aimed at improving the business environment. It also includes a myriad o f more direct interventions, ranging from special economic zones to attract investments to regional tax incentives and subsidies. 1.6 Going beyond the general goal o f improving competitiveness and implementing a complementary agenda for trade integration, one o f the dilemmas faced by policy makers i s where to direct these efforts. In other words, competitiveness policies have a spatial dimension, inthe sense that most decisions about where to direct public spending and incentives involve spatial choices. In principle, rates o f return should govern decisions on resource application. Investments in both physical and human capital in lagging regions can be seen as a second-best policy, as the first-best policy would be to assign public investment purely on efficiency grounds, so as to maximize national output and then carry out any desired redistribution through taxation and subsidies. It should be noted, however, that there usually are limitations on ex-post redistribution because o f limited administrative capacity or political economy reasons; beingthis the case, redistribution through direct infrastructure and human capital provision also becomes part o f a second-best policy package (De la Fuente, 2004). 2 1.7 While lagging regions today are not the most likely places to benefit from greater trade integration, the growth process of Colombia also does not assure that benefits would be redistributed across those lagging sub-regions. But the trade liberalization process may trigger important relocation movements from firms, which must be taken into account by the government. The coastal region of Colombia is often cited as a potential area of attraction for firms given a free-trade agreement (FTA) with the US, due to its proximity to the US. The expected benefits from greater integration with foreign markets may be widely distributed if the linkages between sub-regions and productive capacity are strengthened. As has been the case with other countries in the region, such as Brazil, opportunities for sub-regions to benefit more from national growth are based on improved infrastructure for regional integration. More broadly, lack of cost effective access to major centers of logistics consolidation affects the potential for sub-regions to participate inlarge national markets and international trade. 1.8 Colombia is characterized by the persistence of large disparities between Bogota and the other sub-regions. From 1975 to 2000, only Caldas, Cundinamarca and La Guajira were able to partially close the income gap with Bogota. Inmost other sub-regions, the income gap with Bogota widened. During the period 1975-2000, income grew faster than population in all sub-regions; however, La Guajira was able to double its 1975 income. When Colombia grew at its fastest rates during early seventies (the import substitution model), most sub-regions grew faster than Bogota. After 1990, Bogota grew faster thanany other sub-region inthe country. The crisis of the late-1990s affected the whole Colombian economy, particularly the sub-regions of Atlhntico and Valle. 1.9 In Colombia, manufacturing production is concentrated in the seven main cities. Nevertheless, a significant number of manufacturing enterprises are located in non- metropolitanurban agglomerations,such as small cities in the surroundings of Bogota or Cali. Around 70 percentof Colombia's manufacturingenterprises are locatedinthe metropolitan areas of Barranquilla, Bogota, Cali and Medellin, but they represent only about 55 percent of manufacturing sales revenues. The services sector is more concentrated inthe main metropolitan areas than are manufacturing firms, especially in Bogota. Services for individuals and firms are largely concentrated in Bogota, presenting an increasing concentration trend in spite of the growth of other large urban centers (e.g., Cali and Medellin). Moreover, from 1990 to 2005, the provision of business services has become particularly important in other sub-regions, such as Boyaca, Cauca, Cordoba, Cundinamarcaand Sucre. 1-10 Through an analysis of selected topics focusing on the manufacturing sector, due to data availability, this study aims to help the Government of Colombia fine-tune the mix of policies to assist its sub-regions inmeeting challenges and grasping opportunities from trade integration. It has two main objectives: a) providing the Government with a set of actionable policies tailored to the sub-regions' endowments and conditions with the aim of improving sub-regional competitiveness; and b) identifying the required public and private institutional roles and incentives for policy implementation. 3 Table 1. 2: GDP per capita, annual real GDP growth, GDP composition and educational attainment by sub-region Annual Average Real GDP Per GDP Growth GDP Composition (2005) Educational Attainment (ECH 2005) Sub-region Capita % population w/ % Population W l USD 2005 1995-2000 2000-05 Primary Manufacturing Services Average YrsOf Secondary Secondary or Schooling Education Education Tertiary Colombia 2,174 0.9% 3.3% 18% 15% 44% 6.96 23% 36% Bogota 2,793 -0.9% 4.1% 1% 17% 57% 8.96 22% 47% Santander 2,722 3.8% 5.0% 16% 24% 37% 6.43 22% 35% Antioquia 2,348 1.0% 3.4% 16% 19% 45% 6.78 23% 35% Valle 2,196 0.1% 2.4% 9% 19% 51% 8.19 25% 38% Meta 2,124 2.0% 1.9% 43% 5% 33% 6.25 25% 34% La Guajira 2,115 4.9% 3.7% 61% 1% 24% 6.36 32% 45% Clmarca 1,995 1.5% 3.9% 30% 22% 31% 6.65 26% 35% Caldas 1,727 -1.3% 5.1% 22% 14% 42% 6.17 23% 32% Atlantic0 1,722 0.6% 3.9% 5% 21% 49% 7.62 23% 38% Cesar 1,659 1.8% 6.5% 58% 4% 24% 5.60 25% 34% Tolima 1,584 1.2% 0.1% 32% 7% 38% 5.82 23% 30% Bolivar 1,563 2.0% 5.0% 13% 29% 36% 6.39 25% 36% Huila 1,557 1.2% 2.5% 38% 3% 30% 6.15 27% 35% Risaralda 1,538 -1.6% 4.7% 13% 13% 46% 6.56 24% 35% Boyaca 1,515 -0.3% 2.1% 25% 14% 40% 5.72 20% 29% Cordoba 1,466 3.9% 3.5% 41% 4% 36% 5.98 23% 32% Quindio 1,294 -1.5% 1.2% 22% 5% 51% 6.83 30% 43% Cauca 1,145 3.7% 5.2% 25% 19% 36% 6.73 24% 31% Caquetl 1,104 2.3% -0.9% 45% 3% 28% 5.12 19% 26% Nariilo 1,030 1.4% 4.8% 33% 3% 37% 5.49 17% 25% Magdalena 1,015 -0.2% 3.4% 31% 3% 43% 5.95 25% 34% N.Santander 918 1.7% 1.8% 19% 6% 48% 5.98 22% 30% Sucre 861 0.7% 3.3% 37% 3% 38% 5.62 23% 31% Choco 819 -0.3% 2.5% 37% 1% 38% 5.11 23% 29% Source: Own elaboration based on data from DANE. 1.11 The report is organized into five chapters. The next chapter reviews the main policy instruments for competitiveness and illustrates them with selected experiences in establishing and undertaking strategies for sub-regional development. Two layers are central to the analysis: (a) policies aiming at encouraging national or foreign firms to locate ina specific sub-region, and (b) instruments (e.g., accountability, budgeting and protocols) to promote coordination and cooperation among actors. The chapter also reviews the recent efforts made by the central government of Colombia on the implementation of competitiveness policies at the sub-regional level. 1.12 Chapter 3 explores the trends inregional development and competitiveness in Colombia. An econometric exercise that is presented shows that the impact of the trade liberalization process can indeed be a very important driver of firms' location in Colombia, and that the evolution of firms' locations has to be carefully followed by the central government. 1.13 Chapter 4 explores the 2006 Investment Climate Survey (ICs) on Colombia. It starts with an assessment of the different aspects of the investment climate (IC) in selected Colombian regions, and, in areas where best practices can be followed, a benchmarking exercise was conducted. It used a series of both hard-data and perception-based indicators on the investment climate in four regions - Barranquilla, Bogota, Cali and Medellin. The analysis focuses on four 4 main categories in which sub-regional variances were expected to be significant: (a) infrastructure, (b) markets, (c) skills and technology, and (d) government effectiveness. An econometric analysis was undertaken to evaluate the impact o f the I C on the productivity o f Colombian firms under the assumption that the constraints differ from region to region. Five groups o f I C variables were identified: (a) infrastructure (b) red tape, corruption and crime, (c) finance and corporate governance, (d) quality, innovation and labor skills, and (e) other control variables. 1.14 Chapter 5 reviews the economic interregional links in Colombia, which are generally perceived to be weak; the freight transport difficulties across the territory appears to be one o f the potential causes, which has reduced the potential competitiveness of the country and its sub- regions. The performance and condition o f the freight logistics system has been reviewed, and the findings confirm that Colombia is experiencing some severe shortcomings, not restricted solely to infrastructure but to a more complex range o f transport services regulation, private sector development and trade facilitation. A brief survey o f some value chains suggests that - under the new international trade scenario -the low spatial interaction that characterized Colombia's economy may experience an increase. The chapter ends with policy recommendations on the transport-logistics sector, which may contribute to enhancing sub- regional competitiveness. 1.15 Humancapitalandinnovationarekey drivers ofcompetitiveness andthey are explored in Chapter 6. After discussing trends at the national level, an analysis o f the human capital and technological endowments o f the Colombia's departments is carried out, focusing on years o f education o f the labor force by sub region, and exploring relationship between location o f researchers and investment inR&D, as well as betweenthe latter and GDP per capita. 5 CHAPTER 2 - POLICIESFOR SUB-REGIONAL COMPETITIVENESS INCOLOMBIA This chapter analyzes relevant aspects o f policies aiming to improve sub-regional competitiveness in Colombia. It reviews basic concepts and illustrates them with selected experiences in establishing and undertaking strategies for sub-regional development. Two layers are central to the analysis: (a) policies aiming to encourage national or foreign firms to locate in a specific sub-region; and (b) instruments (e.g., accountability, budgeting and protocols) to promote coordination and cooperation among actors. The chapter also reviews the recent efforts made by the central government o f Colombia on the implementation o f competitiveness policies at the sub-regional level. 2.1 Competitiveness 2.1. Policiesto increasecompetitivenessat the sub-regionallevelare those pursuedas a means of promotinglongruneconomic growththroughincreasesin productivity.These policies are thus a subset o f the policies directed toward regional development. A substantial portion o fthese regional development policies are related to social distress and poverty alleviation, and they involve equity-oriented social policies, such as transfers. Policies focusing on growth span a wide array o f instruments, from direct intervention- credit subsidies and tax incentives to firms locating inareas where the government aims at boosting economic growth - to more passive policies encompassing actions such as the provision o f infrastructure and/or labor training, regional marketing or brandingefforts, etc. Sub-regional competitiveness policies also rely on the effective integration o f economic sectors, including research and development (R&D), innovation andeducation, andthese policies require cooperation from across government levels andneighboring regions. Additionally, they often involve spatial development planning (e.g., development o f infrastructure networks and zoning) within implemented strategies. 2.2. Why should governmentspursue competitiveness policies at the sub-regionallevel? The most compelling case for regional development policies is the failure of market mechanisms to address coordination problems. These problems affect both the private and public sectors. Neither firms nor households may locate in remote regions unless infrastructure and/or skilled labor are already available. But public infrastructure investments in such regions will not pay off unless users are present to take advantage of them, while skilled workers may simply not migrate unless they have the prospects o f finding work. The problem o f coordination failure tends to justify a prominent role for the public sector in coordinating investment and production decisions of different entrepreneurse2 *For a detailed discussion, see World Bank (2007a). There are also socio-political motivations: efforts to promote growth in lagging regions are often prompted by fear of massive migration into cities or for reasons of territorial integrity. 6 2.3. In promoting the competitiveness of sub-regions, central governments sometimes have to face the trade-offs between the spatial distribution of economic activity and economic efficiency. Economic growth does not spread out as evenly throughout an area due to spatial externalities that create cost-savings and greater profits in certain regions compared with others. Whenjobs and firms' locations do not match the population distribution, lagging regions consistently concentrate unskilled population without jobs. Voters' representatives in lagging regions could promote policies to reverse agglomeration trends by deploying substantial fiscal resources to promote competitiveness o f their regions, as i s the case in some countries. As fiscal resources are limited, the required strengthening of competitiveness in leading regions may face financing shortfalls and reduce competitiveness of the whole economy. This does not meanthat most resources should be devoted to leading regions, but it highlights the trade-offs that naturally arise due to the forces behind economic agglomeration. 2.4. Establishing policies at the sub-regional level is further complicated by the interdependency and required interaction of many actors in the development of sub- regional strategies. The achievement o f policy objectives in the sub-regions often requires the intervention of different levels o f government with their respective "weights" being determined by the degree o f decentralization of a specific country. It also requires the participation of different entities (e.g., private and public-private agencies). Such complex arrangements lead to the establishment o f contracts between government levels and to the development of coordination mechanisms with non-governmental agencies3 2.5. Furthermore, competitivenessshould not be seen as a static concept - it changes over time as a reaction to exogenous impacts; policies should take this into account. One such impact is, o f course, related to the trade liberalization process. The design o f competitiveness policies for sub-regions in Colombia, for instance, should take into account not only the observed differentials inincome or productivity but also the expected movements put in place by the change o f relative prices associated with a free trade agreement with the US and other countries. From a theoretical point o f view, there are different views on how the opening up of an economy affects regional competitiveness. Based on New Economic Geography models, one can argue that a liberalization of trade with the U S makes locating closer to the US market more profitable, thereby shiftingfirms toward the Atlantic coast. From an empirical point o f view, it has been observed a remarkable persistence in the pattern o f industry regional distribution over long periods o f time, even after significant changes in the economic en~ironment.~Recent evidence from Mexico before and after NAFTA (Aroca, Bosch and Maloney, 2005) concludes that the post-liberalization period i s not especially tied to moving business activity closer to the U S border. 2.2 PolicyInstrumentsfor Competitivenessat the Sub-RegionalLevelandSome Lessonsfrom the InternationalExperience See OECD (2007). The case of Brazil is interesting, with a seemingly unbending regional distribution of income even after some changesinthe patternof firms' geographicaldistributionafter the liberalization ofthe economy inthe early 1990s. 7 2.6. Below, the main policies and instruments that are used to promote competitiveness at the sub-regional level are discussed. Beginning with more general policies, such as infrastructure and human capital, the discussion then moves on to more specific and targeted interventions. Country cases are provided as examples o f successes and failures o f each specific policy. Infrastructure 2.7. Improved infrastructure and logistics may bring significant short- and long-term benefits to a region. Infrastructure investments remove some o f the characteristics that inhibit potential investors in lagging regions, such as poor transportation links, lack o f suitable sites for expansion and poor telecommunication networks. Investments in physical capital in lagging regions can be seen as a second-best policy, as the first-best policy would be to assign public investment purely on efficiency grounds so as to maximize national output, and then carry out any desired redistribution through taxes and subsidies. However, when there are limitations for ex-post redistribution, because o f limited administrative capacity or political economy reasons, redistribution through direct infrastructure provision becomes part o f a second-best policy package. Recent literature suggests that infrastructure investments are likely to have a strong impact on competitiveness in areas where infrastructure i s scarce, but then the impact can fall abruptly when there i s already significant infrastructure. A recent survey by de la Fuente, cited in World Bank (2007a), concludes that "there are sufficient indications that public infrastructure investments contribute significantly to productivity growth, at least inregions where a saturation point has not been reached. The returns are quite high when infrastructure is scarce and basic networks have not been completed but fall sharply thereafter." 2.8. It is not only important to take into account hard infrastructure, but also soft aspects of logistics.A recent World Bank (2005a) report found that the main weaknesses inthe freight logistics system in Colombia are not restricted to hard infrastructure, but deal with regulations and government-managed processes. The report showed that the freight logistics complex should be approached with a broader perspective, including transport infrastructure and services, business logistics development and trade facilitation processes. Three main problems were detected as particularly critical: (a) the trucking industry, (b) public use ports, and (c) inspections in international gateways (see Chapter 5). A report on the same topic in Argentina also showed that a mix o f investments in hard infrastructure and improvements in soft logistics was necessary in the country; the three areas highlighted in the report that need simultaneous action are trade facilitation, business logistics, andtransportation, infrastructureand services.' 2.9. The European experience is an example of successful policy experiment using infrastructureas an instrumentfor regionalcompetitiveness.The Braziliancase, even with some well known mistakes, also offers good examples. Key instruments for developing regional competitiveness inthe EU are the Structural Funds, which integrate policies destined to improve the competitiveness and productivity o f regions in order to expand income over the long-term. These funds have been crucial to supporting investments in infrastructure, as well as inother areas like human capital, innovation, etc. InBrazil, opportunities for the non-subsidized sub-regions arose from better transport integration. Today, Brazil's most competitive regions are medium-sized urban agglomerations that have attracted firms that left large cities. Infrastructure See World Bank (2006). 8 investments designed to integrate the national economy allowed companies from other sub- regions to expand access to markets. However, mistakes were committed as well, especially in the 1970s and 1980s, as shown by some mega unfinished projects, like the Transamazonica in the Northern region andthe Fura-Fila inSBo Paulo. Human Capital and Innovation Policies 2.10. It is widely accepted today that the ability of regional economies to withstand competition and to adapt to technological change is related to their capacity to innovate. Indeed, innovation and flexibility are the keys to success in coping with globalization (Sepic, 2005). Innovation does not necessarily equal high-endtechnologies but rather any improvement which could be introduced at the level of local firm production, marketing, management or organizational systems. Evidence also exists that regional competitiveness can be enhanced by targeting investments on the quality o f the regional workforce. Education and skill level are powerful determinants for regional competitive advantage (World Bank, 2004~). 2.11. The notion that innovation, understood as both the creation o f new products, devices, methods and processes, and the adaptatiodimprovement o f existing products, devices, methods and processes, represents the lifeblood of competitive advantage is strongly supported by both economic growth theory, and policymaker thinking with respect to local development in both European and non-European OECD countries. Thus, with respect to growth theory, the basic model o f growth which economists use, known as the Solow-Swan model after Solow (1956) and Swan (1956), implies that an economy's competitive position is ultimately doomed to stagnation inthe absence o f innovation. (Roberts and Zhang, 2007). 2.12. There are some caveats, however: education is a long-term policy, with lagged impact on competitiveness, while fiscal costs are borne in the meantime. The financing o f costs may even lead to a decline in competitiveness in the short-run, given the fiscal envelope. Another oft-noted caveat points to the fact that merely increasing the skill level o f a group o f workers may not be effective in promoting growth in a lagging region if it leads to the out- migration of better-skilled workers. Complementary policies, such as investments, etc, should thus be adopted. Finally, innovation requires trust: firms must be flexible and fast in their reactions to changes in markets and technologies, requiring them to collaborate with others and face risks that are not entirely under their control (Nooteboom, 2006). The Business Environment 2.13. Adverse business climates are another possible constraint on economic development in lagging regions. They include soft components like business regulations, credit access, and trade facilitation, and they form part o f any competitiveness policy, with most actions being undertaken by a combination o f central and sub-national governments. Certain elements o f the business climate, such as the legislation governing taxation, labor relations, and trade, are supposed to be applied in a uniform manner throughout the country. But administrative practices usually vary among regions, even when carried out by officials o f the central government. Local governments have beenincreasingly focused on actions to overcome regulatory and bureaucratic obstacles as a means o f attracting new investments. 9 2.14. Building a sound environment for the private sector to operate requires the implementationof a programof reformscoveringdifferentareas. Innovation is decisive, but other microeconomic reforms are also relevant for productivity growth. Since there is heterogeneity among firms in productivity levels and rates of growth, reallocation of resources from low-productivity to high-productivity firms leads to an increase in productivity level and rate of growth. As new firms enter and less efficient ones leave the market, a higher productivity i s achieved, not only directly, but also as a result of more competition. Fairly inefficient factor and product markets, as well as high costs of entry and exit may lead firms to incurring in otherwise unnecessary adjustment costs whenever a shock hits an economy. See, among others, Caballero, Engel and Micco (2004) and Caballero, Cowan, Engel and Micco (2004) for a discussion of the role of microeconomic flexibility on productivity growth inLatin America and inChile intheendofthe 1990s. 2.15. What gets measured gets done. Following the success of the global Doing Business report, publishedannually by the World Bank group, there has been an increasing interest inthe elaboration of national Doing Business reports, allowing for benchmarking the business environment at the sub-national level. By allowing comparisons within a country, these reports are a potent benchmarking exercise, as they often show that the solutions to improving the business environment are "around the corner". InLatin America, reports were already published for Mexico and Brazil, inboth cases displaying wide variations inthe ease of doing business and helping to trigger reforms at the sub-national and national levels in areas like opening and closing a business, paying taxes, etc. A similar report is currently being prepared for Colombia, covering twelve different cities. Other Policies There is a wide array of policies that target sub regional competitiveness and can be implementedeither by local or by central government. This report will not analyze all of them butrather highlight some that are more frequently used. 10 Clusters The cluster approach offers a practicalframework for policy makers to organizepublic and private actions (social capital) centered on competitivemarketforces. The "cluster approach" i s well known through the work o f Michael Porter and i s particularly useful for local governments, as it offers a pragmatic course o f action for action plans to foster competitiveness. While there are debates about the concept and theory o f the cluster approach6, many o fthe policy recommendations from the cluster approach (e.g. emphasis on private sector networking, regulatory environment, constant learning, etc.) are consistent with the emphasis on innovation and human capital accumulation. This tends to be true provided the focus is kept on supporting existing or emerging clusters inthe local economy, rather than trying to create new ones. Tax Incentives 2.16. Credit subsidies and tax incentives to firms locating in areas where growth is desiredare widely used instruments.The debate about credit subsidies and tax incentives is as wide as is their use. The two main problems emphasized in the literature are the fiscal costs to finance the subsidies and the possible deadweight effect, inthe sense that the investments might have occurred anyway. The lesson seems to be that aligning industrial policies with regional strengths and market forces is a complex task that most likely will not pay back. Subsidies and tax-breaks have proved to perform poorly in term o f promoting competitiveness in lagging regions. 2.17. The case of Brazil illustratesthe problems of relying mostly on subsidies from the central governmentto promoteselected industrialsectors in laggingregions; Korea,on the other hand, shows positive results,that could be relatedto other aspects as well. After 40 years o f large repeated subsidies, income inequality between Brazilian sub-regions remains high, while the fiscal cost o f industrial policies puts pressure on the federal budgets8As inthe case o f Colombia, subsidies inBrazil have helpedto relocate economic activity towards lagging regions, but have to be maintained for a long period of time (the incentives for Zona Franca de Manaus are constitutionally mandated, for example).' Carvalho, et. al. (2006) and Lall, et. al. (2004) show that fiscal incentives are weak instruments to promote the location o f economic activities in lagging regions. Targeted protectionist policies may accentuate efficiency problems and divert opportunities arising from trends in spatial agglomeration. Competitiveness inKorea i s based on strong government intervention and orientation toward foreign markets. The central Government deliberately spread industries evenly across the Koreanterritory, setting up strategic partnerships with chaebols (industrial conglomerates), encouraging them to participate by providing tax- breaks, training subsidies and tariff exceptions. Later on, inthe 1980s, the government reoriented 6 Martin, R., Sunley, P. (2003). "Deconstructing clusters: chaotic concept or policy panacea?," Journal of Economic Geography, 3: 5-35. 7 Hon and Fallon's 2002 review o f regional development incentives concluded that, in general, regional incentives merely "inject resources into places that are prone to failure." * Tax expenditures were estimated in 2004 to reach 2.5 percent o f GDP in Brazil, o f which 1.4 percent o f GDP is related to economic affairs (World Bank, 2007b) 9 In Colombia, Ley PCiez (Law No. 218 of 1995) was enacted one year after the river PBez flooded some of its surroundings. I t includes tax exemptions when investments are made in the following sectors: agriculture, construction, manufacturing, miningand tourism. 11 its industrial policy toward high-technology poles. Policies aimed at sustaining a more balanced sub-regional development through spreading out economic poles and supporting rural sub- regions. The comparative advantage o f Korea's urban centers and the mobility o f high-skilled workers ensured that sector shifts did not contribute to worsening regional inequality problems. The country succeeded by strategically reacting to external shocks by changing the pattern and level o f public expenditure, and through its unique combination o f economic openness and strong state institutions. Nevertheless, sub-regional disparities persist. Sector Structure: Supply Chains 2.18. The industrial structure of a region will influence competitiveness, as high value- added sectors will have more influence on regional growth and regional GDP than low added-value sectors. In other words, the competitiveness o f a region will be influenced by the productivity o f its sectors o f activity. The sectors themselves will be influenced by the intra- sector productivity. For example, there can be large differences between high-tech sectors and traditional metal-bashing activities. Another example i s the service sector, where one can have tourism services not associated with particularly highproductivity levels or, on the opposite end, financial and business services characterized by the highest productivity. This i s also true for rural areas where one can find regions with low agriculture productivity and high employment or, on the opposite end, regions with highlevels o f agriculture productivity. TheInstitutional Setting: A Cross-Cutting Theme 2.19. Institutions for sub-regional competitiveness usually refer to necessary arrangements to be undertaken by a country so that the private sector development in specific geographical areas is promoted. Institutions, inthe tradition o fNorth(1984, 1991) and Williamson (1985, 1996), are comprised o f all organizations deployed by either the private or public sectors; they are also made up o f the interaction protocols between different actors and the behavior o f the private sector inits efforts to compete indomestic and international markets. The following are three types o f institutions for sub-regional competitiveness policies: (a) public entities at the level o f central government, (b) semi-private entities at the level o f central government, and (c) private entities. Figure 2.1 shows the four stages for a successful sub- regional competitiveness strategy. First, it i s necessary to establish the institutional mechanisms to undertake such a strategy. Second, specific instruments are deployed and integrated based on the protocols created to promote cooperation and coordination. 12 Figure 2. 1:Stages ofpublic policies for sub-regional competitiveness 1-1 Settingthe National Policy Sub-Regional FineTuning Agenda Implementation Implementation Source: World Bank Analysis. 2.20. Promoting competitiveness relies heavily on the possibility of diversifying production in order to, first, reach out for international markets and, then, promote specialization." Diversification o f supply can be reached through different mechanisms, and institutional aspects can play an important role in this regard, such as technological spillovers induced by strong R&D centers or universities, specific fiscal incentives like free trade zones, and clusters. The establishment o f clusters relies on the ability to strengthen cooperation and coordination mechanisms among firms in a specific sector and region. The role o f the government i s to identify the actions that serve to actually improve the ways that firms interact and serve each other in the cluster. The development of clusters promotes significant productiveness derived from the collective action o f its participants. It promotes productive chains and processes o f innovation. In these cases, the public intervention aims principally at solving coordination and cooperation problems. 2.21. Peru and Chile offer two good examples of coordination and cooperation between government agencies and the private sector. Chile is an example o f best practice insuccessful inter-agency coordination. The example o f the wine industry illustrates this success quite well (Box 2.2). The Peruvian export promotion agency, called PROMPEX, conducts the commercial promotion activities o f Peruvian exports abroad. It also participated in the issuing o f sanitary standards, technical production norms, and the definition of best agricultural practices, which was later adopted by the Peruvianauthorities. The second organization, Frio Aereo, i s a group o f Peruvian exporters o f perishable products, such as vegetables, h i t s , and flowers, whose goal is to improve transport (mainly air) logistics. loSee CAF (2006). 13 2.22. Other interestingcases in Brazil are successful state-led initiatives, such as Bahia's cerrado and those supportingthe development of the agribusinessand footwear sectors in the state of Ceari. The common feature o f these cases is that they seem to have focused on improving the supply o f inputs for the development o f local comparative advantages (i-e., infrastructure, labor training and technology), rather than trying to direct product market outcome^.'^ 2.23. In sum, institutionalbuildingplays a decisive role in crafting sub-regionalpolicies for competitiveness, as the identification of political economy forces and economic opportunities will always be imperfect. There are inherent difficulties for the central government or planner when aiming at identifying what are the key opportunities at the sub- regional level. First, it has to balance the ability o f sub-regions to undertake the strategies with the fact that some agents will push hard for opportunities. It is difficult for the government to know whether the sub-regions are actually capable o f handlingthe responsibilities. By selecting those that provide information it could miss highly profitable alternatives. By selecting best alternatives technically it could end-up with policies that are inconsistent with sub-regional capacities. A holistic approach should be adopted in solving the puzzle o f matching opportunities to sub-regions. On the one hand, the central government has to consult the local political economy and its capacities, as was done in Colombia with the Internal Agenda, a large effort aimed at identifying sub-regional opportunities based on the regional political economy. On the other hand, it should check the realism o f opportunities being considered by trying to select few opportunities that are consistent with capacities at the local level. 'I World Horticultural Trade & US. Export Opportunities: World Asparagus Situation & Outlook, Foreign Agricultural Service, U.S. Department of Agriculture (August 2005) at 1 (data provided for 2004). The United States "is Peru's top market, accountingfor 75 percentof Peru's fresh asparagusexports in2004." 12See Rodriguez and Humberto (2004). 13For details on local economic development illustrated with cases at the state-level, see World Bank (2005). 14 2.24. A tailor-made approach to regional development based on the diagnosis of a particular region's comparative advantages and the constraints on its development seems to be the most sensible strategy. Sub-regions have different characteristics (urban, rural, industrial, intermediate, etc) that demand specific and diverse policy and investments needs. Therefore, to improve sub-regional competitiveness, policies must be capable o f adapting to these different needs. In each stage, a non-trivial conjunction o f ingredients i s necessary. For example, coming up with a successful strategy for coordination and cooperation among agencies requires more than merely the willingness to do it. Other stages and their components also imply finding the right institutional arrangement, which could be either the organizations themselves or the protocols usedto group together their actions. 2.25. Finally, as growth strategies are, to large extent, experimental exercises, the implementation of monitoring and evaluation systems, and retaining flexibility inthe use of instruments, is decisive. In lagging regions it is necessary to develop capacity-building for monitoring and evaluation because the policies, as previously discussed, depend critically on the local institutions' participation and interaction with the central government. Eventhe best system l4Visser (2004). 15 of monitoring and evaluation will be of little use if there is no flexibility in adapting the instruments. 2.3 Colombia'sInstitutionsfor Sub-RegionalCompetitiveness 2.26. Institutions for competitiveness in Colombia have existed for more than twenty years. Several years exchanging best practices on doing business with international partners led to public agencies in charge of trade promotion and competitiveness with similar structure and capacity level of their international peers. Trade promotion has included subsidies to specific sectors, commercial missions, datacenters, subsidized credit lines, and many other instruments. Rather than evaluating what has worked and what has not, it is more important to recognize that institutions have evolved creating an environment that represent one of the strongest opportunities for Colombia's competitiveness. The recently created System for National Competitiveness has consolidated a very modern set of transversal policies with the ability to respondto local and sector demands. 2.27. The ColombianEconomicModernizationProgramof the early 1990s was intended to make the export sector one of the engines of economic growth. These constitutional reforms, in addition to reforms inthe financial sector, banking, labor and trade, were promoted, and institutions were transformed createdto make economic modernization viable. The Strategic Export Plan of 1999-2009 set the road map for strengthening the Colombian production sector and steering it toward the international market. 2.28. In 2006 the National Productivity and Competitiveness Committee (NPCC) was created. This entity has concentratedits efforts on firm-level competitiveness by improvingthe investment climate, increasing and diversifying exportable goods in an effort to consolidate and increase foreign investment. This scheme i s coordinated by the National Competitive Commission (CNC), which is part of the Senior Advisory for the Competitiveness and Productivity. Its main purpose i s to recommend to the national government a general and sector policy on competitiveness, productivity and foreign trade ingoods, technology and services. 2.29. Colombia has many agencies in charge of promoting the country's competitiveness, such as: 0 Proexportis responsible for promoting non-traditional exports; it provides support and comprehensive advice on international marketing to Colombian entrepreneurs. It also identifies marketopportunities, devises market-penetration strategies, and develops trade fairs inColombia andabroad. Bancoldex (Foreign Trade Bank of Colombia) is a semi-public entity that operates as second tier bank offering financial services to companies involved in Colombia's foreign trade. Through partnerbanks, it offers financing for importers of Colombian goods and services. 0 The Ministry of Commerceis responsible for leading international negotiations related to investment inthe framework of the WTO; identify opportunities for improvement of the legal framework oriented towards investment climate. Coinvertir (The Invest in Colombia Corporation) is a semi-public entity that promotes and facilitates the development and consolidation of foreign investment initiatives in Colombia. Itprovides legal assistance, economic informationand direct supportto potential investors. 16 0 Red Colombia Compite (Colombia Competes Network) was designed as a National Productivity and Competitiveness Policy program whose basic objective i s to preserve and boost the positive aspects o f the production environment, to correct or eliminate the factors that hamper efficiency, and to introduce the elements needed to enhance and modernize that environment as a function o fthe demands o f the national and international markets. 0 Colcienciasis a network supported by the National Science and Technology System. Itis incharge o ftechnology innovations for newproducts andprocesses as well as training to builda new business culture that is innovative and capable o f assuming the challenges of international trade. 2.30. The current administration has sought to forge links among current agents, programs and instruments. Efforts are focused on the following activities: linking research centers and export competitiveness agreements; increasing business sector interaction with universities, including technical training; creating financial instruments for technological development; linking the Science and Technology Network with the National Learning Service (SENA); andpreparingtraining and certificationprojects ininformation technology. 2.3 1. The adoption of pro-exportpublicpoliciesis a key component ofthese entities,butit i s not enough. The public sector will have to play an important role in correcting market imperfections, strengthening the institutional mechanisms that favor discoveries, and facilitating the collective action of companies that take part in productive clusters that are hampering the innovative process. Inaddition, the state should try to attract foreign direct investment inorder to maximize the spillover effects on the local economies and provide productivity benefits. To makeuse of the benefits that this investment can contribute to the sub-regional economies, there must be improvements to the capacity of the domestic absorption of new organizational processes and productive technologies. 2.32. Sub-national governmentshave an important role in the preparationand execution of competitiveness policies in Colombia. Better macroeconomic conditions and consequent greater focus on micro determinants o f growth have led, throughout the world, to an increased involvement o f sub-national government in economic development policies. Given the full spectrum of microeconomic factors that influence the quality o f the business environment and ultimately productivity, the number o f relevant decision makers naturally increases. The central government i s o f course still a key actor, but no longer the only one. Local and regional governments are often much closer to the conditions that affect companies most, and they control many o f the investments that are needed to upgrade business environments. In Colombia sub- national governments are responsible for a significant part o f the expenditures related to competitiveness, like education and transport, and may increase their participation with the expected increase intransfers. As the efficiency in spending at the sub-national level shows high heterogeneity (World Bank 2007), it seems important to adopt measures to improve quality o f spending. Suggestions presented in a recent World Bank report (WB 2007), like a clearer assignment o f responsibilities inthe area o f transport, work on a more complete decentralization in education and expanding un-earmarked resources for departments are key to provide conditions for more effective role o f sub-national governments inthe competitiveness agenda. 17 2.2. It is essential that the sub-regions consolidate a favorable environment for the discoveryof new activitiesthat encourage innovation.The promotion of innovation systems is a mechanismthat promotes the coordination and cooperation among companies, universities and public entities. This symbiosis can generate a major productive diversification and expand the export of goods and services. 2.33. A more active strategy aimed at promotingFDI can be worthwhile. Foreign direct investment generates the transfer of tangible and intangible assets, which brings along technology and training to the workforce, creates jobs, develops production processes, and strengthens trade links and the country's export capacity, thereby improving its competitiveness. More importantly, quality FDIhelps to create spillovers that strengthenclusters in specific areas by establishing a supply chain andhelping to improve highly qualified inputsthat can be used in other related industries, such as a stronger pool of engineers. Colombia has made substantial efforts at improving the investment conditions and climate for Colombians and foreigners wishing to invest inthe country. This improvement should be complementedby a more strategic quest for investment partners, who may needmore thanmerely transversalconditions. 2.34. The design and implementation of programs to strengthen the competitivenessof SMEs and MSMEs in laggingregionsis based on stronger use of technology. Sub-regions with lower human capital, weaker education systems, and fewer technology and computing resources may still be the most affected by the lack of connectivity and competitive base. Today, higher competitiveness of the agricultural sector, a very important sector for the development of Colombia's lagging sub-regions as clearly shown by theAgenda Interna, requires the adoption of technology in the production process, as well as communications and information. Therefore, sub-regions should increasetheir use of informationtechnologies for education inrural areas and provide training on educational-entrepreneurialactivities based on computing. 2.35. Colombian entrepreneurs require a great deal of training in information and communications technologies. New technologies and an increased adoption of technological innovation are essential elements for increasing productivity and improving competition. A regional study is required to identify the main challenges and needs facing information and communications technology enterprises resulting from regional competition. This initiative will help to define the policies and measures aimed at implementing adequate standards for participation ininternational markets. 18 CHAPTER3 - COMPETITIVENESSOF THE SUB-REGIONSOF COLOMBIA This chapter presents the trends in regional development and competitiveness in Colombia. It starts with a discussion of national trends in total factor productivity, followed by an overall description of sub-regional economic structure, showing its diversity across sub-regions, followed by a discussion on trends and regional convergence inthe country. Next, it focuses on competitiveness and total factor productivity (TFP) at the sub-regional level. Finally, an econometric exercise is presented, focusing on the drivers of firms' location inColombia. 3.1 Productivityand Competitivenessat the NationalLevel 3.1. Colombiahas been lagginginterms of its productivityperformancein the lastthree decades(Table 3.1 andFigure 3.1). Its performance since 1995has been quitedisappointing. For the periods 1996-2000 and 2001-05, TFP was -1.40 and 1.36, respectively. Colombia's productivity performance was below the LAC average for the period 1996-2000 and about the regional average for the 2001-2005 period. As for the relevance of TFP in explaining growth fluctuations, the last row in Table 3.1 indicates that the share of cross-country growth fluctuations explained by TFP would have rangedbetween 56 percent inthe 1981-85 periodand 91 percent in the 1991-95 period. For the whole period being considered, the average would be 75 percent (Le., for the LAC sample, differences inTFP may explain almost 75 percent of cross- country differences in growth rates). Ifthe average TFP rate over the period 1971-2005 were to be computed, Colombia would have experienced an average productivity growth of 1.5 percent per year. This rate would be similar to that for Sub-SaharanAfricanand below that for East Asia (3.8 percent) and the OECD (2.8 percent). Thus the urgency for Colombia to improve conditions to secure higher TFP growth. 3.2. Yet LAC performance was generally disappointing when compared to other regions.Figure 3.2 presents results of the exercise for the median country of each regiodgroup of reference. It suggests that TFP growth inLAC has not beenparticularly high. Infact, with the exception of the periods 1971-75 and 1991-95, LAC'STFP growth rates have beenbelow those of most of the other regions. For example, in the period 2001-2005, the only regiodgroup that had a TFP growth lower than Latin America (1.1 percent per year, on average) was the OECD (0.77 percent per year, on average). The opposite happened to the productivity performance of East and South Asia, the Middle East and North Africa and even Sub-SaharanAfrica, where the average TFP growth rate was around 2 percent per year. Similarly, over 1976-1980, TFP growth was around 2 percent per year for Latin America, but this was a period when East Asia had a TFP growth rate above 4 percent per year and the Middle East andNorth Africa, South Asia and the OECD had TFP growth rates of around 3 percent per year. More dramatically, during 1981 - 85 and 1996-2000, LAC i s clearly the worst performer in the sample with average annual TFP growth rates of -1.83 percent and 0.20 percent, respectively. 19 Table 3. 1: Total factor productivity inLatinAmerica, selected countries Country Growth Period Component 1970-75 1975-80 1981-85 1985-90 1991-95 1996-2000 2001-05 Argentina GDP 3.10 2.81 -2.54 -0.47 6.55 2.58 1.99 Capital -0.84 0.72 -1.30 -1.66 0.70 2.09 -0.73 Labor 1.53 0.93 1.24 1.40 1.70 1.40 1.39 TFP 2.5 1 1.97 -2.76 -0.65 5.25 0.90 1.45 Bolivia GDP 5.79 2.05 -1.93 2.21 4.10 3.44 3.01 Capital -3.10 -1.29 -3.18 -1.83 0.93 4.26 0.88 Labor 2.43 2.54 2.40 2.56 2.35 2.40 2.45 TFP 5.18 0.77 -2.49 1.10 2.21 0.42 1.08 Brazil GDP 10.27 6.67 1.09 2.01 3.13 2.24 2.19 Capital 1.95 3.96 1.08 1.73 0.77 1.68 1.06 Labor 3.09 3.11 2.66 2.37 2.30 2.25 1.74 TFP 7.58 3.27 -1.02 -0.14 1.36 0.19 0.68 Chile GDP -1.36 7.26 0.89 6.73 8.69 4.16 4.39 Capital -2.63 -1.92 -1.07 2.09 6.58 7.26 5.46 Labor 2.59 2.69 2.23 1.94 1.71 1.75 1.75 TFP -1.81 6.46 0.01 4.73 4.98 0.15 1.11 Colombia GDP 5.65 5.37 2.24 4.94 4.13 0.92 3.42 Capital -1.39 0.29 2.12 2.60 4.35 2.44 2.07 Labor 3.23 3.27 3.12 2.53 2.44 2.25 2.07 TFP 4.04 3.15 -0.52 2.39 1.03 -1.40 1.36 Costa Rica GDP 6.04 5.24 0.28 4.59 5.47 4.93 3.71 Capital -1.07 2.41 -0.62 2.25 4.05 4.72 4.56 Labor 3.74 4.05 3.39 2.69 2.89 3.30 2.89 TFP 3.60 1.64 -2.03 2.02 2.27 1.24 0.37 Ecuador GDP 8.71 5.27 1.37 2.73 2.67 0.94 4.95 Capital 3.21 6.19 2.43 1.06 1.51 0.58 1.72 Labor 3.26 3.29 3.34 3.21 2.89 1.99 1.91 TFP 5.47 0.96 -1.65 0.27 0.27 -0.55 3.11 ElSalvador GDP 4.61 -0.02 -2.78 2.06 6.18 3.06 2.19 Capital -2.74 -0.41 -3.13 -1.52 2.33 3.26 3.03 Labor 2.95 2.36 1.12 2.26 3.19 2.55 2.28 TFP 4.05 -1.22 -2.11 1.39 3.35 0.21 -0.41 Jamaica GDP 1.77 -3.26 0.40 4.99 2.39 -0.08 1.52 Capital 0.05 -3.06 -0.99 0.66 3.12 1.91 0.60 Labor 2.13 2.79 2.85 1.08 1.20 1.05 1.15 TFP 0.47 -3.71 -0.91 4.08 0.42 -1.47 0.59 Mexico GDP 6.26 7.11 1.94 1.68 1.53 5.45 1.82 Capital -0.01 2.33 2.18 0.65 2.43 3.28 3.35 Labor 3.19 3.27 3.33 3.26 2.78 2.13 1.71 TFP 4.5 1 4.26 -0.87 -0.41 -1.09 2.81 -0.63 Panama GDP n.a n.a 3.44 -0.67 5.49 4.63 4.16 Capital n.a n.a -3.21 -4.20 0.38 3.09 1.03 Labor n.a n.a 3.19 2.86 2.61 2.37 2.18 TFP n.a n.a 1.98 -1.62 3.48 2.07 2.29 Paraguay GDP 6.72 11.07 1.67 3.89 3.24 0.72 1.90 Capital -0.60 6.79 5.32 3.31 3.66 2.21 -0.53 Labor 3.17 4.00 3.17 3.38 2.93 3.27 3.04 TFP 5.48 5.65 -2.60 0.55 -0.07 -2.01 0.68 20 Table 3. 1: Total factor productivity inLatinAmerica, selected countries (cont) Country Growth Period Component 1970-75 1975-80 1981-85 1985-90 1991-95 1996-2000 2001-05 Peru GDP 4.99 2.28 0.32 -1.90 5.48 2.46 4.09 Capital -1.39 -0.63 0.09 0.04 1.59 3.67 1.63 Labor 3.09 3.18 3.05 2.79 2.37 2.18 2.10 TFP 3.87 0.78 -1.43 -3.49 3.45 -0.38 2.20 Trini. & Tobago GDP 2.74 7.88 -2.25 -2.24 1.39 4.95 7.68 Capital -0.42 2.37 1.85 -2.08 2.72 2.83 2.38 Labor 2.15 2.44 1.92 0.47 1.67 1.93 1.16 TFP 1.38 5.46 -4.14 -1.92 -0.61 2.74 6.14 Uruguay - . GDP 1.50 4.55 -3.78 3.87 3.94 2.11 0.99 Capital -3.85 0.75 -1.75 -2.34 0.74 2.05 -1.62 Labor -0.03 0.57 0.54 0.69 0.79 0.61 0.81 TFP 3.14 3.90 -3.36 4.46 3.17 0.89 1.20 Venezuela, RB GDP 2.95 2.45 -0.93 2.59 3.45 0.75 2.30 Capital -2.05 1.92 -1.14 -1.36 0.19 3.15 1.17 Labor 4.34 4.41 3.50 2.75 2.77 2.61 2.45 TFP 1.61 -0.79 -2.24 1.77 1.89 -2.11 0.46 YOof growthfluctuations explained by TFP 62 64 56 84 91 78 86 Note: n.a means not available. Source: World Bank. (2007e). Calculations use WDI data for GDP, labor growth and gross investment. Capital is computed using an inventory rule assuming a 7 percent depreciation o f the capital stock and an initial capital to output ratio o f 5. The share o f capital comes from Loayza, et. al. (2004). Figure 3. 1: TFP change inselected Latin American Figure 3. 2: TFP: RegionaVgroup comparisons, countries, 2001-2005 (percent) 1970-2005 7 6 5 - 4 0 E 3 & 2 1 0 -I 71-75 76-80 81-85 86-90 91-95 96-00 01-05 a > v - 0 - Source: World Bank. (2007e). Source: World Bank. (2007e). 3.2 Sub-RegionalEconomicStructure 3.3. Colombia has a widely diversified distribution of economic structures across sub- regions.Figure 3.3 presents the structure of production inColombia's sub-regions, broken down into five economic sectors - (a) primary, (b) manufacturing, (c) services, (d) commerce and (e) construction and government. Sub-regions are sorted by the sum of non-primaryhon- government sectors to give a sense of specialization in more modeddynamic sectors, which 21 should initiate the challenge o f defining strengths and weaknesses o f the largely different sub- regions. Figure 3. 3: Economic structure of Colombia's sub-regions, 2005 (percent) 100% 80% Government 60% 40% 20% 1Commerce& Construction 0% Source: Based on DANE'SSub-Regional Accounts. 3.4. Manufacturing production is concentrated in seven main Colombian cities. However, an important number of manufacturing firms are located in non-metropolitan urban agglomerations,such as smaller cities near Bogota or Cali. Almost 70 percent of the manufacturing firms are located in Colombia's four main metropolitan areas (Le., Barranquilla, Bogota, Cali and Medellin), but they represent only around 55 percent o f manufacturing revenues. A relevant number o f firms in the food & beverage and metal sectors are located in small urban areas in Cundinamarca, Norte de Santander, Tolima and Valle (Table 3.2). An instrument to assess industry concentration i s the Herfindahl-Hirschman index (HHI); as sub- regions reduce their participation in the manufacturing sector, the HHI increases. Figure 3.4 shows that concentration i s higher in cities with relative lower number o f firms (i.e., specialization rather than diversity drives their competitive advantage). The group o f specialized cities includes Cartagena, Manizales and Pereira. Inthe other extreme, diversified cities are those in which the strength of local markets is higher and includes Bogota, Cali and Medellin. Bucaramanga and Barranquilla are in between (but increasingly specialized): neither as diverse as the large cities nor as specialized as the first group. The group `other' i s not relevant to this analysis because it i s does not represent a single urban agglomeration. 3.5. Services are more concentrated in the main metropolitan areas than are manufacturing firms, especially in Bogota. Traditionally, with manufacturing gentrification, the services sector increasingly agglomerates in former industrial areas, as it is the case in the United States and Brazil. More efficient urban centers have become the locus o f industrial activities, like in the Chinese or Midwest American cities. Services for both firms and individuals are substantially concentrated inBogota, presenting an increasing trend inspite o f the growth o f other large urban centers, such as Cali and Medellin. In addition, other sub-regions, such as Boyaca, Cauca, Cordoba, Cundinamarca and Sucre, have become increasingly important in the provision of business services during the period of 1990-2005. In addition to Bogota, 22 regarding personal services, Atlantico, Bolivar, and Cordoba have also increased their participation between 1990 and 2005 (Figure 3.5).15 Table 3.2: Colombia: Distribution o fmanufacturing Figure 3. 4: Concentration o f Manufacturing Firms f m s by sub-region and sectors, 2002 (percent) 0 25 =v 020 g 91 1993 E2002 $ 015 c5 010 t,E005 0 00 Source: Based on DANE's Annual Manufacturing Survey: Figure 3. 5: Colombia: Distribution of services by sub-region, 2005 (percent) (a) Business services (b) Personal services 45% 35% 40% 30% 35% 30% sa 1990 25% a1990 25% a2005 20% 02005 20% 15% 15% 10% 10% 5% 5% 0% 0% g $ $ g i S $ ; g B s " ' " cod" P 8 2 k i j i j s p $ j g j3i 8% % 0 5 Source: Based on DANE's Annual Manufacturing Survey. 3.6. The share of Colombian firms exporting their products increased substantially between 1993 and 2002. The Colombian manufacturing sector widely increased its export orientation in this 9-year period (Figures 3.6a and 3.6b). Firms located in regions other than Bogota are more export-oriented, which points to the following two facts: (a) the attraction generated by local demand i s stronger in Bogota, and (b the competitiveness of Colombia's firms may also rely on the buying ability of local markets.' 2 15 Business services comprise activities such as finance, insurance, real state, software, and consultancy, while personal services refers to activities like lodging, restaurants, and entertainment. 16See Chapter 4 for additional evidence fiom the Colombia 2006 Investment Climate Survey. 23 Figure 3.6: Proportion of exporting f m s , 1993 and 2002 (percent) (a) By metropolitan area (b) By sector 40 35 35 Sl993 ill2002 30 ai993 m 2002 30 25 25 20 20 15 15 10 10 5 5 0 0 3.3 EconomicConvergenceamong Sub-Regions 3.7. Colombia is characterized by the persistence of large disparities between Bogota and the other sub-regions. Studies have found mixed results on whether or not there is convergence on regional income per capita in Colombia. Over a 25-year period, no sub-region presented an income per capita greater than 50 percent of Bogota's (see vertical axis in Figure 3.7). From 1975 to 2000, only Caldas, Cundinamarca and La Guajira were able to partially close the income gap with Bogota (Group A). In most of the other sub-regions, the income gap with Bogota widened; these sub-regions can be classified into three groups based on their income level: (a) low-income (Group B), including Caqueta, Cauca, Choco, Cordoba, Magdalena, Nariiio, Norte de Santanderand Sucre; (b) middle-income (Group C), includingBolivar, Boyaca, Cesar, Huila, Meta, Quindio, Risaralda, Santander and Tolima; and (c) high-income (Group D), including Antioquia, Atliintico, Valle and Bogota. 3.8. Duringthe period 1975-2000,income grew faster than populationin all Colombian sub-regions; however,L a Guajirawas able to doubleits 1975-income.When Colombia grew at its fastest rates during the early 1970s (the import substitution model), most sub-regions grew faster than Bogota. After 1990,Bogota grew faster than any other sub-region inthe country. The crisis of the late-1990s affected the whole Colombian economy, particularly the sub-regions of Atlantic0 and Valle (Figure 3.8). 3.9. Colombian sub-regions differ significantly in terms of competitiveness, ranging from those that can and are competing in international markets to those that require special attention. Using indicators of economic and social development (e.g., GDP level and growth, and export orientation) and knowledge and innovation indicators (see chapter 6), the sub regions in the country can be grouped in four different categories (Figure 3.9). The first two groups are comprised of sub-regions that can compete internationally. The difference between them being the degree of local specializatioddiversification: Antioquia, BogotdCundinamarca and Valle already show a degree of economic diversification, while Bolivar, Boyaca and the coffee region are highly specialized sub-regions. In addition, Atliintico and Santander are in betweenthese two groups. The third group, called "poles for local development", is formed by 8 24 sub-regions, which present an intermediate level o f economic development and knowledge indicators but lack important instruments to compete and grow. Finally, the last group lacks most o f the instrumentsfor growth and competitiveness, requiring special attention. Figure 3. 7: Colombia: Sub-regional incomeper capita relative to Bogota's (percent) 50% * 0 N .-C VI m CI g 40% A: Regionsthat partially rn 0 closed income gap with 0 Bogota CI a, B: Low-income -?! ._ 30% C: Middle-income D:High-income $a20% -E810% 10% 20% 30% 40% 50% Income per Capita Relative t o Bogota's in 1975 Source: Based on CEGA. Figure3.8: Average income per capita growth by sub-regions or group of sub-regions 8% 6% 4% 2% I I 0% 2% I I -2% 0% I -4% \ -2% -6% .1 ....................................... -4% - - 8 -Lowlncome Middle Income -Antioquia BogotQ Caldas -Cundinamarca ____ ---LaGuajira -AtlOntico _- ~ -Valle -x- Bogotl ~"_ "__._-I " " Source: Based on CEGA- ___. ~_1 _I--" I I ~ ~ 3.10. Most empirical analyses found that persistence of sub-regionalincome disparitiesin Colombia persisted. Birchenall and Murcia (1997) classified sub-regions by three types: (a) a few sub-regions where income per capita was consistently higher than the national average, (b) the majority o f sub-regions with decreasing income per capita (inreal terms) compared with the 1960s' levels, but raking in the same initial position, and (c) a few sub-regions with per capita income levels around the average but showing convergence towards the income per capita o f 25 leading region^.'^ Bonet and Meisel (1999), using bank deposits as a proxy for regional revenue, found that lagging regions grew faster than leading regions but without reducing inequalities in income per capita distribution for the period 1925-1960. On the other hand, for the period 1960- 1995, the evidence for both more rapid growth of lagging regions and better income distribution are weak. They concluded that instead there has been a polarization inthe distribution of the sub- regional income. Soto (1998), using a data panel approach for the period 1960-1995, concluded that there was no faster growth of lagging sub-regions (absolute convergence), but that there was some convergence conditional to high school education and population growth. Bonet (2006) showed that the process of polarization between Bogota and the other regions persisted during the entire periodof 1975-2000. Figure3. 9: Sub-regionalcompetitivenessinColombia .. - _ _ _ - - . --I ._ EconomicDevelopment Source:World Bank Analysis. 3.1 1. Other researches found that disparities have either increased or decreased over time in Colombia. Rueda(2004) found evidence of divergence on regional income per capita during the period 1960-1994, disagreeing with Birchenall and Murcia (1997) on the convergence of middle-income regions. She also found that the period 1985-1996 was characterized by a persistence of inequalities, while divergence occurred from 1960-1998. Cardenas and Escobar (1995) found an annual rate of absolute convergence (faster growth of lagging regions) of 4.0 percent during 1950-1989. When a shorter period of time was considered (1960-1989), this convergence rate was reduced to 3.2 percent. 3.12. Little interregional interaction seems to be one of the causes of the absence of convergence in Colombia. The existing literature agrees that most sectors are self-sufficient at a regional scale in Colombia, with little interregional interaction, and most forward and backward "FollowingQuah(1993a, 1993b,1996and1997). 26 linkages are concentrated inside the richest regions; these results suggest that the regional polarizationmay be perpetuated. '* 3.13. The few sub-regions inwhich income per capita converged to the national average either relied on the availability of natural resources or were relatively better positioned geographically in terms of market access. Birchenall and Murcia (1997) suggested that distance to Bogota, such as from Cundinamarca, and a strong share of sub-regional GDP in the mining and oil sectors, such as in La Guajira, explained the observed convergence in a few sub- regions. However, in the majority of sub-regions, which demonstrated neither of these two characteristics, the gaps on income per capita persisted. 3.14. Transfers from the central government to sub-national governments have not contributed to convergence.Rocha andVivas (1998) found that fiscal transfers from the central government to sub-national governments were negatively related to sub-regional GDP growth. Consistently, Baron and Meisel (2003) found that the decentralization process that began with the 1991Constitution didnot helpto reduce sub-regional disparities. Rueda(2004) indicatedthat public expenditures, especially public investments, improved the relative position of some regions but not the dynamics of income distribution. However, it i s important to point out that transfers and investments made by the central government could have played a major role in reducing income divergence. Infact, Rueda (2004) found that public investmentshave prevented polarization by redistributing resourcesfrom leading sub-regionsto the lagging ones. 3.15. Investments in infrastructure seem to have had an important impact on sub- regional growth. Cardenas and Escobar (1995) suggested that public investments in infrastructure played a substantialrole inpromoting regional income convergence. Meisel (2006) showed that the variables with the major effects on the growth of income per capita and its level in cities were human capital, the physical infrastructure endowment and the quality of institutions. 3.4 Productivity and Location of ManufacturingFirms 3.16. The average TFP for the manufacturing sector does not present significant variations across industrial agglomerationsin Colombia. However, between 1994 and 2002, TFP has grown inBogota, Cartagena, Medellinand smaller urban agglomerations,while it declinedinall other industrial agglomerations,especially inManizales (Figure 3.10). Results are only presented at the city-level average for statistical representation. These productivity measures were estimated usin the dynamic panel estimation methodology developed by Levinsohn and Petrin (2O03).lF An inverse relationship between average capital and labor intensity was found for the manufacturing sector. While Bogota, Manizales, Medellin and Pereira have relatively capital intensive firmshectors, Barranquilla, Cartagena and smaller industrial agglomerations have labor intensive firms/sectors (Figure 3.11). In terms of developing commercial advantages inconsumer markets basedon new products (as presentedby I8 In Chapter 5, this issue is explored in detail, and evidence from input output matrix and from other analyses is presented. l 9DetaileddiscussionaboutTFP inColombiacan be found inHaltiwangeret a1(2006). 27 CAF, 2006), all metropolitan areas seem to have the same potential. Sub-regions not located in metropolitan areas, where there is lumpy employment, may have problems competing with the mainurbanagglomerations except for a few cases inwhich there i s specialization. 28 Figure 3.10: TFP inlevels(2002) and growth (1994- Figure3. 11: Intensity inthe use of factors, 2002 2002) by region 1 1.2 1 ,3% 2% 1% Other OX -1% Caltagena Barranquilla4 -2% 4% -4% + Bucaramanga+ Cali -5% -6% a, -7% -J35 ~ Pereiramanizales +4 +BogOt' Medellin 30 1 > Note: Columns are TFP levels and dots are TFP growth Source: Based on DANE's Annual Manufacturing rates for the period 1994-2002. Survey. Source: Based on DANE's Annual Manufacturing Survey. 3.17. Firms' responses to competitionmay be drivinglocation.An econometric model was estimated to analyze drivers of firm location - a Probit model where the dependentvariable is a dummy that takes the value one if the firm is located in Bogota, for example, and zero if it is located somewhere else inthe year of 2002. The independentvariables are: TFP, which measures how the mix of production factors (i.e., capital, labor, energy and materials) allows for output maximization and, therefore, profit maximization, given input prices. It i s estimated using the full panel 1994-2002 to capture the dynamics of TFP using Levinsohn and Petrin (2003) procedureto correct for simultaneity ininput coefficients; Agglomeration forces, as suggested by the New Economic GeographyY2'with a version of own industry concentration included to capture different processes that occur at the local level, such as access to a specialized labor pool, suppliers and technological spillovers, among other industry agglomeration mechanisms;21 A measure of the impact of international trade to check whether increased competition with foreign firms drives the differential location of firms in the sense of Schumpeter's (1942) postulates. Both export-orientation and import-exposure are included. Export-orientation measures whether an industrial sector ina given sub-region exports more thanthe same sector in all other sub-regions, such that it would be this sector in the given sub-region that i s strongly competing with foreign firms (i.e,, it would not be a Colombian comparative advantage but the 20 See Fujita and Krugman (2004) for a general discussion. For a more technical analysis, see Neary (2001) and Ottaviano and Thisse (2004). 2 'See La11et a1(2004). 29 sub-region comparative advantage). Import-exposure measures whether this sector is facing strong competition from imports, with data available by sector at the national 3.18. Conceptually, these models should be estimated in a system of equations in which the decision to locate in one city is made comparing simultaneously all alternative 10cations.~~ However, in the case of Colombia, a few areas concentrate all manufacturing firms making it impossible to identify the role of location drivers from local unobservable factors. In order to tackle this identification problem, a second equation is included with instruments to control for the fact that agglomeration depends on the higher profits/pecuniary costs that firms face from locating together. Natural candidates for instruments are the same agglomeration variables but lagged up to 1994 as DANE'Sdataset is a panel (1994-2002). These instruments recognize that agglomeration is also an endogenous location process by attributing agglomeration to historical conditions. It leaves unanswered the question on what created agglomeration in the areas under analysis but statistically cleans the model for 2002. 3.19. Higher export orientation drives manufacturing firms toward the Atlantic Coast and smaller urban agglomerationsversus Bogota, Cali and Medellin,while higher import exposure affects mostly Barranquilla, Cali and smaller urban agglomerations. After controlling for productivity and industry concentration, results show that the most competitive location for firms inthe manufacturing sector is Cartagena, as it benefits from export-orientation but does not suffer from import-exp~sure.~~However, Cartagenalacks industrial diversity, which requires the region to envision itselfas a specializedpole. Barranquilla, inturn, i s highly affected by import competition. Bogota and Medellin are similar, since the export-orientation drives firms somewhere else, while the impact of import-exposure is reduced. Bogota and Medellinrepresent the most qualified markets in terms of local buying capacity with a long distance from ports. Thus, while firms operating inthese two regions are naturally "protected," they pay a heavy toll to compete in foreign markets. Cali faces both problems: export-orientation does attract firms and import-exposure substantially affects firms. Bucaramanga seems to be less affected by import competition but export-orientation drives firms elsewhere. Smaller urbanagglomerations in different localizations within the country ("other" inTable 3.3), even near large urbanareas, may have specific competitiveness drivers that are not accounted for in this exercise. Nevertheless, results are consistent with what is has been observed in other large urban agglomerations, such as Si30 Paul0 (Brazil), where manufacturing activity has consistently declined since the 1 9 7 0 ~ ~ ~ 22 Note that there are multiple endogenous processes taking place. In particular, profitability depends on agglomerationvariables as f m s benefit and suffer from locating together. Therefore, the model requires a second stage to control for proper instruments. Natural candidates are the same agglomerationvariables but lagged up to 1994. 23 See Bayer andTimmis (2007). 24 This analysis is basedonhistoricalinformationusinglargeurbanagglomerationsfor statisticalrepresentativeness. 'IiSee WorldBank(2007). The overall experienceofBrazil is describedinmore detail in Chapter 2 ofthis report. 30 Table 3.3: Two-stageProbit: Determinants ofthe probability of locating ina sub-regionvs. any other sub-region Absolutevalue of z-statistics in parentheses *significant at 5%; significant at 1% Instrumented: Total Factor Productivity Instruments: Own Industry Concentration,Export Orientation, and Import Penetrationin 1994 Source: Basedon DANE'SAnnual Manufacturing Survey. 3.20. This econometric analysis seems to reinforce the view that trade liberalization may have a significant impact on firm's location. It i s important to convey that conclusions are drawn upon historical data and may not equally work in the future as there is no counterfactual with data available (i.e., industrial areas with other sub-regions of Colombia where such processes did not occur but could have occurred we not being compared). For instance, NATFA did not drive massive relocation of Mexicanfirms to the border with the U S as one would have concluded analyzing the balance between exposure to trade and transportation costs. The opposite, firms relocating due to trade exposure, had been also observed in many cases like the U S and China. Despite this methodological limitation, what is most likely to occur given what i s currently observed is that the flows of imports and exports seem to be important drivers of firms' location in Colombia. In designing its sub-regional competitiveness program, the central government should carefully follow the evolution of firms' location, as important changes may take place inthe country. 31 CHAPTER4 - INVESTMENTCLIMATE INSELECTEDREGIONSOF COLOMBIA 4.1 Introduction 4.1. The investment climate i s recognized as a key pillar o f World Bank Group work to promote economic growth and poverty alleviation in developing countries.26 A dynamic private sector, inwhich firms compete with each other and seek to improve productivity by investingin human and physical capital as well as in technological capacity, is the main propellant o f sustained economic growth - a sine qua non condition for reducing poverty and improving living standards. A sound business environment is essential to establish the appropriate incentives for firms to invest. Such an environment must deal with the factors constraining the effective functioning o f product markets, financial and non-financial factor markets and infrastructure services, including weaknesses inan economy's legal, regulatory and institutional framework. 4.2. Adverse business environments are one of the constraintsto economic development in lagging regions. The degree to which these constraints are particularly binding at the sub regional level i s less significant than at the national level, giventhat legislation on taxation, labor relations and trade are fixed at the national level and meant to be uniformly applied throughout the country. However, administrative practices vary across regions, even when performed by central government officials. For instance, tax administration i s particularly susceptible to local discretion inthe form o f disruptive and costly inspections, selective enforcement and associated extortion. Moreover, adopting a wider concept o f the investment climate (IC), in order to incorporate the availability and quality o f infrastructure services and labor force, leads to significant differences across regions. 4.3. Several dimensions of the business environment in four Colombian regions were captured by a firm-level survey undertaken in 2006. The Colombia Investment Climate Survey (ICs) sampled 1,000 establishments from three industries, including four distinct manufacturing sectors and two specific services sectors (for details, see Table 4. 1). The questionnaire was based on the standard Investment Climate Assessment (ICA) questionnaire so as to contribute to the World Bank's cross-country data on business environment issues. The World Bank's staff selected the sample, which included all firm sizes, whose classification was based on the standard World Bank criteria - small (less than 20 workers), medium (20-99 workers) and large (100 or more workers). Finally, trained enumerators carried out the survey through face-to-face interviews with company management. 26"The central challenge in reapinggreater benefits from globalizationlies in improving the investment climate - that is, in providing sound regulation of industry, including the promotion of competition; in overcoming bureaucratic delay and inefficiency; infighting corruption; and inimprovingthe quality of infrastructure. While the investment climate is clearly important for large, formal sector f m s , it is just as important - if not more so - for small and mediumenterprises(SMEs), the informal sector, agricultural productivity, and the generationof off-farm employment. For these reasons, the investment climate itself is a key issue for poverty reduction" (Nicholas Stem, World Bank, ChiefEconomist, March22,2001). 32 Table 4. 1: Colombia ICs: Sample composition by region and industry, 2006 Indust Barranquilla Bogota Cali Medellin Apparel 13 12.0 60 14.1 35 22.4 64 20.6 172 17.2 Chemicals 8 7.4 99 23.2 13 8.3 40 12.9 160 16.0 Services 29 26.9 146 34.3 31 19.9 63 20.3 269 26.9 Note: Nmeans number of observations. Source: Colombia ICs-2006. 4.2 RegionalBenchmarking 4.4. In order to assess the different aspects of the investment climate in selected Colombian regions inareas where best practices canbe followed, abenchmarking exercise was conducted. It used series of both hard-data and perception-basedindicators on the investment climate in four regions - Barranquilla, Bogota, Cali and Medellin. The analysis focuses on four main categories inwhich sub regional variance was expectedto be significant: (a) infrastructure, (b) markets, (c) skills and technology, and (d) government effectiveness. Infiastructure 4.5. Overall, the quality of infrastructure services is lower in Barranquilla than in the three other regions. Infrastructureis one of the determinantsof firms' productivity,and, therefore, the quantity and quality of infrastructure services are key components of the investment climate (Dollar, et. al., 2004). In addition, the availability and the quality of infrastructure services are important determinants of a firm's location decision (World Bank, 2007a). The share of total sales losses due to electricity outages can be used as a proxy for the quality of electricity services in a region. In Colombia, when the four regions are compared, firms in Barranquilla suffered more from power outages than those located in other cities. The share of firms that reported power outages in Barranquilla (82.4 percent) is 50 percent higher than that in Bogota and more than 2.5 and 4.0 times those in Cali and Medellin, respectively. In addition, firms in Barranquilla reported a higher average number of outages in 2006 than firms located elsewhere. Nevertheless, the average sales loss for firms located inBarranquilla was only significantly higher than the average for firms operating inMedellin (Figure 4.1). Manufacturing firms responded to the unreliability of the electricity supply with precautionary investments in self-generation systems (Figure 4.2). Even though self-generation may be an effective way of 33 guaranteeing supply, it is not efficient for the economy as a whole as firms are channeling financial and managerialresources to non-core activities. Figure 4. 1: City-region quality of electricity services Figure 4. 2: City-region: Share ofmanufacturing f m s with ownership ofpower generator (percent) I Source: Colombia ICs-2006. Source: Colombia ICs-2006. 4.6. Firms in Barranquilla and Bogota Figure 4.3: City-region: Share of f m s rating or electricity and transport as major or severe obstacles severe perceive infrastructure sewicesgrowth obstacles to their as majorand to their growth (percent) operations to a larger extent than firms in Cali and Medellin. Even though firms in Barranquilla are provided with electricity services with the lowest quality among the four regions, only half o f them rated this service as a major or severe obstacle to their operations. On the other hand, even though around 50 percent o f firms in Bogota suffered from power outages in 2006, about 60 percent o f them considered electricity a majodsevere obstacle to their growth and operations. This finding i s similar to that for firms in Medellin. On transportation, firms in Barranquilla and Bogota were the ones SOurce: that felt more constrained by this service in terms o f growth and operations. The share o f firms with this perceptioninBarranquilla, for example, was about two times that of Cali and Medellin (Figure 4.3). Markets 4.7. Exporting is an intense activity in Medellin, while the other regions are more focused on the domestic market. Overall, the operations o f Colombian firms focus on domestic markets. Exports (either direct or indirect) as a share of total sales in 2006 for firms located in Barranquilla, Bogota and Cali averaged only between 3 and 5 percent. Even though this share was also low in Medellin, exporting was a more important activity in the region - around 13 percent o f its firms' total sales were sold inforeign markets (Figure 4.4). 34 4.8. More than one-third of firms Figure 4.4: City-region: Extent ofmarkets(percent ) operating in Barranquilla and Medellin sell their products to other citieshegions, but manufacturing companies located in Medellin present the highest exporting intensity. Firms in the manufacturing industry located in Barranquilla and Medellin made around 35 percent o f their sales to buyers from other citieshegions (Figure 4.5). However, firms in Barranquilla were more geared towards the domestic market than firms inMedellin. In2006, 10 percent of manufacturing sales o f firms located in Medellin reached foreign markets, while this share was only about 5 percent in Source: Colombia ICs-2006. Barranquilla, 4 percent inBogota and 1percent Cali (Figure 4.6). Figure 4. 5: City-region: Average share of sales to Figure 4. 6: City-region: Extent of markets of other regionsby manufacturing frms (percent) manufacturing firms (percent) ET3Local National -rlr Foreign Source: Colombia ICs-2006. Source: Colombia ICs-2006. 35 Skills and Technolod7 4.9. Interms of educationalattainment, the Figure 4. 7: City-region: Average educational production workforce in Colombia seems to attainment of workforce (percent) be relatively high-skilled. More than half o f surveyed firms in Colombia reported that the average education of a typical production worker was at least 13 years. This share rose to almost 60 percent in Medellin and about 65 percent in Barranquilla. At the same time, firms in Barranquilla and Cali presented the highest averages o f low-skilled workers, which are those with less than7 years of education (Figure 4.7) 4.10. Firms located in Bogota and Medellin take advantage of information technology Source: Colombia ICs-2006. (IT) to a larger extent than firms operating in the other two regions. Around 90 percent o f Figure 4.8: City-region: Share of firms using I T firms with businesses in Bogota and Medellin used e-mail to contact their clients and suppliers, while around 75 percent o f the establishments in Barranquilla and 80 percent in Cali used email. Some firms also used their own websites for communication purposes, but this technology was used less frequently by Colombian enterprises than was e-mail. About half o f the firms operating in Bogota and Medellin used websites, while this share fell to about one-third o f the enterprises located in Barranquilla and Cali (Figure 4.8). Source: Colombia ICs-2006. Government Effectiveness 4.11. In terms of government effectiveness, dealing with government regulations is a burdensome process throughout Colombia. When complying with such regulations means fulfilling complicated and extensive processes, the investment climate deteriorates. In 2006, managers o f firms in Medellin and Barranquilla spent 12 and 13 percent, respectively, o f their time dealing with government regulations. Meanwhile, enterprises in Bogota and Cali spent between 20 and 25 percent o f their time dealing with such bureaucratic procedures (Figure 4.9). In addition, entrepreneurs' perceptions on the statement "the interpretations o f laws and regulations by public officials are consistent" varied across regions. For instance, only half o f the managers working in Barranquilla agreed, while this share reached almost 60 percent in Bogota and Medellin (Figure 3). 27See Section4.2 for abriefdiscussionon innovation inputs and outputs. 36 Figure 4. 9: City-region: Average share of Figure 4. 10: City-region: Share of firms agreeing managerialtime spent dealingwith government that governmentofficials' interpretations of laws and regulations, 2006 (percent) regulations are consistent (percent) 20 20 - x s j 13 12 0 1 0 . _ _ . - ~ rI. BarranqJilb Bogota Cali Meuetlin Banaoquilla Bogota Cali Medellin Source: Colombia ICs-2006. Source: Colombia ICs-2006. 4.12. On average, firms inBarranquilla declare lower shares of annual sales and labor for tax purposes; therefore, a larger share of enterprises was inspected by tax officials in 2006. Firms operating inBarranquilla declared 70 percent o f their annual sales for tax purposes, while this share rose to between 82 and 87 percent inthe three other regions. On labor declared for tax purposes, firms inBarranquilla also declared the lowest share o f the workforce (85 percent), but it was close to what was declared in the three other regions (Figure 4.11). Thus, the share of companies operating in Barranquilla, which was inspected for tax purposes in 2006, was the highest (one-third o f them), a figure close to that o f Bogota. On the other hand, about one-fourth o f firms inCali and Medellinwere inspected by tax officials in2006, even though their shares o f sales and workforce declared for tax purposes were close to those o f Bogota. Nevertheless, the average number o f inspections in 2006 per inspected firm (three) was the same in the four cities (Figure 4.12). Figure 4. 11: City-region: Average share of sales and Figure 4. 12: City-region: Tax inspections, 2006 labor declaredfor tax purposes Totalworkforcedeclared for tax purposes (X, average) Source: Colombia ICs-2006. Source: Colombia ICs-2006. 37 4.13. The time it takes for courts to reach a Figure 4. 13: City-region: Judicial decisions(weeks) judicial decision and enforce it varies significantly across the four regions, with courts in Cali performing the best. In 51 A 7 52 Barranquilla, it took the longest for courts to t 2 4 0 reach a judicial decision (68 days on average), 25 ~ VI 8207 but it was the city with the second shortest time L to enforce a decision (7 days). Courts in Bogota 5 and Medellin averaged the same amount o f time to reach a decision (around 50 days), but courts located in Medellin enforced their decisions in Time for courts to reachjudgement -A- Time for court enforcement I almost half o f the time it took the courts in Bogota (25 days) (Figure 4.13). Source: Colombia ICs-2006. 4.3 A Closer Look at Technology and Innovation 4.14. Are there significant differences across the four Colombian regions in terms of technology and innovation? Using the data collected by ICs, it was possible to analyze with more detail the adoption o f innovation inputs and the development o f outputs, which were found to vary significantly across regions. Training o f the workforce and research and development (R&D) were the innovation inputs that were most promoted by Colombian firms. In addition, Medellin was the region in which firms promoted the most innovation inputs (Le., R&D, technology license, quality certificate and worker training). Given the fact that Medellin was the most exporting intensive region, the highest share o f firms with quality certificates (ISO) was found in this region. On the other hand, the lowest share o f firms investing in innovation inputs and outputs was found in Cali, with the exception o f improved production line. Eventhough the share o f firms investing in innovation inputs was lower than that in Medellin, Barranquilla and Bogota excelled inthe development o f innovation outputs (Figure 4.14). Figure 4. 14: Colombia: innovation inputsand outputs by region (percent) (a) R&D, licensing, and I S 0 (b) Training, improved line, andnew product Source: Colombia ICs-2006. 38 4.15. Cross-regional comparisons need to take into account some firm-level characteristics that are associated significantly with investments in innovation, including size, sector, and exporting and foreign ownership status. As reported in Table 4. 2, the probability of investing inR&D and other innovation inputs and outputs generally increasedwith firm size, and it is usually higher for exporters. For example, for a firm with less than 20 employees, the probability of investing in R&D was only 21 percent lower than that of firms with more than 100 workers. Similar relative probabilities are found for acquiring an I S 0 certification (22 percent), providing worker training (3 1 percent), having a technology license (13 percent), developing a new product (13 percent) and improving a production line (13 percent). Exporters also had higher probabilities of investing in innovation. For example, their odds of spending in R&D, I S 0 certification and worker training, as well as developing a new product, were respectively 25, 15, 11 and 12 percent higher than non-exporters of the same size and sector that operate in the same region. Firms with some degree of foreign ownership were more likely (by 11 percent) to only report technology licensing than were the purely domestic firms with the same idiosyncratic characteristics (Table 4.2). 4.16. Location has few impacts on the promotion of innovation inputs and outputs by Colombianfirms. However, firms operating inCali were 14 and 12 percent less likely to invest inR&D and worker training, respectively, thanfirms ofthe same size, sector, and exporting and foreign ownership status operating in Medellin. Inaddition, firms operating in Cali were also 13 percent less likely to develop new products than were similar companies located in Bogota. A similar result was found when Bogota and Medellin were compared: firms located in Medellin had an 8 percent lower probability of developing innovation inputs than that of similar firms operating inBogota (Table 4.2). 4.4 ProductivityDeterminants2* 4.17. There i s growing econometric literature investigating the relationship betweeninvestment climate (IC) and economic performance at the firm level. Many of these papers have widely discussedindicators to measure IC and economic performance. For instance, to assess the impact of seventeen IC indicators in China, Hallward-Driemeier et al. (2003) used four measures of economic performance - employment growth, investment rate, sales growth and total factor productivity (TFP). Dollar et al. (2003) and Escribano and Guasch (2005) proposed pooling observations across countries to measure the impact of IC variables on TFP. Inaddition, Dollar, et. al. (2003) considered the effects of IC variables on the growth rate of sales, employment, output, fixed assets, average wage and gross return as well as their effects on capital. The drawback of this approach i s that it estimated common IC parameters for all the countries inthe sample. Bastos and Nasir (2004) proposedusing the principal components approachto assess the impact of a set of indices of rent predation, competition and infrastructure on productivity. They not only interpretedthe coefficients of IC variables, but they also adjusted the methodology of Kruskal (1987), based on the concept of partial correlation coefficients to rank the three IC indices. This sub-sectionis basedonEscribanoet a1(2007). 39 Table 4.2: Marginaleffects on innovationinputs and outputs inColombia Worker Tech. New Improved Independentvariables R&D I S 0 training licenses product line 20 to 99 workers 0.077* 0.062*** 0.193*** 0.041** 0.036 0.055 [0.0461 [0.026] [0.045] [0.022] [0.041] [0.040] Exporter 0.250*** 0.155*** 0.114* 0.010 0.121** 0.057 [0.0571 [0.044] [0.060] [0.024] [0.049] [0.052] Barranquillavs. Bogota 0.008 -0.009 -0.039 -0.029 0.015 0.022 ro.0791 ro.0381 r0.0781 ro.0251 r0.0721 r0.0701 Cali vs. Bogota -0.090 -0.018 -0.062 -0.026 -0.134** 0.017 [0.063] [0.026] [0.065] [0.021] [0.061] [0.057] Medellinvs. Bogota 0.056 0.024 0.055 -0.009 -0.085** -0.066 Bogota vs. Medellin -0.056 -0.022 -0.055 0.010 0.083** 0.065 [0.0491 [0.019] [0.050] [0.021] [0.043] [0.044] PseudoR2 0.080 0.243 0.089 0.081 0.034 0.013 Notes: *Significant at the 10 percent level. **Significant at the 5 percent level. ***Significant at the 1percent level. Robust standard errors are in brackets. For brevity purposes, sector variables were not included in Table 4.2. Textiles is the omitted category for sector. Small is the omittedvariable for size. Source: ColombiaICs-2006. 4.18. One o f the main contributions o f the Escribano and Guasch (2005) methodology i s how they modeled the I C effects. Usingthe dataset for Colombia - a broad list of more than 100 I C variables that intended to recreate conditions in which firms operate - rather than considering a narrow group o f I C indicators, was employed inthe model, and, as a result, the firm-level fixed effects explaining the idiosyncratic differences between firms could be determined. Thus, the larger and more reliable the I C group o f variables is, the more comprehensive will be the characterization o f the investment climate conditions. Inthis section, the analysis was based on the regional level. But the more the analysis is disaggregated, the more heterogeneity the I C will have. The disadvantage o f this process is the loss o f asymptotic efficiency due to the reduction in the number o f observations used in the regressions. Apart from the differences on economic performance measures and other differences in the construction o f the I C indicators, the way Escribano and Guasch (2005) approached the robustness of the empirical results i s what differentiates their paper from other studies. The emphasis i s on the robustness o f the empirical I C elasticities and semi-elasticities to several alternative productivity measures based on the ICs. Since there is no single salient measure of productivity, policy analysis focuses on robust empirical I C results using several productivity measures. 40 4.19. The evolving literature on IC has recognized the importance of having data at the firm- level to disentangle the real impact that IC has on the firm's economic performance. The results should be interpreted with caution due to the usual caveats applied to cross-sectional datasets. Using cross-sectional data, it is difficult to make conclusions on causal relations between the variables; and this kind of dataset makes it difficult to use the traditional IV approachto address the endogeneity problem. Results at the national level 4.20. The econometric analysis at the national level confirms the expected association between investment climate variables and productivity of firms. Using all firms of the survey, a number of investmentclimate indicators drawn from the Colombia Investment Climate Survey were econometrically related to measures of productivity. For the econometric analysis the investment climate (IC) variables were allocated to five categories: infrastructure, governance, including red tape and crime, finance, quality, innovation and skills and control variables. The estimation was performed holding constant basic firm characteristics, including location, industry affiliation, firm size, firm age, and other factors. 4.21. Overall results of the assessment show that variables in all four categories of investmentclimate affect Colombian firms in terms of their productivity. This can be seen in Table 4.3, which presents the elasticities and semi-elasticities of the investment climate variables with respect to TFP obtained using different specifications. The results indicate that IC variables related to all categories are statistically related to productivity. Redtape, corruption and crime- related variables have in general a significant impact upon productivity - with the exception of payments to speed up bureaucracy. For example, an increase of 1 percent in the crime losses results in a decrease of around 0.06 percent in productivity. Among the infrastructure variables relevant for productivity, time to clear customs for imports, power outages, waiting time for electricity supply and shipment losses appear with a strong negative effect. An increase of 1 percent in the number of days to clear customs for imports reduces productivity by 0.4 percent. Education - as measured by the share of staff with university education -increaseproductivity and its impacts are stronger than inthe cases of infrastructure and red tape, corruption and crime. Finally, with respect to finance, the existence of a checking or saving accounts positively impacts productivity while sales paid after delivery and working capital from family and friends have a negative association with the firm's productivity. For example, those firms that have a checking or savings account are 0.4 percent more productive thanfirms without suchaccounts. 4.22. Variables related to red tape, corruption and crime and to infrastructure present the higher relative impacts on productivity. Figure 4.15 shows the impact of average IC variables on aggregate (log) productivity. The results show that for infrastructure the two most important factors for averageproductivity are days to clear customs average and shipment losses. For red tape and crime, work force reported to taxes, security costs and crime losses have a relative high impact on aggregate productivity. Finally, regarding quality, innovation and labor skills, their overall relative impact i s lower than the one it was found in other Latin American countries, although the results havethe expected sign. 41 Table 4.3: IC elasticities and semi-elasticitieswith respectto productivity Infrastructure Paymentsto obtain a contract with -0.072*** -0.070*** -0.052*** -0.057*** -0.053*** -0.077*** governance ing or saving accoun 0.877** 0.598 0.700* 0.412 Other control Observations 570 570 570 570 570 570 Notes: significant at 10%; * ** significant at 5%; ***significant at 1%. Eachregression includes a set of industry dummies and a consta term. (a) Variables instrumented with the industry-region-size average. (b) Variables approximated with a proxy (only missing valu replacedby the industry-region-sizeaverage). Source: Colombia ICs-2006. Results at the sub-regional level 4.23. The empiricalobjective here is to evaluate the impact of the IC on the productivity of Colombianfirms under the assumptionthat its constraintsdiffer when one moves from one region to another. Five groups of IC variables were identified: (a) infrastructure (b) red tape, corruption and crime, (c) finance and corporate governance, (d) quality, innovation and labor skills, and (e) other control variables. In order to pick up those differences, the robust econometric methodology proposedby Escribano and Guasch(2005) and extended by Escribano et al. (2007) was applied for the regional analysis (i.e., region by region). The procedure used in the selection of IC variables was performed for each one of the three regions considered: (a) Bogota; (b) Medellin; and (c) Barranquilla and Cali. For the joint regression of the metropolitan areas of Barranquilla and Cali, the parameterso f some IC variables were allowed to vary region by region with the introduction of interaction terms. Once the statistically significant IC impacts on firms' productivity were selected, the relationship proposedby Escribano et al. (2007), among the terms ofthe Olley and Pakes (O&P) decomposition (see Box 4.1) and the investment climate factors affecting productivity ofthe Colombian manufacturing firms, was analyzed. 42 Figure 4. 15: Relative I C effects on aggregate productivity (decomposition in logs) 6 0.0 0.0 0.0 0.0 D O 0.0 T.l 11 12 13 14 15 16 T.2 21 22 23 24 2.5 26 T.3 3.1 32 3+34 ,7,4 . 4.1 4.2 4.3 T.5 5.1 5.2 T.lTota1 Infrastructures 2.3SeQcitYmsts otal quality, Innovationand labor ll-to d w a s t o m t o i w r t 2 4Clrmlosses skills 12Loseesdust0powrollages 25 P m sto obtainawntradm h t b 4.1-for qualitydifmtion 13AveaeecCratianofponeroutage0 go- 42cUso~n4rg 1 4 W t f o r d e c t r i c s ~ 26P W Sto speedup- 4 . 3 S W f - u i v e s i t y ~ i a n 159ip-m-d Iosses,inpolt T.3 Total flnanco and corporate T.6 Total othorcontrol variables T.2 Total redtape corruptlonand 5.lhnnyforp&kcgnal (loVOmPIW0 crlme 3.1SslespddaRerdEii 5.lhnnyfor FDI 2IvbrkfOrm~ltedtO txes 32Wrkiwcgnai:fcmilyfdmds 5 2 ~ f o r m r e 5 w n - @ i i o r s 22hnnyforwnRictsinwuts 3.3Checkirgoroauiweoootnt Source: Colombia ICs-2006. Box 4. 1: Olley andPakesDecomposition The Olley and Pakes (1996) decomposition o f productivity is useful to evaluate the allocation efficiency o f fm by region and ineach year, duringa given period o f time. It allows determiningifan increase inproductivity is due to a reallocation towards more productive f m s or to an increase in average productivity growth. The Olley and Pakes (1996) decomposition is given by the expression: where E, is the average productivity o f industry k and 5 measures the covariance between productivity r=1 and output-shares. Ifthe covariance is positive, then the larger it is, the higher the share o f sales that goes to more productive f m s (Le., allocation efficiency) andthe higher is the indu productivity. Ifthe covariance is negative, it cannot be interpreted as allocation inefficiency, since the more negative it is, the higher is the output share that goes to less productive h s , reducing the industry'sproductivity. 43 4.24. Figure 4.16 reports the results of the Figure 4. 16: O&P decompositioninlevels of O&P decomposition in levels for each aggregate productivity (restricted Solow residual) metropolitan area, using the restricted Solow residual as the productivity measure (see Escribano et. al. (2007)). In this decomposition, the 5 percent upper and lower values of the probability distribution of the productivity were dropped. Results show that there were no significant differences in terms of average productivity across regions. Inaddition, the most efficient regions in terms of resource reallocation were Bogota and Medellin, which considerably improved the aggregate Aggregate productivity DAverage productivityIEfficiencyterm productivity of these two regions with respect to Notes: The productivity measure used is the restricted Barranquilla and Cali. Solow residual [equation (1) of Escribanoet. al. (2007)l. 4.25. Figure 4.17 presents the relative Source: Colombia ICs-2006. weight of each group of IC and control variables on aggregateproductivity, average productivity and the efficiency term through: (a) the percentage contribution of each group of variables in the O&P decomposition in logs (Figure 4.17 a), and (b) the simulations of improvements in IC and C variables on the O&P decomposition in levels (Figure 4.17 b). The impact of red tape, corruption and crime group, in terms o f average log-productivity, was more important in Barranquilla, Bogota and Cali, with weights between 20 percent for Medellin and 70 percent for Cali. The infrastructure group was the second most important, with the largest impact inMedellin. The relative impact of the finance and corporate governance group was low compared with the other groups in all regions. The quality, innovation and labor skills group only had a significant impact on the averageproductivity of firms inBogota. 4.26. Regarding regional allocative efficiency, the red tape, corruption and crime group had the highest relative impact in Bogota and Cali, whereas infrastructure was the most important group in explaining efficiency in Barranquilla and Medellin. The relative weights of the remaining groups were considerably smaller, and only the impact of the finance and corporate governance group inMedellinis worth mentioning. 4.27. Once the groups that have the largest relative impacts on efficiency are recognized, in what follows, the focus will be on the description of the results per IC variable obtained from the decomposition in logs by each Colombian region. Each variable may have two effects: one i s the effect on the log-productivity o f the average Colombian firm, and the other i s the effect on the efficiency of the region as a whole. Both effects may be positive or negative, and a positive effect on average productivity does not necessarily mean a positive effect on efficiency and vice-versa. The total effect on efficiency is the coefficient of the variable on the productivity equation multiplied by the covariance of the variable and the market share (see Annex I). 44 Figure4. 17: Relative IC effects by group of variables on aggregate productivity, averageproductivity and efficiency, by region (a) Percentagecontribution of IC and C variables to the O&P decompositionofthe aggregateproductivity inlogs d (b) Simulations of achange inIC variables andpercentage change inaggregate productivity andthe componentsof the O&P decomposition -rp(leRoductMty LJ Infrastructures Redtape, corruptionand crime 0 Financeand corporate governance 0 Quality,innomtionand laborskils Other controlmriables Notes: Decomposition inlogs and simulations of a 20 percent improvement in IC and C variables. In(a), relative contributions computedaccordingto equations(19)-(21) ofEscribanoet al. (2007). In(b), relative contributions computedaccordingto equations(30)-(32) of Escribanoet al. (2007). Source: Colombia ICs-2006. 45 Bogota 4.28. The variable with the largest relative contribution to the average productivity in Bogota was sales reported for tax purposes. Its effect was negative on both the average productivity and the efficiency term, and firms with the largest market shares were the ones that declared the lowest shares of sales for tax purposes.29 The variable training of production workers was also important, and its effect on average productivity was positive and large, while the effect on efficiency was low and negative. Therefore, firms with the largest market shares provided training to a lower share of their production workers, an activity that enhances productivity; as a result, efficiency decreased. Other variables with considerable impacts on average productivity were water from public sources (positive) and sales paid after delivery (negative). Other variables with large impacts on efficiency were the dummy for conflicts in courts (positive) and the number of tax inspections (positive) (Figure 4.18). Medellin 4.29. For Medellin, once again informality on taxes was the most relevant variable for average productivity and efficiency. The impact of sales reported for tax purposes was negative and large on both average productivity and efficiency, while the effect of the labor reported for tax purposes was positive on both average productivity and efficiency. Other IC variables affecting the average productivity were duration of monthly power outages (negative) and water from public sources (positive). Factors that affected efficiency to a large extent were mainly related to electricity supply - duration of power outages (positive), electricity from own generator (positive) and waiting time for an electricity connection (negative) (Figure 4.19). Cali 4.30. For Cali, informality on the taxes was, once again, the most relevant variable for average productivity and efficiency. The impact of variable labor reported for tax purposes accounted for more thanhalf of the total contribution of IC variables to average log-productivity inCali -presentingapositive impact. Transportation cost had a considerably negative and large impact on average log-productivity. Labor reported for tax purposes and the dummy for security were the variables with the largest impacts on efficiency. The former had a positive impact, which means that firms with more market share reported higher shares of their labor force for tax purposes. The impact of the security dummy was negative, implying that firms with larger market shares did not incur security expenses, an activity that reduced productivity in Cali (Figure 4.20). 29The covariance betweenmarket share and sales reportedfor tax purposes is positive. 46 Figure 4. 18: Bogota: Relative IC effects onthe Olley and Pakes decompositioninlogs 1- (a) Relative IC effects on averageproductivity (b) Relative IC effects on efficiency ana rnrp.gov *I$ 32 45 40 35 30 23 8 25 20 8 5 jj;i 8 n7 0 5 J, 00 0 T.l 11 12 13 T2 2.1 22 2.3 2A T.3 3.1 32 3.3 3.4 T.4 4.1 42 4.3 4.4 T,5 5.1 52 5.3 1.1 11 U 13 T2 2.1 22 2.3 2A T.3 3.1 32 3.3 3.4 T.4 4.1 42 43 4.4 1.1 infrastructures Total T.3 Total flnance and corp. governance T.4 Total quality, Innov. and l a b r skills 1.1Average duration of power outages 3.1 Initial investment:public banks 4.1 Dummy for environmental programs 1.2Water from public sourcas 3.2 Sales paid after delivery 4.2 Staff-female workers 1.3Wait for a water supply 3.3 working capital:famiiy/friands 4.3 Staff -umversityeducation 3.4Value of the collateral 4.4Trainin9 to productionwohers T.2 Total red tape corr. and crime 2.1 Sales reported to taxes Source: Colombia ICs-2006.6Total other control variables 2.2 Dummy for conflicts in courts 5.1 Dummyfor limitedcompany 2.3 Number of inspections 5.2 Dummy for FDI 2.4 Payments to obtain a contractwith the government 5.3Dummy for localmonopoly XT-*-. IYULC. CUICII CUIIIC)UIICIII F--L I-\ --A d aiu (u) CUIIIC~ ~iuiii IL\ c-- E--AL-..,. ,.+ L.DLIIU~~U ai. {LUU/ J. GL - 1 /1nn7\ Source: Colombia ICS-!dO6 I.~ Figure 4. 19: Medellin: Relative IC effects on the Olley and Pakesdecompositioninlogs (a) Relative IC effects on averageproductivity (b) Relative IC effects on efficiency _- IMrastrLC1ues Redtape,wrrlqtionL Finueeh QrslltylmJV 6 Mh3rmdrol Yrs(nrLC1uss Rdtspe corrption(i Fimneb ~ d l t Y l m J V(i Mtwwdroi Ulna cor0 oov labor sblis vnablss cnm corp gov labor sblis wiablm 40 35 30 25 20 8 XI 5 42 1 5 51 52 5 0 11 I1 12 13 14 15 16 12 21 22 2 3 24 1 3 31 3 2 3 3 34 T 4 4 1 T 1 11 12 13 14 15 16 T Z 21 22 2 3 24T3 3 1 3 2 3 3 3 4 1 4 4 1 42 T5 51 52 53 w T.3 Total finance and corp. governance T.l Total infrastructures 3.1 Largest shareholder 1.1 Duralion of power outages by monlh 3.2 Working capital: familylfriends 1.2 Electricityfrom a generator 3.3 Owner of the lands 1.3Wait for an eleclric supply 3.4 Dummy for loan with coliateral 1.4Water from public sources 1.5 Dummy own transport T.4 Total quality, lnnov. and labor skills 1 6 Shipment losses, domestic 4.1 Dummy for foreign technology 4.2 Staff -universityeducation T.2 Total red tape corr. and crime 2.1 Sales reported tot taxes T.5 Total other control variables 2.2 Workforce reportedto taxes 5.1 Age 2.3 Dummy for security 5.2 Exporting experience 2 4 Losses due Io crime 5 3 Dummv for more than 5 comostilori Note: Each component a and (b) comes from Escribano et al. (2007). Source: Colombia ICs- 06 Id 47 Figure 4.20: Cali: Relative IC effects on the Olley and Pakesdecompositioninlogs (a) Relative IC effects on averageproductivity (b) Relative IC effects on efficiency Rtdtrpe,mnlption&mnn I 9c ~i-a m i y i m . & uhmrtroi m s68 586 M) 9) 40 32 1 30 20 2 3 0 0 21 22 23 24 2' '4 4 1 42 15 51 52 T,1 11 12 13 T2 2.1 22 2.3 2.4 25 T.3 3.1 32 T,4 4.1 42 T5 5.1 52 Leaend 1.1 Total infrastructures T.3 Total finance and corp. governance 1.1 Losses due to power outages 3.1 Initial investment: private banks 1.2 Wait for an electric connection 3.2 Working capital: familyflriends 1.3 Average duration of water outages 1.4 Transport cost 1.4 Total quality, Innovationand labor skills 4.1 Dummy for quality certification T.2 Total red tape corruptionand crime 4.2 Outsourcing 2.1 Workforce reported to taxes 2.2 Dummy for security 1.6 Total other control variables 2.3 Dummy for crime 5.1 Exporting experience 2.4 Managefs time spent In bur. issues 5.2 Dummy for local monopoly 2.5 Dummy for payments to obtain a contract with the gov. 5.3 Dummy more than 5 competitors Note: Each component a and (b) comes from Escribano et al. (2007). Source: Colombia ICS-!ldO6 Barranquilla 4.3 1. For Barranquilla, informality on taxes was, once again, the most relevant variable for average productivity and efficiency - with positive and large impacts. Other variables with relatively large contributions to average productivity were transportation cost (negative), waiting time for an electricity connection (negative) and the security dummy (positive). Transportation cost was the most important determinant o f efficiency, presenting a positive impact, which means that firms with the largest market shares paid the lowest transportation costs. Other determinants o f efficiency were waiting time for an electricity connection (negative) andprior exporting experience (positive) (Figure 4.21). Summary 4.32. In sum, six different productivity measures were used in order to get consistent and robust estimates (elasticities) of IC determinants of productivity, following the methodology o f Escribano and Guasch (2005). Most of the I C variables had the expected signs, and the estimated elasticities or semi-elasticities were within reasonable ranges for the six different productivity measures that were considered. The robustness o f the empirical results, across productivity measure, allows policymakers to obtain consistent empirical results for policy evaluations o fthe I C effects on aggregate productivity by region. 48 I Figure 4.21: Barranquilla: Relative IC effects onthe Olley andPakesdecompositioninlogs (a) Relative IC effects on averageproductivity (b) Relative IC effects on efficiency I hfmt1~3uas Wt~mmptionaairm A m a Qslhyimw. abmiiml mm.oov. al&rwb nri- I I I75 6 4 43 39 36 5033 T I 04 12 08 14 11 13 n n 48 04 16 n 1 0 4 I T.1 Total Infrastructures T.5 Total flnance and corp. governance 1.1 Losses due to poweroutages 3.1 Initialinvestment: private banks 1.2Wait for an electricconnection 3.2 Working capital:familytfriends 1.3 Average duration of water outages 1.4 Transportcost T.4 Total quality, innovatlon and labor skills 4.1 Dummyfor quality certification T.2 Total redtape corruptlon and crime 4.2 Outsourcing 2.1 Workforce reportedto taxes 2.2 Dummyfor security T.6 Total other control variables 2.3 Dummyfor fflme 5.1 Exportingexperience 2.4 Managet`s time spent in bur. issues 5.2 Dummy for local monopoly 2.5 Dummytor paymentsto obtaina wntract with the gov. 5.3 Dummy more than 5 competitors Notes:Eachcomponent in (a) and (b) comes 6-omEscribanoet. al. (2007). Source: Colombia ICS-2006. 4.33. Strong statistical evidence was found o f relationships between aggregate productivity and I C variables by region. The results were consistent with firms' perceptions on the main obstacles to their growth. The IC variables with the strongest impacts on average productivity and efficiencywere (tax) informality,corruptionandinfrastructure. 4.34. The policy implications are clear: the investment climate matters and the relative size o f the impact o fthe various IC variables is not the same ineach region, indicating that the areas in which reform efforts should be placed differ from region to region. 49 CHAPTER 5 - PRODUCTIVEINFRASTRUCTUREAND SUB- REGIONALCOMPETITIVENESSINCOLOMBIA 5.1. This chapter reviews the economic interregional links in Colombia, which are generally perceived as weak; the freight transport difficulties across the territory appears to be one of the potential causes, which has reduced the potential competitiveness of the country and its sub- regions. The performance and condition of the freight logistics system has been reviewed, and the findings confirm that Colombia is experiencing some severe shortcomings, not restricted only to infrastructure but to a more complex agenda of transport services regulation, private sector development and trade facilitation. A brief survey of some value chains suggests that-under the new international trade scenario-the low spatial interaction that characterized Colombia's economy may experience some increase. The Chapter ends with policy recommendationson the transport-logistics sector that may contribute to enhancing sub-regional competitiveness, and taking stock of the emerging opportunities. 5.1 Introduction 5.2. The new international trade scenario presents a favorable horizon for Colombia's economic growth; the new opportunities might reinforce the current geographical pattern, characterized by a high concentration of activity and income in few regions, or the new opportunities could be shared throughout the country. New opportunities for economic growth emerged in the context of trade liberalization; international trade (exports plus imports) doubled between2000 and 2006, growing at an annual rate of 12.5 percent and are expected to keepgrowing at similar rates due to the globalization of trade and, particularly, the FTA withthe US. The benefits from trade may spill over from territory to territory, or it could just concentrate inthe regions where the output is already concentrated. This outcome will depend on the inter- sector and interregional links, which are the result of the spatial pattern of the backward and forward linkages of the productive activities, which, in turn, are related to a myriad of other factors influencing sub-regional competitiveness, transport and logistics. There is a considerable amount of research on the interregional links in Colombia, and there i s some on the role that transport plays in explaining these links, but the researchis usually constrained to infrastructure network limitations. Colombia is an example of a growing concentration of economic activity. Therefore, the new growth may reinforce the current pattern, or, accompaniedby the appropriate policies, may helpto change that pattern and promote a more balancedregional development. 5.3. Infrastructure services constitute a relevant factor in the spatial organization of economic activities; transport and logistic-he infrastructure sector on which this Chapter will focus-can ease or worsen interregional connectivity, influencing the spatial interaction among economic activities and sub-regions in terms of competitiveness. In the case of Colombia, logistics costs are particularly relevant because of the country's geography: the three major economic hubs (Bogoth, Medellinand Cali), which concentrate 75 percent of the national GDP (excluding mining activities), are located in the Andean regions, as is shown in Figure 1.3; the average distance to/from ports is severaltimes longer than inother LAC countries (World Bank, 2005, pag. 13). The topography of the country, with three main mountain ranges 50 crossing through it, makes freight transportation particularly difficult. Although freight logistics is a major factor in sub-regional competitiveness, other infrastructure services are also very relevant, particularly the provision o f electricity. As discussed in chapter 4, power outages and the number o f outages in Barranquilla was considerable higher than other urban centers. This fact constitutes a clear competitiveness disadvantage. 5.4. The objective of this chapter is to review the current perception of the economic interregional links and the current performance of the logistics system, contrasting the these resultswith the results from a recent survey of selected value chains,which suggest that the low spatial interactionthat characterizedColombia's economy may be changing. The Chapter ends with policy recommendationson the transport-logisticssector that may contribute to enhancing sub-regional competitiveness under a new trade scenario. The chapter consists o f four sections. (i) The first section includes a review of the spatial structure o f the interregional and inter-sector links o f the Colombian economy, looking at the results from diverse approaches, which end up with a general consensus: there is limited spatial interdependence, links are weak, Colombia's regions are "closed". This view suggests that a potential increase in demand from products generated in one region will not have relevant backward and forward linkages to other regions. (ii)The second section assesses the performance o f the county's logistics system, identifying its main weaknesses and discussing to what extent the transportation difficulties are one o f the major the determinants o f the limited internal spatial interaction. The analysis i s based on a conceptualization o f logistics performance that goes beyond the conventional infrastructure-based perspective, based on a recent WB ad-hoc assessment and an across-country logistics performance comparison that led to the estimate o f a worldwide Logistics Perception Index. (iii) A new avenue is explored in the third section: the analysis o f the sector-regional links o f selected value chains, surveying several firms and capturing the logistics problems they currently face and the location strategy they are adopting to expand their activities. Although a small sample was surveyed (two chains and five firms ineach chain), the results suggest that the weak spatial interaction paradigm that characterizes Colombia's economy may be changing, at least partially, and can be changed further if the logistics performance i s improved. (iv) The fourth section presents conclusions and recommendations inorder to maximize sub-regional competitiveness. 5.2 The GeneralConsensusofLowInter-SectorandInterregionalLinksinColombia's Economy 5.5. The spatial structure of Colombia's economy has been the subject of considerable research;severalavenues havebeen explored, demonstrating,with consistent results,weak inter-industry and interregional interactions. Many o f the analyses were carried out in relation to the policy discussion on the decentralization model implemented after the 1991 Constitution, particularly by the Center for Regional Economic Studies (Centro de Estudios Econdmicos Regionales, CEER) at the Banco de la Republica. The results, as will be seen, coincide in that the weak sector relationships contribute to preserve, and even exacerbate, regional disparities, expressed in a growing gap between rich and poor regions (Galvis Aponte 2001, Baron and Roca 2003, Baron 2003, Bonet 2006, Roca and Romero 2007). Three approaches have been selected to review the inter-sector and interregional links in Colombia: (i) 51 the analysis of the existing Input-Output (IO) matrix, (ii)a parsimonious approach, aimed at capturing the essence of the regional interactions, and (iii)an interregional IO matrix model, which estimates the sector and regional links. The first approach is based on the DANE databases, and the other two approachesreproduceresearchmadeby Jaime Bonet in2005. 5.6. The input-output matrix gathered in 2004 at the national level shows an updated picture of the economic structure, displaying a weak interaction among the 36 goods- producingsectors; the transactions among these sectors represents only 23 percent of the GDP, well below the ratio in other more integratedeconomies. The input-output (IO) matrix i s the model that summarizes the relationship among sectors within the economy, displaying the extent to which a sector constitutes the inputs of another. In addition to intermediate inputs, the IO matrix also includes the final consumption of goods and services. In Colombia, DANE has generated a use matrix, the "Matriz de Utilizacion'y,the latest version from 2004, which includes the relationships among 60 activity sectors, out of which 36 are goods producing sectors. The matrix shows that 14 pairs out of the potential 36x36 (1,296) concentrate half of the goods producing sectors, and that 57 pairs explain 75 percent of the transactions. (Table 5.l).30 Results put on view two types of interactions: inter-sector and intra-sector. The results dependto a large extent on how sectors are defined. The secondtype of interaction represents about 24 percent of the interaction between goods producing sectors. Table 5.1 shows the 14 transactions between sectors that make 50 percent of the domestic intermediate goods flows. Transactions between garment and apparel, or between oil by-products and crude oils are examples of inter-sector links. Transactions betweentransport equipment, livestock and chemicals i s an example of intra- sector links. Resultsclearly dependon the industry classification. 5.7. An analysis of regional economic interaction- based on parsimonious analysis - shows that the main seven regions in the country compete with, rather than complement, themselves: the growth of one of them results in a decrease of share in the other regions. The analysis is based on researchcarried out by Bonet (2003, 2005), who adoptedthe so called Dendrinos-Sonis approach to assess the competition or complementarity among regions3'. The country is divided into seven regions, and the period covered by the analysis i s 1960-96. Results are shown in Table 5.2, displaying estimates of the relationship among every pair of regions. If the coefficient between two regions has a negative sign, it signals that there is a competitive relationship (if one region were to increase its share of GDP, the other would reduce it). If the coefficient is positive, there is a complementary relationship (if one region were to increase its share of GDP, the other would increaseas well). Measurement i s made again a numeraire region; the Bogota region - with the higher GDP per capita - is utilized as the numeraire. The table reveals a high level of competition among regions. "The regions that have the major proportion of the national GDP - Bogota, West-Central and Pacific - as well as the one with the highest growth rate-New Departments -exhibit a competitive relationship. This means that an increase in the share of the most dynamic economies will result in a decrease in the share of other 30 In other countries the domestic intermediatematrix represents more than 23 percent of GDP, as is the case of Colombia. 31 The Dendrinos-Sonis (1988, 1990) developed a nonlinear model that measures the relative share of GDP of diverse regions over time, comparing whether the growth of some regions is at the expense of other regions, or whether they are complementary.Elasticitiesare estimatedfor eachpair ofregions. 52 regions." (Bonet, 2003:9). The New Departmentsexhibit only a positive relationship with itself, reflecting its poor integration with the rest of the country. Table 5. 1: Inter-industrylinkages fiom the 2004 input-output matrix Production in this sector ... ...demands from this sector YOinteraction Meat and fish Livestock and animal products 11.5 Basic chemicalproducts(except plasticsand Indystrial chemicals(except plasticsand 6.5 M;il pioducts Other agricultural products 5.0 Garment andapparel Garmentand apparel 3.1 Livestock and animal products Coffee 3.O Diary products Livestock and animal products 3.O Refined oil products Crude oil, gas 2.9 Metal products Metalproducts 2.8 Livestock and animal products Livestock and animal products 2.6 Rubberand plastic products ' Ind,ustrial chemicals(except plastics and 2.4 Paper, paper products and printing Paper,paper productsand printing 1.9 Other agriculture products Glass andother non metalic mineral products 1.8 Transportequipment Transport equipment 1.8 Transportequipment Other machinery 1.3 Source: elaboratedon the basis of DANE `s Matriz de Utilizacibnde Productos2004. 5.8. A multiregional IO matrix, which estimates the inter-industry flows between and within regions, allows the identification of the key sectors with backward and forward linkages in each region and the degree of interregional input trade in the country, which is low when 1997 is analyzed: most regions show a high degree of independence, the New Departments being the exception. The IO matrix can be disaggregated at the regional level, reflecting the geographical origin of the inter-industry links. The data collection for an IO multiregional (MR) matrix is cumbersome; it can be also estimated, utilizing location quotients, which has been done for Colombia's 1985, 92 and 97 IO matrices (Bonet 2005). The IO MR matrix is made the comparison of 7 regions and nine aggregated economic sectors. Bonet tries two analytical techniques to depict the linkages among sectors and regions. The first one is the standardized pure linkages index, which reflects the interregional linkages among sectors. The second one i s the MultiplierProduct Matrix (MPM), inwhich the row-column crosses (reflecting backward and forward linkages) are sorted indescending order, allowing the visualization of the regional economic structure. Table 5.2: Qualitative analysisofthe competitive and complementaryrelationships amongregions De ts, New West- Central North- Southcentral Caribbean Pacific Caribbean + + + + Bogota Central New Depts. + South-Central Note: + indicatescomplementarity while - indicatescompetition. Source: Bonet (2005). 5.9. Table 5.3 identifies the sectors with the most relevant backward and forward linkagesin each regionfor the lastyear for which the analysis has beendone. At the national 53 level, the sectors with the strongest backward impact are non-durable manufacturing and construction; each region shows a different profile. The MPM results, based on a high level of aggregation (seven main regions), show that in the regions where economic output is concentrated (Caribbean, West Central, Bogota and Pacific), more than 90 percent of the inter- industry trade is made within the same region. In South Central and North Central this ratio is around 80 percent, and on1 the New Departments shows a relatively high interregional inter- industrytrade (40 percent).3Y 5.10. In summary, the existinganalysis concurs inthat most sectors are self-sufficient at a regional scale in Colombia, with little interregional interaction, and most forward and backward linkages are concentrated inside the richest regions; these results suggest that the regional polarization may be perpetuated. The result of the extensive researchcarried out by Bonet concludes that interregional links are indeedweak inColombia. One implication is that growth, under this economic spatial pattern, will likely not help to reduce the current regional imbalances, but could even reinforce them. Previous studies, perhaps not as sophisticatedas this one, hadreachedsimilar conclusions. (Galvis Aponte, 2001,Roca and Romero, 2007). 5.11. Some analysts perceive the transportation costs, associated with weak infrastructure, as one of the leading causes of the low spatial interaction. Transport infrastructure, and its associated services, constitute the physical link between production and consumption centers, as well as the key trade gateways. With different perspectives, several analyses have pointed out that transport sector weaknesses may be reducing the potential for economic activity in Colombia, constituting an obstacle to greater interregional links within the economy (among others: World Bank 2004c, 2005; Cardenas, et. al., 2005; and Perez 2005). In some cases, the analysis is focused on transport infrastructure networks, while other analyses also take into consideration transport services and the institutional factors behind their performance. The next section presents a review of the condition and performance of this sector, characterizedas the freight logistics system. Table 5. 3: Backwardandforward linkageorientedsectors, basedon standardizedpurelinkagesindices Region Backward linkage Forward linkage oriented sectors 1997 - oriented sectors 1997 - Non-durablemanufacturing Agriculture Caribbean Durablemanufacturing Durablemanufacturing Government Utilities Privateservices Agriculture South Central Non-durablemanufacturing Durablemanufacturing Government Utilities Private services - - ..--- --...--- Non-durablemanufacturing Durablemanufacturing Bogota Durablemanufacturing Utilities Construction Privateservices 32FromBonet(2005:33). 54 Region Backward linkage Forward linkage orientedsectors 1997 - orientedsectors 1997 - Wholesale and retail Utilities Durable manufacturing Mining North Central Wholesale andretail Utilities Utilities Government Private services Pacific Mining Agriculture New Departments Non-durable manufacturing Mining Utilities Utilities Government Private services Source: Adapted from Bonet (2005), Table 8. 5.3 The FreightLogisticsSystem Responseto the SpatialInteractionDemands 5.12. A complex logistics system, composed of transport infrastructure and services, businesslogistics practicesand trade facilitationprocedures,is responsiblefor the physical flows; a comprehensive framework integrating the logistics system is emerging, which substitutesthe one that focused exclusively on infrastructurenetworks.Several World Bank studies have analyzed the link between competitiveness and the physical flows of goods (World Bank 2004c, 2005, 2006), concluding that three major areas have to be dealt with in order to ensure the flow of goods throughout the logistics chains: (a) transportation, (b) business logistics, and (c) trade facilitation. One conclusion derived from this perspectiveis that the policy levers to improve logistics competitiveness are not just infrastructure provision, but services regulations, private sector development and the direct provisions of some services by the Government (Le., inspections at gateways). This framework encompasses the flows of international and domestic trade as well, although some components can be more relevant in one case or another (customs inspections for international trade, trucking industry performance for domestic trade). Most productive activities comprise export, import and domestic flows along the value chain. 5.13. A recentBank report-following this framework-found that the main weaknesses inthe freight logisticssystem in Colombiaare notrestrictedto infrastructure,butdealwith regulationand government-managedprocesses. Colombia's logistics weaknesses are usually perceived as a lack of adequate infrastructure. The World Bank (2005) report-based on a supply-side analysis and a survey of users' viewpoint-found that, although some relevant infrastructure bottlenecks do exist, the freight logistics systemhas other critical problems. Three main areas were detectedas particularly critic: (i) trucking industry, (ii) use ports, and the public (iii) inspections ininternational gateways (Table 5.4). 55 HIGHWAYS Bottleneckin critical segments, geometry TRUCKING INDUSTRY Low efficiency and quality of service Lack of efficient standards, procedures and informationsystems: I e port-truck SCM ORGANILATION SMES with logistics costs three time higher than average CONTROLS High inspectioncosts, lack of coordination among agencies Source: Adapted from World Bank (20070. 5.14. The regulatory organization of the trucking industry-by far the most important surface transport mode in the countryls curtailing its efficiency; this does not represent higher freight cost and lower quality of service, but hampers the firms' ability to innovate their logistics strategies. 81 percent of internal freight is moved by truck in Colombia; it is by far the most important transport mode for internal traffic and a relevant one for exports and imports to/from countries in the region (mostly Venezuela and Ecuador) (World Bank 2004c, 2006). Figure 5.1 shows truck flows across the country. The performance of the industry is low: trucks average around 50,000 kmper year (should be about twice that figure), there are too many trucks for the country's geographical and economic dimensions, the fleet i s aging, the industry structure is too fragmented, and only a few companies are evolving into logistics operators. This fact may result not just in reducedproductivity and increasing costs, but also in lower service quality, which is becoming progressively important inthe modern supply chain practice. Several studies show that the modernization of the trucking companies has a strong impact downstream, as it allows users (shippers) to innovate their supply and distribution strategies. The reason for the low performance of Colombia's trucking industry is likely to result from-among other factors-its regulatory organization: the multilayer system, the multiplicity of intermediaries adding little value, and the rates reg~lation.~~ There is a relevant cultural factor in the sector performance, as well as a delicate social issue, and reaching a solution faces complex political economy obstacles. 33Exogenous factors: security on the roads, difficult topography, fieight demand imbalances. 56 Figure 5. 1: Flows of trucks on Colombian roads Source: Roda -USAID. 5.15. The public ports improved substantiallysoon after the 1993 reform, but they seem to have been left behind afterwards, although performance varies a great deal among ports; the organizationalstructure that was adopted-a hybrid between a landlord model and a tool port model-may be one of the main reasons for the current constraints. Colombia's port system was reformed in the early 1990s, concessioning the key public ports, which improved substantially compared with their previous conditions. But after more than 12 years, the public ports service quality in Colombia looks to be loosing ground when compared with other ports in the region. The picture is uneven, with some ports delivering better services (like the Cartagena container terminals) and others showing some serious congestion problems (like the Buenaventura multipurpose A recent Latin America Competitiveness Review 2006 (preparedby the WEF) shows Colombia as #58 in a 117 country rank, but number 85 in regards to ports. The explanation for the public ports difficulties may lie in the following explanations: the changes to the maritime and port industriesthat took place after the reform, the weaknesses of the port structure organization adopted, or the flaws inthe reform implementation, such as government oversight and control. A recent analysis of the evolution of ports in Latin America finds that the application of a landlord model has been one of the key success factors, particularly inthe efficiency of terminal operations. Inthe case o f Colombia, the model that was adopted is a hybridbetween a landlord anda tool port organization structure. The lack of service standards requirements, the multiple-operators scheme, the high participation of shippers inport 34Cargo Systems, 2007, highlights Buenventura problems from the users' perspective. 57 governance, concessionaires fee determination, and the absence o f environmental and security clauses are some o f the concession design aspects that may be hampering better development o f port activities. 5.16. The inspectionsin export-import gateways are performed by many agencies, which control a proportion of the international trade that is several times higher than in comparable countries; in addition to some flaws in each agency's procedures, the coordination among them is indeed weak. Inspections in ports are currently more than 30 percent focused on exports and 20 percent on imports, approximately three times the standard in developed and MI countries. Many agencies participate in the process, there are some flaws within each agency's control procedure (i.e., the physical customs inspection damaging exports), and weak coordination among them. As a result of this cumbersome process, the extra cost generated in container movement has been estimated at more that US$ 300 per unit (World Bank, 2005). 5.17. A recent worldwide survey of operational staff in freight forwarding shows that Colombia logistics performance is frail, even by LAC regional standards, particularly in the management of customs and border procedures; hard data on logistics performance demonstrate that the inspection process is particularly critical. The Logistics Perception Index (LPI) was estimated for 150 countries, based on seven sub-indexes; hard data on the logistics environment and physical performance indicators were collected as well for 110 countries.35 Colombia ranked 82 out o f 150, with some sub-indexes performing very low, particularly "Efficiency o f Customs and Other Border Procedures". Table 5.5 summarizes the LPI and its sub-indexes. It shows that Colombia's weaknesses are not just infrastructure, but mostly in the inspection procedures and logistics service organization. Table 5.6 confirms that general cargo i s inspected at too high a proportion (35 percent), that damage in the inspection process i s relevant, and that the cost of moving containers is higher than in other comparable countries. 35World Bank -GFP -Turku: Measuring Global Connections. 58 59 Colombia 82 116 85 74 86 71 80 87 Chile 32 24 34 34 35 37 115 44 Argentina 45 51 47 49 44 46 93 46 MBxico 56 63 53 54 57 48 101 51 Peni 59 49 57 53 61 67 59 80 USA 14 19 7 20 14 10 144 18 Source: World Bank (2007f). Table 5. 6: Trade facilitation indicatorsfor Colombia and other selected countries Average Average 40' Cont. 40' Cont. Country Time Inspec- release tion as Yo Damaged export import export import shipmt. lead time lead time cost cost Colombia 7.0 8YO 4.0 10.0 2000 2000 Source: World Bank (2007f). 5.18. As a conclusion,different sources agree that freight logisticspresentseveral relevant problems in Colombia, and that the major weakness is not the lack of adequate infrastructure-although there are some clear deficiencies-but in many regulatoryissues, in the development of business logistics practices, and particularly in the inspection customs andborder processes managedby official entities. 5.4 Perspectivesfrom a Micro Approach 5.19. An alternative approach, the analysis of value chains, has also been pursuedin order to further analyze the inter-sectoral and inter-regional links, looking at the supply chain current spatial patterns and-particularly-the potential trends facing the FTA. Two value chains were selected as examples: fruit-horticulture and glass ceramics. The value chain approach intends to overcome one of the I O matrix weaknesses: results depend on industry classification and aggregation. Instead of the traditional sectors, activities are grouped as echelons ina chain, from the extractive activities to final production of goods, going through intermediary goods, following client-supplier links. For example, the textile chain covers the products from cotton production to garment delivery, including all intermediate stages. Colombia has done intensive work on recognizing and analyzing the key productive chains, motivated by the challenges expected from the FTA with the U S (DNP, 2006). This approach may yield further insights into 60 the inter-industry and interregional relations; it allows exploring what are the supply chain current spatial patterns, where the inputs to each chain echelon come from, what are the obstacles they face nowadays, and, if activity grows, what are the current location trends. Two value chains with high growth otential were selected and explored through interviews: fruit- horticulture and glass ceramics.3 9 5.20. In juice production-a growing activity within the fruit-horticulturechain-inputs havediversifiedregionalorigins, dependingon the fruit, and elaborationplantsare located in the main industrialcenters (particularlyValle del Cauca). About half of the product is exported, but the domestic market, concentrated in the three main cities, also remains important. Firms are willing to keep their current locationto serve the domestic market, while expandingtheir activitiescloser to the coast in order to deal with the rising exports. Juice production i s the most relevant activity in this productive chain (44 percent o f the chain output), with mango as the main input. Juice i s mostly produced inValle del Cauca; mango pulp i s produced on the Atlantic Coast and partially in Tolima and Huila. The flow is seasonal. Mango has two harvests a year, which gives it a competitive advantage for Colombia (only one harvest in the maincompetitor producer, India). Other fruits come from diverse regions: bananas from Uraba and Magdalena, mora from Cundinamarca and Antioquia, oranges and ZuZo from Antioquia, maracayd from Huila, Antioquia and Valle del Cauca. Half o f the juice i s exported in containers (mostly to the US, Germany and Ukraine), and the other half i s consumed domestically, distributed through Bogota. The firms' location strategy consists in developing new export-oriented plants on the Atlantic coast, in the Barranquilla area, while simultaneously developing suppliers close to the Valle del Cauca plants. 36 The selectionwas made in agreement with DNP 61 5.21. The internaltransportationof mango pulpcosts aroundthree times the productfree alongside ship (FAS) value; the main problems the firms face are customs and border procedures,rural roads and truckingactivities.Total logistics costs have not been estimated, but interviews indicate that surface transportation alone costs more than three times that of the FAS product value. The main problems along the logistics chain are customs and border procedures, low quality of rural roads linking the fruit production units to the main highways network, andthe low quality of the domestic trucking services. Recurrent problems inCartagena are forcing exports from the Valle del Caucathrough Cartagena. 5.22. In the glass and ceramics production chain, raw material location leads the placement of the product manufacturing plants, whose output is still directed, to a large extent, to the internalmarket; port location(mostly Cartagena) determinesthe locationof the logisticsplatformsof those firms that importand distribute.New plants are expected to be locatedon the Atlantic coast. In ceramics, wall and floor tiles make two third of the chain output; the main materials are gres 37-mostly oriented at the domestic market-and standard ceramic-mostly for export; sanitary porcelain is also relevant. Most inputs originate from Norte de Santander, where clay and sand have outstanding quality, and Cundinamarca, close to Bogota. Some feldspar is transported from Santander and kaolin from Tolima. Manufacturing plants are located close to the mines in the Bogota and Cucuta areas. Some inputs are imported, via Cartagena, from the US, Europe and Mexico. Today more than 80 percent of the production is for the domestic market, of which almost three-forths is directedto the three main urban centers. Exports go to the US, Europe and Central America. Facing a growth in exports, firms are considering the installation of new plants on the Atlantic coast, close to Barranquilla. 5.23. Glass-ceramics firms main logistics problems are customs and border procedures, port costs and service quality, and the domestic truckingservices costs and transit time. In the export flows, inspection procedures frequently damage the cargo ina value that firms assess as equal to the their business benefit. In terms of port and maritime services, the sector firms complain that the Cartagena container terminal prioritizes transfer traffic instead of national export-import traffic; therefore, shipping companies sometimes cannot drop off all containers, and imports must make detours to other ports. All these practices extended transit times. Buenaventura port is used for flows to and from China and the South American Pacific coast. Firmsfind port coststhere to be particularly high. It is a common practice to consolidate and de- consolidate containers in port yards in order to better utilize truck capacity, which adds to port congestion. The domestic trucking industry is blamed for high costs and transit time; some firms that are experimenting with thirdparty logistics providers are giving very satisfactory responses. 5.24. The conclusion from a preliminary review of two productive chains is that, confrontingthe increasein exports and imports, firms are willingto expand their capacity to produce/distributeon the Atlantic coast, which may raise the spatialinteractionthrough growing forward and backward linkages in the Caribbeanregion. Moreover, the logistics problems that firms declare are only partially infrastructure shortages and are also problemsrelatedto regulationand governmentmanagement. Inthe two analyzedproductive chains, the main location factors for the producing plants are the domestic market, the raw material sources andthe import gateways. With the perspectiveof increasedexports and imports, 37Fromthe Frenchword grds,a ceramic paste with little porosity after cookedandusedfor wall and floor coverage. 62 firms are planning to increase their capacity close to ports, particularly on the Atlantic coast and in the Barranquilla area (no interviewed firm showed interest in locating activity close to Buenaventura). This trend may produce some increase in the interregional linkages: new agriculture products originating from diverse regions and going to manufacturing plants, flows of raw or semi-processed material from inner regions to plants located in port areas, and development of logistics platforms on the coast supplying the inner domestic market. Most of the flows derived from this activity will follow the existing infrastructure corridors. The logistics problems that firms are going to confront confirm the hypothesis of the previous section regarding the importance of factors such as the trucking industry, ports, and customs and border procedures. 5.5 Conclusions andRecommendations 5.25. The previousanalysisleads to the followingsummarized conclusions and suggested policy recommendations: Colombia's economy is currently characterized by weak inter-industry and interregional economic links; most forward and backward links are concentratedinthe richest regions. Transport logistics i s clearly one of the major factors constraining spatial interaction. This constraint is not only due to the lack of coverage of infrastructure networks (ie., roads) or inadequate standards and capacity, but also to a set of diverse interacting features such as transport regulatory issues, private sector (shippers and carriers) business logistics development, and government managedprocesses (mainly customs and border procedures). 0 The increasingly open trade scenario offers opportunities to redefine spatial backward and forward linkages, as firms' strategies frequently consider the relocation of production and distribution facilities inorder to take full benefitof the new situation. The production chain analysis is a useful tool to identify the locational trends, the emerging spatial supply chain patterns, and the type of reform of the freight logistics systemthat may help increase spatial interaction. 0 The analysis made (at a preliminary level) in two productive chains-fruit-horticulture and glass-ceramics-shows that firms want to continue utilizing their current facilities, located according to raw material sources and domestic markets, to serve internal demand. These firms plan to expand capacity in order to serve export markets, locating new facilities on the Atlantic coast, particularly inthe Barranquilla area. 0 The materialization of these plans may increase interregional links. New agricultural products originating from diverse regions and traveling to more scattered manufacturing plants, flows of raw or semi-processedmaterial from inner regions to plants located in port areas, and developmentof logistics Platforms onthe coast that supply the inner domestic market. 0 Government policies may help to facilitate these new potential flows; these policies include enhancing infrastructure networks, but the analysis confirms previous findings that 63 special attention should be placed on the trucking industry performance, on the organization and management of ports, and on the inspectionprocesses at gateways. 0 Competition and service innovation on road freight logistics has strong downstream benefits, as it enables user industries to develop more efficient supply and distribution strategies. This innovation is not merely a reduction o f the carriage cost. There are international examples: MCxico, Czech Republic, Hungary, and Poland (Dutz 1995, 2005) are focusing on the impact o f service innovations on user industries. 0 As a result o f a relative low public investment inproductive infrastructure duringthe 90s (World Bank, 2 0 0 4 ~ )the focus in recent years has been on infrastructure development, ~ sometimes overseeing the importance o f the "soft" side of the agenda: policies to enhance transport services, business logistics organization and trade facilitation. The shift towards a "wider" agenda, including all those soft aspects, requires an update o f the institutional organization. Two initiatives may be considered in this regard: a reform and modernization o f the Ministry o f Transport, and the creation o f a National Logistics Council, coordinating public and private actions, at bothnationaland sub regional levels, to increase competitiveness. 64 CHAPTER 6 - HUMANCAPITALAND INNOVATION: NATIONALAND SUB-REGIONALANALYSIS 6.1. Innovation and human capital are major drivers o f economic growth and productivity. Roughly half o f cross-country differences in per capita income and growth are driven by differences in total factor productivity, generally attributed to technological development and innovative capacity (Dollar and Wolf 1997, Hall and Jones 1999). According to Prescott (1998), to understand large international income per capita differences, it i s necessary to explain differences and growth in total factor productivity (TFP). The argument i s that one o f the main candidates to explain those gaps i s resistance to the adoption o f new technologies and to the efficient use o f current operating technologies, which in turn are conditioned by the country's institutional and policy arrangements (investment climate variables). 6.1 HumanCapitalEndowment 6.2. What role do investmentsinhumancapital play inproductivity growth?38While there are often disagreements about the contribution o f human capital to productivity, there are several stylized facts that can be gleamed from the theoretical and empirical literature. 0 Increases in human capital (both health and education status) have a positive impact on productivityand economic growth. From a theoretical perspective, ina now classic paper, Lucas (1988) makes a strong argument about the spillover effect o f education on other workers and economic growth.39 From the empirical side, the work o f Barro and Sali-i-Martin (best summarized in Barro and Sali-i-Martin, 1999) show broadly the contributions o f human capital to growth and productivity. 0 The quality of human capital is at least as important as the quantity of human capital. Many early studies focused strictly on the accumulation o f human capital (for example, on the number o f years enrolled). More recent evidence shows that what and how well students learn i s at least as important as how much they learn and that by controlling for school quality and cognitive ability, the impact o f education i s often substantially greater. Haunushek and Wobmann (2007) rovide detailed evidence on the relationship between education quality and economic growth.4 9 0 Workers with higher humancapitalearn more and are moreproductive.Regardless o f the debate on the macroeconomic impact o f human capital, there i s clear evidence that people with higher levels of human capital (both health and education) have higher incomes and are 38Human capital is a broad topic (Schultz, 1961) that encompassing "investments" in health, knowledge, and educationat different ages. 39Morerecentreviews are providedby Aghion andHowitt(2003), from a macroeconomicperspectiveandBasu (1997), from a developmentperspective. 40By itsnature, it is moredifficult to differentiatebetweenhealthquantity andhealthquality. 65 more productive. Early investments tend to have long term impact. Healthy and well-educated individuals have long-term advantagesthrough their lives. 6.3. Traditionally, Colombia has had low levels of human capital. In recent years, it has invested a significant amount in health and education which have lead to some positive results. The government has set ambitious goals: full enrollment inbasic education (grades 1 to 9) and full enrollment in the managed competition insurance system by 2010. At the same time, the government has also committed to substantially increase the number of graduates from higher education. 6.4. In internationalcomparisons, Colombia generally performs lower than expectations in education outcomes, given its income level. Table 6.1 compares education spending and education enrollment rates in Colombia with other peers in LAC. Colombia does not appear to have a particularly high enrollment rates compared to other countries in the region, even taking into account the fact that Colombia is a poorer country. Colombia also appears to be one of the higher spenders on education, interms of percentageof GDP. Table 6. 1: International Comparisons Country GNI per Public Education Secondary Net Tertiary Gross Capita Expenditure (% GDP) Enrollment Rate Enrollment Rate Argentina $4,470 4.0 80.8 61.1 Boiivia $l;olo 6.2 72.7 40.6 Brazil $3,550 4.2 74.5 20.1 Mexico $7,3 10 5.3 63.8 23.4 Paraguay $1,040 4.4 n.a. 25.9 Peru $2,650 2.4 68.8 33.4 Colombia $2,290 5.2 54.9 28.3 L A C $4,045 4.5 68.1 27.2 Upper Middle Income $5,630 4.4 74.5 44.5 Notes: GNI= Gross National Income; n.a.=not available. Source: Ed Stats. 6.5. Local governments play an important role in education at all levels. As a percentage of their expenditures, education takes by far the largest share of their budget, averaging around 25 percent in 2006. With the exceptions of a few national universities, all public educational establishments are under the responsibility of one of the levels of local governments. Private education plays an important role in all levels of education and there are a range of private educational institutions that serve all socio-economic classes, particularly in larger municipalities. The public primary and secondary education system are run by the departments and by larger m~nicipalities.~~Following strict rules and salary guidelines set by the national government, they employ teachers, which are the main input in any education system. They also have a great amount of flexibility to invest ineducation quality. The central government has long encouraged local authorities to be creative in the actual delivery of services. Colombia has also beenquite opento public-private partnershipsinall levels of education. 41All municipalities with a population of 100,000 or more in 1993 census are "certified," meaning that they directly provide public primary and secondary education. 66 6.6. The financing for primary and secondary education is largely from the central government, although local governments do have the option of "topping off' the central government's subvention. Currently, around 85 percent o f public spending on primary and secondary education i s directly from the central government. Inprinciple, education transfers are on per student basis ("student-centered financing"). However, the central government makes significant transfers on the basis o f historical costs as well. While larger municipalities often make significant contributions to the education system, this i s the exception and not the rule. 6.7. All municipalities are given central resources to invest in education quality and considered the main level for improving quality. This often creates a contradiction, as many teachers are departmental employees. With the bulk o f resources committed to paying teachers' salaries, many municipalities have to devote their education resources to basic recurrent expenses such as maintenance and public services, instead o f continuous investment inquality. 6.8. Higher education is also largely a local responsibility at least in terms of administration. Most departments administer public universities and many o f the larger municipalities also operate universities. There are several national universities located around the country, with a heavy presence in the capital, Bogota. Even though the local government's role infinancing is quite small, universities are still considered to beprimarily a local responsibility. 6.9. Publictertiary education is also largely financed directly by the central government. In2000, 63 percent of fundingcame from national transfers and only 7 percent came from local governments' discretionary budget. The higher education budget i s generally transferred on the basis o f historical costs and i s done through direct transfers from the central government rather thanthrough the transfer system that is used to finance primary and secondary education, health, water, and other programs. Tuition and funding from other sources (likely to be research grants) has played an increasing role in higher education financing, accounting for 30 percent o f total spending compared to 22 percent in 1992. This is generally a healthy trend, involving more cost sharing from households that directly benefit from higher education (World Bank, 2004a). At the same time, Colombia has beena pioneer in Latin America in expanding student credit for higher education students. This has helped facilitate entrance into higher education, particularly for students from poorer backgrounds. 6.10. The National Training Service (SENA) plays a major role in non-university tertiary and technical education, but there are concerns about the efficiency and effectiveness of its training program. SENA i s financed by a wage tax, equal to 2 percent o f payrolls paid by employers. With its core financing, SENA operates a large number o f training centers and programs throughout the country. SENA provides around a quarter o f training inthe country. It targets, imperfectly, poorer segments o f society. In rural areas, SENA's courses are offered for free and they are often quite cheap or free in urban areas, even for subject matters in great demand. SENA directly offers courses but is also expected to play a greater role as an accreditation agency and working with private sector training institutions. While SENA does have links to the private sector, it is often inflexible in adjusting its training programs and i s often more expensive (on a per student basis) than private sector training providers. Despite these limitations, SENA has a generally good reputation among workers, firms, and the general public (World Bank, 2004b). 67 6.11. Increasingspendingon social services does not automaticallylead to better results; the focus should be on efficiency and quality of expenditures. Recent evidence from Colombia shows that there is a wide range of efficiency of local governments in running the health and education sectors (World Bank, 2007b). While some jurisdictions have substantially more locally raised resources available for education, there is no relationship between fiscal capacity and efficiency in delivery of services. A recently approved reform to the Constitution guarantees more resources for education, through local governments. Care i s needed to ensure that these resources will be used efficiently. The situation i s similar in higher education. The Colombian university system appears to be inefficient, with unmet demand in some areas and oversupply in other areas. It focuses primarily on undergraduate education, which i s done at the expense of graduate education and often, in the long-run, "cheats" undergraduates of a quality university education (World Bank, 2003). 6.12. Education policy plays an important role in competitiveness. As a middle income country, Colombia can no longer rely on cheap labor to produce manufactured goods and will increasingly have to look for strategies to increase the value added of its goods and services. In particular, there will have to be greater emphasis on services, both for export and for the domestic market (Farrell, et. al., 2005). 6.13. The basic education system (primary and secondary) plays a critical role in providingskills for workers. A well educatedworkforce will be more productive and probably more reliable. Inaddition, basic education can provide important basic skills, such as knowledge of foreign language and information technology that play an increasing role in the day-to-day operations of many firms. More importantly, an individual with a good basic educationcan adapt to changing employment circumstances. In a modern economy in the globalized world, this adaptability is crucial bothfor the employee and for firms. 6.14. The higher education system (universities and similar post-secondary institutes) plays a dual role in competitiveness. As educational establishments, higher education institutions play a key role intraining managers and skilled workers. Incountries like Colombia, 68 where a significant percentage of the population now finish formal education, technical institutions can play a major role inthe higher education system. Ina service economy, workers with higher education are an absolutenecessity. 6.15. Beyond the function of universities as centers of education, they are also important centers for both basic and applied research. Even in developing countries, this role can be very important as they can play a role as a bridge betweenresearch in industrial countries and local adaptation. Universities can work in partnership with private industry and provide highly skilled research support when needed. For example, universities and research centers in developing countries played a major role in spreading new agricultural technology that greatly increasedthe supply of food and reducedthe risk of famine inthe 1960sand 1970s. This built on research done in high income countries and introduced local adaptions to ensure that the new varieties would function. 6.16. The education level of the workforce plays a major role in determining a firm's productivity and its capacity to adapt new practices and technology. It i s also a good proxy ofthe skill level inthe work force. Inaddition areas with a high concentration of educatedpeople are likely to generate positive externalities; educated people attract other educated people and having a higher density tends to increase the demand and contribution of education. The education level of workers is low and varies greatly by region. Although Colombia has made an important effort to increase coverage of the education system, the education of the existing "stock" of the labor force is still quite low. Figure 6.1 shows the years of education of the labor force by Department. Figure 6.2 shows the percentage of the work force by sub-region, with university and higher education. 6.17. Both these figures show that the highest levels of education are located in Bogota and Valle, both in terms of overall education levels and the percentage of the population with professional qualifications. Antioquia and the eje cafetero (Caldas, Quindio, and Risaralda) are above the national average but lag significantly Bogota and Valle. The Atlantic coast also has some areas with high education, primarily in Barranquilla and Cartagena. Having ready access to workforce with a minimum level of education is an important In general, larger firms have more educated workers than smaller firms. 6.18. In the Central and Pacific regions only 4 percent of the work force has technical and professional qualifications, while in Atlaintico, Valle, Antioquia, and Bogotai that rate is substantially higher. This level of education represents the labor pool for firms. A region with reduced number of trained people may face impacting skill shortages, and therefore, weaker horizontal and transversal linkages among firms, critical for moving up in the value and knowledge chain and overall competitiveness. 69 Figure 6. 1: Years ofEducation of labor force, by Figure 6.2: Percentageo f work force with post- department secondary education, by sub-region 12 0.16 1 10 L- 0.14 0.12 0.10 0.08 0 06 0.04 0.02 AtlhSco Oriental Central Pacifica Bogota Antioquia Valle Source: Own elaboration based on ECV2003. Source: Own elaboration based on ECV2003. 6.19. Current trends tend to reinforce the existing stocks. Figure 6.3 shows the percentage of the population between the ages of 18 and 23 in each department that is enrolled in ost- secondary (higher) education:* Bogota still has the leading positioninterms of enrollment: but Antioquia and Santander have significant enrollment indicating that these departments will gradually increase the percentage of workforce with degrees. At the same time, Valleys enrollment is below the national average, indicating that the region is at some risk of losing its advantage in education. Although Atlhtico, with its capital of Barranquilla, has a high level of enrollment, the rest of the Atlantic coast regions perform poorly. To some degree, this represents the tendency of many students to enroll inBarranquilla butprobably also reflects atrendtowards increasing concentration inBarranquilla and, to a lesser extent, Cartagena. Also worrisome is the low enrollment rate inthe Central and Oriental sub-regions, none of whose departments is at the national average. This will tend to reinforce the low level of education achievements in the region andreduce the attractiveness of these areas for investors. 6.20. Investment in human capital have (a long lagged) impact on income per capita after long periods as illustrated by the recent dynamics of human capital formation, which does not match income per capita growth. Sub-regions with higher increases in schooling were not the fastest growing economies within Colombia between 2001 and 2005. Human capital endowments and income per capita among Colombia's sub-regions are strongly correlated (0.70 in 2005). Bogota leads all sub-regions in human capital endowments and is also the richest department. Choc6 belongs to the other extreme of the distribution with half years of schooling and one third of Bogota's income per capita. Some sub-regions have between 6 and 7 years of schooling while having substantially different levels of income per capita, which suggest explanatory variables for differences in income per capita (Figure 6.4). Atlhtico and the departmentsof the Eje Cafetero are positive outliers, while Antioquia and especially Santander are negative outliers, with levels of education below what i s expected for departments of their income levels. 42This is similar to the gross enrollment rate for tertiary education. 43Bogota is likely to be a net "importer" o f higher education students and the figures here probably overstate the tertiary enrollment rate in the district. 70 Figure 6.3: CurrentEnrollmentRates inHigherEducation I d Source: Ministerio de Educacidn Nacional, 2005. 6.21. Colombia has a relatively sophisticated system of tracking education quality in different education levels. The SABER assessment i s given every three years to students in 5th and gthgrade, in a number o f subject areas. Inthe final year o f secondary schooling, 1lth grade, most students take the State Exam (commonly called the ICFES exam), which covers a wide range o f subject areas. The central government i s gradually introducing the ECES exam for university graduates. This exam, which will become obligatory, will cover core subject matter for different degree programs. While students are not required to "pass" the exam, it can be useful information for employers and to judge the quality o f higher educational institutions. 6.22. Both the secondary and tertiary education systems have been criticized for not being aligned from the needs of the labor market. Inthe case o f secondary education, there does not appear to be good information about student's options at the tertiary level and many students do not have resources to continue with higher education. The curriculum i s also traditional with little input from the private sector. Universities often have contacts with the private sector, but this often does reflectthe informationthat students receive about employment opportunities. 71 Figure 6.4: Schooling YS.Income per capita, by department . BogotaD C Santander . ,Antioquia ,Valle MSaa Guajlra - . Atldntico BOlivzRisaralda . . .Quindio 0 N SpJt.agdal 0 .Na o Chocb I I I - 1 ~~ 5 6 7 8 9 AverageYears of Schooling 2005 Source: Based on DANE CuentasNacionales Departamentales (2001,2005) and Encuesta Continua de Hogares. 6.23. Colombia has not fared well in international quality comparisons but there are important initiatives being undertaken at the governmental level. Colombia participated in 1997 TIMSS international comparison and had the second worst performance. The best scores obtained by Colombian students in mathematics and language rank among the lowest obtained by their peers of Singapore, which was first inthe test. Bogoth, Atlhtico, Boyaca, and Santander have the highest levels of quality in education, however their performance fares more than 15 percent below the international average in the TIMSS assessment (Sarmiento, et. al., 2002). Colombia participated in the 2006 PISA evaluation, which is sponsored by the OECD and has increasingly become the international standard to benchmark education quality. SABER results confirm that around one third of students inninthhave strong analytical capacity inmathematics (defined as levels D and E). At the supply side, the central government is actively introducing mandatory accreditation of new university programs and of some existing programs to ensure that they meet a minimum level of quality. Existing programs may also apply for accreditation, butthis is voluntary inmost areas of studies. 6.2 Innovation and TechnologicalDevelopment 6.24. Colombia's innovation system has a low capacity to translate investments into innovation. It performs below par compared with similar economies at its income level. This deficiency can be largely attributed to weaknesses in all four pillars of the knowledge economy: (i) economicincentiveregime;(ii) educationsystem;(iii) innovationsystem;and(iv) the the the access to information and communication technologies. As a result of these limitations, Colombia has a very low science and technology skill base. In 1995, there were 0.3 researchers per thousand workers in Colombia compared 0.6 in Mexico, compared with 0.8 researchers in China (which increased to 1.1 researchers in 2002), 1.7 in Argentina, and an average of 5.8 researchers per thousand workers in OECD countries. Most researchers are inthe public sector, 72 with only about 25 percent of all R&Dpersonnel working inindustry in2004 versus 32 percent as the average for LAC. By contrast, in the OECD around two-third o f researchers are in the private sector. 6.25. Total R&D investment, public and private, is also far below the OECD levels: expenditures on R&D as a percentage of GDP in Colombia are about 0.6 percent as of 2006, about average for the LAC region, but below those of Mexico, Chile, and Brazil. It is also inferior to other emerging economies such as China 1.2 percent, and India 1.3 percent. All Latin American countries, including Colombia, have low levels of investment in R&D below their projected levels (Lederman and Maloney, 2002). A recent ranking made by the Economist Intelligence Unit ranked Colombia 61, while close L A C competitors are more than ten positions above (Table 6.2). Technologically successful countries such as China, Finland, Korea, Ireland, andIndiaare investingabove their projected levels. Table 6.2: International Innovation PerformanceRanking, 2002-2006 Country InnovationRanking Argentina 38 Venezuela 40 Mexico 45 CostaRica 46 Chile 47 Brazil 48 Colombia 61 Note: The innovation performance index measures innovation output or performance, and is based on international patentsdata. Source:Economist Intelligence Unit(2007). 6.26. Inthe poorest sub-regions, research and development endowments and activities are almost inexistent. Most R&D i s concentrated in Bogota and Antioquia, with Valle and Santander as minor centers. 35 percent of research and technological development centers (RTDC) are located in Bogota, follow by 25 percent in Antioquia, 11 percent in Valle, and 10 percent in Santander. As can be seen in figure 6.5, the top three sub-regions concentrate more than two thirds o f researchers. Likewise, 45 percent o f firms' investments in research and development occur inBogota. 6.27. The private sector plays only a small role in innovation activities. Private investment inR&D,both as a proportion of GDP and as a percentage oftotal nationalR&D expenditures, is very low. Firms' investment in R&D in Colombia i s about 0.25 percent of GDP compared with 0.4 percent in Brazil, 0.3 percent in Chile, and 0.8 percent in China, and 1.5 percent in the OECD. The private sector also has very little contact with public sector R&D. This implies limited private sector access to basic and applied research, reduced relevance and commercialization of R&D, and limited success in productive innovation. Likewise Colombia, as most other L A C countries have relatively few and scattered links with foreign knowledge centers. 73 Figure 6. 5: Concentration of Researchers by Department, 2004 P a Source: Colciencias. 6.28. Colombialags behindthe LAC regioninterms of use and investmentininformation and communications technologies. The average number of internet users in LAC is 110 per 1,000 in2004 while inColombia, it was 85. The number of internet user inLAC is significantly below those in successful and innovative countries, such as Chile and Korea. That number i s often as a proxy for the technological readiness and capabilities of the country to move up the technological value chain. Investment in ICT per capita shows that Colombia is investment less thanhalfof Chile inper capitaterms (Figure 6.6). 6.29. Establishing a National Quality System with minimum standards of quality is critical for access to international markets. Much as the standardization of weights and measures in the 19' century facilitated international commerce, current efforts to establish internationally recognized quality assurance system increase transparency in international trade. Without a reliable system, buyers are often reluctant to purchase local goods as they cannot easily compare to goods produced by competitors or ensure their final customers a minimum standardof quality. Colombia has made significant progress inthe last few years and inthat area ranks among the top for the LAC region (Figure 6.7). 6.30. Using the number of firms with I S 0 9000 and 9001 certification as a proxy for quality adoption, we see that Latin American countries lag other regions. Brazil is the leading country in the Region with 6,120 firms with I S 0 9000, Argentina with 4,149 firms, Colombia with 4,120 firms and Mexico with only 2,508 firms with I S 0 9000 and 265 firms with I S 0 9001. China has 75,755 firms with I S 0 9000 and 9001 and 40,997 certified firms with I S 0 9001. China's GDP is not significantly larger than that of Mexico's. This i s a key component of China's export success. Chinese firms understoodvery quickly the need to use and adopt quality standards, setting up a quite effective National Quality System. 74 Figure6.6: Per capita investment ininformation and technology, 2006 I Source: WDI. Figure 6. 7: I S 0 9000:Normalized manufacturing value-added, 2003 IS0 9000/ US$ bln de valor afiadidomanufacturer0en2003 Valorariadidomanufacturr (US$ corrientes) 2003 (Ohdel PIB) 13 IS0 9000 par US$ Mn de mior afiadido manufacturemen 2003 (US$ comentes) IValor afiadido manufacturemen 2003 (% del PIB) Source: World Development Indicators, World Bank; The I S 0 Survey o f I S 0 9001:2000 and I S 0 14001 Certificates - 2003, ISO. 6.3 Conclusions and Recommendations 6.3 1. Colombia faces certain deficiencies in human capital and innovation that need to be tackled as part of a broader sustained growth agenda. There are deficiencies inthe quality o f education as measured by international reading literacy tests. The use and investment in information and communications technologies seem also very weak comparing with LAC countries not to mention that o f the fast growing economies. The technological base seems very 75 weak, and to some extent unexploited, with limitedinvestment inR&D, even when adjusting per GDP per capita. R&D activities are highly concentrated inuniversities with limited linkages with the productive sector, therefore associated with the limited effectiveness o f R&D investments. However, Colombia does better on use and adoption o f quality and standards when compared to L A C countries but it i s still below international benchmarks o f successful countries. Strengthening the Colombia National Innovation System should be a key priority, more and better spending in supporting R&D and facilitating linkages among university and industry should be at the top o f the national agenda. 6.32. The distributionof humancapitaland R&D capacityis spreadvery unevenlyacross sub-regions. Most o f the human capital and technological endowments are concentrated in the three major productive centers- Bogota, Valle, and Antioquia. There are significant differences across Departments in highly correlated three factors: labor skills, educational attainment, and quality o f education. 6.33. There is an apparent misallocation and mismatch of resources for research and development among sub-regions, along with the prevalent lack of linkages among productive sector and knowledge centers. There seem to be a weak relationship between location o f researchers and investment in R&D, along with a weak relationship between the existence and location o f technology centers and investment in R&D. In fact, Cundinamarca, Sucre and Cauca have no technology centers, yet they are the three top Departments in terms o f investment in R&D. There i s no clear relationship between investment in R&D and GDP per capita, indicating to some extent, the validity o f the hypothesis o f the lack o f productive focus o f R&Dactivities. Santander appearsto be a balanced department, with the secondhighestGDP per capita, a fair stock o f researchers and technology centers, but less than expected R&D investments. 6.34. "Connecting the dots" or linking the productive base with the department's capacities is the first element in moving forward. The second is exploiting the existing technological capabilities at the sub-regional level, and strategically developing missing ones to match productive needs and structure, after proper evaluations. In terms o f endowments in the area o f human capital and innovation, the departments can be classified into five groups, form better to worse, as shown in Table 6.3. Assistance and intervention should be tailored to the conditions o f each group. For group A, the strategy would be to build on strengths, eliminating bottlenecks in logistics, increase integration o f value chains, and increase efficiency o f technological capabilities and resources. For group B (Atlhtico, Cundinamarca, Caldas Cauca and Magdalena) along with Santander, there appears to be some potential, in terms of unexploited capabilities andtrends inR&D. This can be supported by targetedprograms 6.35. In moving forward, Colombia should seek to increase its productivity through knowledge-basedintegrationinto globalvalue chains and improvementsin human capital and technologicaldevelopment. Several steps needto be taken to make progress in improving humancapital and inthe buildingo fa coherent innovation system with a more active role for the private sector and stronger academic-industry linkages. The quantity and quality o f education needs to be significantly upgraded and complemented with demand and incentive driventraining programs. Efforts should focus on defining and sustaining a long-run Science, Technology and 76 Innovation and Quality strategy that is translated into multi-annual budgets that clearly identify national priorities aimed at increasing competitiveness and productivity. Specific recommendationsinclude: 0 The central government should provide more resources for education quality for basic education, ensuring that marginal resources be provided to both municipalities and departments. At the same time, through both incentives and regulations, efforts should be made to ensure that these resourcesactually are usedfor efforts to increase education quality. Focus on the demand side of higher education finance, given the general oversupply of seats in universities and other higher education institutions. Colombia has been a pioneer in making student loans and this effort should continue and intensify. 0 Continue efforts to reform SENA, including initiatives to increase the accreditation and supervision role of agency while encouragingmore public-private partnerships intraining. Public finance for SENA should increasing focus on subsidizing poorer workers. 0 Continue with efforts to modify and improve education curricula, aimed at ensuring its relevance for productive activities. Colombia has had a long tradition of public-private partnership in education and this can continue with more private support in developing and delivering education models at all levels. 6.36. Increasingand strengtheningfunding mechanisms and incentives for the National Innovation and Quality Systems across all areas and levels of government is also important.Centralized leadership, increasedand consolidated funding, and setting up highlevel incentive-driven programs-such as matching grants-are key components for an effective reform program. Demand support programs need to be complemented by strengthening the regional infrastructure that provides technological services. Interactions between R&D institutions and industry can also be facilitated through incentive driven funding. Consortia programs can be viewed as entry points for such interactions. To make science more relevant to industry requires providing more funding to enhance such linkages (support programs focusing explicitly on this objective) but also a deep reform to enhance incentives of research organizations to cooperate with industry.Finally, to secure effective knowledge transfer, there is a need to support internationalization policies, by financing activities that strengthen the integration of the national innovation system into the global system and by engaging successful expatriates abroad inan international knowledge network. Particular recommendations include: Continue to provide support for basic public goods in innovation, such as strengthening the patent system andimproved quality assurance systems for private enterprises. Focus government support on basic researchthat has a highpayoffto the entire economy, while encouraging universities and other research institutes to work with the private sector. Competitive grants can play an important role inensuring adequate basic research. 77 c I I I I I P b CHAPTER7 - CONCLUSIONS 7.1. Policies for sub regional competitiveness are only one component o f sub-regional development, which encompasses also policies aimed at alleviating poverty and social distress. Competitiveness policies aim at unleashing the forces that drive productivity and are thus focused on long rungrowth. 7.2. Notwithstanding its merits, the traditional definition o f convergence should not necessarily be seen as the sole objective o f sub-regional competitiveness policies. The convergence perspective may imply an excessive orientation toward supporting lagging sub- regions whose productive base is reduced compared to the leading sub-regions. It may also limit needed resources for most competitive sub-regions and, therefore, hurt the overall competitiveness o f the country. Pro-competitiveness policies should encompass both lagging and leading sub-regions through the pursuit o f different goals based on the different types o f sub- regions. As an illustration, the fastest way to achieve convergence would be to allow leading sub-regions to wait for the lagging ones to catch-up, but this policy would damage overall competitiveness. Instead, a proactive convergence would allow for an increasing gap between sub-regions as long as the lagging sub-regions improve at a higher pace than their historical trends. 7.3. Sub-regions were grouped according to endowments and economic performance, on one hand, and knowledge and innovation indicators, on the other. Leading regions are positioned as top world class business locations. This group includes strong competitive metropolitan areas such as Cali, Medellin and Bogota. Bolivar, Boyaca and the coffee region are also world class, although highly-specialized sub-regions. In addition, Atlhtico and Santander are in between these two groups. The third group, called "poles for local development", is formed by 8 sub- regions, which present an intermediate level o f economic development and knowledge indicators but lack important instruments to compete and grow. Finally, the last group lacks most of the instruments for growth and competitiveness, requiring special attention. 7.4. Consistently with their economic development, each group has a different institutional strength. Most o f the policy lessons stemming from international best practices have already been implemented inthe leading sub-regions o f Colombia. Leading regions can therefore be an important source o f learning, and the strategy o f the central government, when establishing sub- regional policies, should benefit from these experiences. 7.5. Additionally, a great opportunity to strengthen institutional capacity in the other three groups is through coaching the leading sub-regions. As part o f this study, two workshops were held in Bogota and Barranquilla with representatives from twelve different sub-regions. The conclusion o f the workshops was that it would be worthwhile to implement a competitiveness strategy in the most competitive sub-sector o f each sub-region, with the support o f the leading sub-regions and the national public and private entities. Lessons learned by local actors, from a sub-sector selected by them, would improve their institutional capacity to continue working with their other opportunities. A preliminary model o f intervention, drafted from Bogota's recent experience, can be developed by answering four questions. 79 Figure7. 1: Four questions for sub-regionalcompetitiveness strategies 1. Identlficationof pZ->Leadenhipwlth adivitiesandthdr Implementationplan PartialResults forfollowupand Monitoring S`ource: Ownelaboration tlasedon two "Workshops for Sub1-regionalCompetitiv`eneISS in Colombia". 7.6. A word o f caution is required regarding the impact of violence on the location of economic activities. Despite the fact that the overall objective o f the study relates to all Colombia's sub-regions, higher private sector involvement in some sub-regions may be difficult to achieve under lack o f good security conditions. However, it i s difficult to assess to what extent relatively low frequency variables, like homicides rates, may affect the location decisions of firms in specific business districts. 7.7. It is crucial, though, to recognize the value of improved security conditions on the investment climate. For instance, inChapter 5 a survey o f entrepreneurs o f the main four cities o f Colombia shows a strong negative effect o f losses associated with crime (thefts, burglary, etc) on productivity. Yet other variables like red tape and corruption have also substantial negative effects inthe same model. 7.8. On the other hand, the assumption o f security as a prerequisite for investments, location o f firms and competitiveness although valid in general may need to be analyzed differently in a complex security context such as Colombia. Operating in a secure environment may be a business objective and not just an expected previous condition. There are examples o f industries that became successful, among other factors, by dealing directly with questions o f poverty and inequality in the surrounding environment as part o f their business logic. They show that a particular way o f doing business may provide better security than traditional security measures usedby other firms and businesses. 7.9. Overall the investment climate in Colombia i s being influenced positively by recent trends in security conditions. The effects on specific sub-regionsfrom national improvements in security are difficult to identify, as many other local attributes enter into action to attract specific firms. In certain sub-regions o f Colombia the package for improved competitiveness should include security along with other variables like infrastructure, human capital and financing, among others. For a complete analysis, see World Bank ( 2 0 0 6 ~2007d). ~ 7.10. We revised the recent literature on the subject, focusing on the main instruments that have been applied as well as on some international experience related to the field o f regional growth-oriented policies. The most successful experiences seem to be based on a combination o f 80 instruments, tailored to face the specific needs while leveraging the endowments o f the different regions. The institutional capacity o f each region must be carefully assessed; achieving institutional efficiency may demand assembling certain interventions at a higher level o f aggregation, a macro region, for instance. 7.11. The lesson o f different policies for different sub-regions i s further reinforced by an econometric exercise based on the Investment Climate Survey carried out in 2006 in Colombia. The results clearly showed that, even restricted to a comparison among four o f the leading regiondcities o f the country, there are significant differences in terms o f productivity determinants. While there are clearly common elements, such as the impact o f informality on productivity, there are also specificities o f each region that necessarily lead to different emphasis interms o fpolicy recommendations. 7.12. At the national level, Colombia faces low total factor productivity growth, particularly since 1990s. Investment in infrastructure, measures to reduce informality, corruption and crime and improvements intechnology and skills are the main elements o f the agenda. An econometric analysis o f the determinants o f productivity at an aggregate level for the country shows that variables related to red tape, corruption and crime and to infrastructure present the highest relative impacts on productivity. The impact o f variables linked to quality, innovation and labor seems a bit lower, but the results have the expected sign and are thus important determinants o f productivity as well. 7.13. The bottom-up process, known as the Agenda Interna, was a useful instrument for the identification o f the strengths and weaknesses o f the regions. Some proposals (apuestas) were defined that should be implemented so that the regions can progress in the promotion o f their own competitiveness. The regions are anxious to take the next steps o f competitiveness promotion, and there are no clear answers for them, neither on the theoretical level nor based on country case experience. One suggestion could be to develop a project to accompany each region inthe development o f an apuesta productiva. The ultimate objective would be to develop local institutional capacity that could be redirected to other initiatives. 7.14. An econometric exercise presented in Chapter 3 shows that the impact of the trade liberalization process can indeed be a very important driver o f firms' location in Colombia, and that the evolution o f firms' location has to be carefully followed by the central government and the Competitiveness Commission. One suggestion is to create an observatory that could be linked to this commission. The design o f competitiveness policies for sub-regions in Colombia has to take into account not only the observed differentials inincome or productivity, but also the expected movements that will take place due to the change o f relative prices associated with the signing o f a free trade agreement with the US and other countries. Anecdotal information from the meetings held in Barranquilla and Bogota during the preparation of this report tends to reinforce this view, as does the preliminary but useful information collected in a value-chain exercise for fruit-horticulture and glass-ceramics. 7.15. Interms of policy instruments, infrastructure and logistics can play a fundamental role. The existing analysis agrees that most sectors in Colombia are self-sufficient at a regional scale, with little interregional interaction, and most forward and backward linkages are concentrated in 81 the richest regions. These results suggest that regional polarization may be perpetuated. Transport logistics is clearly one of the major factors constraining spatial interaction, which is not only due to infrastructure networks (i.e,, roads), lack of coverage or inadequate standardsand capacity, but this constraint i s also due to a set of diverse and interacting features, particularly transport regulatory issues, private sector (shippers and carriers) business logistics development, and government managedprocesses (mainly customs and border procedures). 7.16. The increasingly open trade scenario brings opportunities to redefine spatial backward and forward linkages, as firms' strategies frequently consider the relocation of production and distribution facilities in order to take full advantage of the new situation. The production chain analysis made (at a preliminary level) in two productive chains - fruit-horticulture and glass- ceramics - shows that firms wants to continue utilizing their current facilities, located according to raw material sources and domestic markets, in order to serve internal demand. These firms plan to expand their capacity in order to serve export markets located in new facilities on the Atlantic coast, particularly inthe Barranquilla area. 7.17. The materialization of plans such as the following may increase interregional links: new agricultural products, originating from diverse regions and traveling to more scattered manufacturing plants, flows of raw or semi-processed material from inner regions to plants located in port areas, development of logistics platforms on the coast that supply the inner domestic market. Government policies may help to facilitate these new potential flows and may include enhancing infrastructure networks, but the analysis confirms previous findings that special attention should be placed on the perfonnance of the trucking industry, on the organization and management of ports, and on the inspectionprocesses at gateways. 7.18. The strengthening of both higher education and the innovation system should be at the center o f sub-regional competitiveness policies. Most of the human capital-technological endowments are concentrated in the three major productive centers, Bogota, Valle and Antioquia. There is no relationship between investment in R&D and GDP per capita, indicating to some extent, the validity of the hypothesis of the lack of productive focus of R&D activities. MainrecommendationspresentedinChapter 6 ofthis report are: 0 Colombia should seek to increase its productivity through knowledge-based integration into global value chains and improvements inhuman capital and technological development. "Connecting the dots" or linkingthe productive base with the department's capacities is a key element in moving forward. Strengthening the relationship between universities and the private sector could help to make sure that the available resources respond to the needs of the productive sector. 0 Increasing and strengthening funding mechanisms and incentives for the National Innovation and Quality Systems across all areas and levels of government is also important. 82 REFERENCES Aghion, P. and P. Howitt (2003) Endogenous Growth Theory. Cambridge, MA: MIT Press. Aroca, P., M. Bosch and W.F. Maloney (2005) "Spatial Dimensions of Trade Liberalization and Economic Convergence: Mexico 1985-2002." World Bank Policy Research Working PaperNo. 3744. Baron, J.D (2003) "QuC Sucedi6 con las Disparidades Econ6micas Regionales en Colombia entre 1980y el 2000?" Documentos de Trabajo sobre Economia Regional, No. 38. Banco de laRepublica, Cartagenade Indias. Baron, J.D. and A.M Roca (2003) "La Descentralizaci6n y las Disparidades Econ6micas Regionales en Colombia en la DCcada de 1990." Documentos de Trabajo sobre Economia Regional, No. 36. Banco de la Republica, Cartagenade Indias. Baron J and A. Meisel, A. (2003) "La Descentralizacion y las Disparidades Econdmicas Regionales en Colombia en la DCcadade 1990." Coyuntura Econdmica, Vol. XXXIII (2), September. Barro, R. and X. Sala-i-Martin (1999) Economic Growth.Cambridge, MA: MIT Press. Bastos, F. and J. Nasir (2004). "Productivity and the Investment Climate: What Matters Most?." Working Paper. Mimeo. Basu, K. (1997) Analytical Development Economics. Cambridge, MA: MIT Press. Bayer, P. and C. Timmins (2007) "Estimating EquilibriumModels of Sorting Across Locations." TheEconomic Journal 117(518), pp. 353-374. Belli, P. and 0. Appaix (2003). "The Economic Benefit of Investing in Child Health." HNP Discussion Paper. Birchenall, J. and G. Murcia (1997) "Convergencia Regional: Una Revisi6n del Caso Colombiano." Desarrolloy SociedadNo. 40, September. Bloom, E., V. de Wit, and M.J. Caarangal-San Jose (2006) "Potential Economic Impact of an Avian FluPandemic on Asia." Asian Development Bank ERD Policy BriefNo. 42. Bonet, J. (2003) "Colombian Regions: Competitive or Complementary?" Revista de Economia del Rosario 6(I), pp. 53-70. Bonet, J. (2005) Regional Structural Changes in Colombia: An Input-Output Approach. Mimeo. Bonet, J. (2006) "Desequilibrios Regionales en la Politica de Descentralizacibn en Colombia." Documentos de Trabajo sobre Economia Regional, No. 77. Banco de la Republica, Cartagenade Indias. Bonet, J.and A. Meisel (1999) "La Convergencia Regional en Colombia: Una Visi6n de Largo Plazo, 1926-1995." Documentos de Trabajo sobre Economia Regional, No. 8, Banco de la Republica, Cartagenade Indias. Caballero, R; K.N. Cowan; E. Engel; and A. Micco (2004) "Effective Labor Regulation and Microeconomic Flexibility." NBERWorking Paper 10744. Caballero, R.; E. Engel; and A. Micco (2004) "Microeconomic Flexibility in Latin America." NBERWorking Paper 10398. CAF - Corporaci6n k d $ a de Foment0 (2006). Camino a la Transformacidn Productiva en Ame`ricaLatina. Cardenas, M.and A. Escobar (1995). "Infraestructura y Crecimiento Departamental 1950-1994." Planeacidny Desarrollo 26(4), October-December. Cardenas, M., A. Gaviria and M. Melendez (2005) "La Infraestructura de Transporte en Colombia." Fedesarrollo. 83 Cardenas, M., A. Ponton, and J.P. Trujillo (1993) "Convergencia y Migraciones Interdepartamentalesen Colombia: 1959-1989." Coyuntura Econbmica 23(l), pp. 111- 137. Carvalho, A., S.V. La11and C. Timmins (2006) "Regional Subsidies and IndustrialProspects of Lagging Regions." World Bank Policy ResearchWorking PaperNo. 3843. Dasgupta, P. (1993) An Inquiry into Well-Being and Destitution. Oxford: Oxford University Press. De la Fuente, A. (2004) "Second-Best Redistribution through Public Investment: A Characterization, an Empirical Test and an Application to the Case of Spain." Regional Science and Urban Economics 34, pp. 489-503. Dendrinos, D. and M. Sonis (1988) "Nonliniear Discrete Relative Population Dynamics of the US.Regions." AppliedMathematics and Computation 25, pp.265-285. Dendrinos, D.and M.Sonis (1990) Chaos and Socio-Spatial Dynamics. Springer-Verlag. DNP - Departamento Nacional de Planeacih (2006) Cadenas Productivas, Estructura, Comercio Internacional y Proteccibn. Dollar, D., and E.N. Wolf (1997). "Convergence of Industry Labor Productivity among Advanced Economies, 1963-1982." InE.N. Wolff (ed.), The Economics of Productivity. UnitedKingdom: Elgar. Dollar, D., M. Hallward-Driemeier, and T. Mengistae (2003) "Investment Climate and Firm Performance in Developing Economies." London School of Economics STICERD Working Paper. Dollar, D., S. Wang, L.C. Xu, and A. Shi (2004) "Improving City Competitiveness through the Investment Climate: Ranking 23 Chinese Cities." World Bank Development Research - Group. Mimeo. Dutz, M. (2005) "Road Freight Logistics, Competition and Innovation: Downstream Benefits and Policy Implications." World Bank Policy ResearchWorking PaperNo. 3768. Economist Intelligence Unit (2007) Innovation: Transforming the Way Business Creates. May 2007. Escribano, A. and J.L. Guasch (2005) "Assessing the Impact of the Investment Climate on Productivity Using Firm-Level Data: Methodology and the Cases of Guatemala, Honduras, andNicaragua." World Bank Policy ResearchWorking Paper 3621. Escribano, A,, J.L. Guasch and J. Pena (2007) "Colombian Metropolitan Areas: Investment Climate Assessment on Product." Mimeo. Fan, E. (2003) "SARS: Economic Impact and Implications." Asian Development Bank ERD Policy BriefNo. 15. Farrell, D., A. Puron, and J. Remes (2005). "Beyond Cheap Labor: Lessons for Developing Economies." McKinsey Quarterly.2005, No. 1, pp. 99-109. Fujita, M. and P. Krugman (2004) "The New Economic Geography: Past, Present and Future." Papers in Regional Science 83, pp. 139-164. Galvis Aponte, L. (2001) "La Topografia Econbmica de Colombia." Documentos de Trabajo sobre Economia Regional, No. 22. Banco de la Repfibha, Cartagenade Indias. Hall, R.E. (1991) "Invariance Properties of Solow's Productivity Residual." NBER Working Papers, Paper No. 3034. Hall, R.E. and C.I. Jones (1999) "Why Do Some Countries Produce so MuchMore Output per Worker tha nOthers." The Quarterly Journal of Economics 114(1), pp. 83-116. 84 Hallward-Driemeier, M.,S. Wallsten and L.C. Xu (2003) "The Investment Climate andthe Firm: Firm-Level Evidence from China." World Bank Policy Research Working Paper. No. 3003. Hanushek, E. and L. Wobmann (2007) Education Quality and Economic Growth Washington, DC: World Bank. Hewings, G.J.D., M. Sonis, F.A. Cuello, and F. Mansouri (1996). "The Role of Regional Interaction in Regional Growth: Competition and Complementarity in the US Regional System." Australasian Journal of Regional Studies 2, pp. 133-149. Hon, V. and P. Fallon (2002) "Regional Development Policies: Theory and a Review of the Evidence." InWorld Bank (2002) "Brazil: Maranhao State Economic Memorandum." Kruskal, W. (1987) "Relative Importance by Averaging Over Orderings." The American Statistician 41, pp. 6-10. Lall, S.V., R. Funderburg, and T. Yepes (2004) "Location, Concentration, and Performance of Economic Activity inBrazil." World BankPolicy ResearchWorking PaperNo. 3268. Lederman, D. and W. Maloney (2003) "Research and Development (R&D) and Development." The World BankPolicy ResearchWorking PaperNo. 3024. Levinsohn, J. and A. Petrin (2003) "Estimating ProductionFunctions Using Inputsto Control for Unobservables." Review of Economic Studies 70(2), pp. 317-342. Loayza, N.,P. Fajnzylber and C. Calder6n (2004) "Economic Growth in Latin America and the Caribbean: Stylized Facts, Explanations, and Forecasts." Working Papers Central Bank ' of Chile No. 265, Central Bank of Chile. Lucas, R.E. (1988) "On the Mechanics of Economic Development." Journal of Monetary Economics 22, pp. 3-42. Meisel, A. (1999) "Dutch Diseaseand BananaExports inthe Colombian Caribbean, 1910-1950." Presented at the Meeting of the Latin American Cliometric Society (LACLIO) in Cartagena, Colombia, August 27-28. Meisel, A. (2006) "Polarizacion del Ingreso per Capita Departmental en Colombia, 1975-2000." Documentos de Trabajo sobre Economia Regional, No. 76, Banco de la Repfibha, Cartagenade Indias. Neary, P. (2001) "Of Hype and Hyperbolas: Introducing the New Economic Geography." Journal of Economic Literature, June, pp. 536-561. Nooteboom, B. (2006) "Trust and Innovation." Essay written for the Dutch Ministry o f Economic Affairs as background to the 2006 Innovation Lecture on trust and innovation. Mimeo. North, D.C. (1984) "Transaction Costs, Institutions, and Economic History." Journal of Institutional and Theoretical Economics 140, March 1984, pp. 7-17. (1991) "Institutions." Journal of Economic Perspective 5( l), pp.97-112. OECD (2007) Linking Regions and Central Governments: Contractsfor Regional Development. Paris, France:OECD. Olley, G.S and A. Pakes (1996) "The Dynamics of Productivity in the Telecommunications EquipmentIndustry." Econometrica 64(6), pp. 1263-97. Ottaviano, G.I.P. and J. Thisse (2004) "Agglomeration and Economic Geography." In: J.V. Henderson and J. Thisse (eds), Handbook of Regional and Urban Economics, Vol. 4, pp. 2564-2608. Amsterdam: North-Holland. Quah, D. (1993a) "Galton's Fallacy and Tests of the Convergence Hypothesis." LSE Economics DepartmentWorking Paper, December. 85 (1993b) "Empirical Cross-section Dynamics in Economic Growth."European Economic Review 37, pp. 426-434. (1996) "Empirics for Economic Growth and Convergence." European Economic Review 40, pp. 1353-1375. (1997) "Regional Cohesion from Local Isolated Actions: 11. Conditioning." Centrefor Economic Performance, DiscussionPaper No. 379. Perez, G.V. (2005) "La Infraestructura del Transporte Vial y la Movilizaci6n de Carga en Colombia." Documentos de Trabajo sobre Economia Regional, No. 64. Banco de la Republica, Cartagenade Indias.. Roberts, Mand Zhang, M.(2007) What Makes Cities Competitive? A Selective Reviewof Theories and International Experiences, The World Bank, preocessed. Roca, A.M. and J. Romero (2007) "Igualdad de Oportunidades para Todas las Regiones." Documentos de Trabajo sobre Economia Regional, No. 83. Banco de la Republica, Cartagenade Indias. Rocha, R. and A. Vivas (1998) "Crecimiento Regional en Colombia: LPersiste la Desigualdad?" Revista de Economia del Rosario 1, pp. 67-108. Rodriguez, L. and M. Humberto M. (2004) Asi se hizo Chavimochic. Trujillo, Peru: Ediciones Carolina. Rueda, L.A. (2004) "Gasto Pfiblico y Convergencia Regional en Colombia." Ensayos sobre Politica Econbmica, Revista ESPE 45, Junio 2004, pp. 222-268. Sarmiento A., L. Tovar and C. Alam (2002) "Situacion de la Education Bhsica, Media y Superior en Colombia." Document produced for Educacibn, Compromiso de Todospor Corpoeducacibn. Schultz, T. (1961) "Investment inHumanCapital." American Economic Review 51, pp.1 17. - Sepic, D. (2005) "The Regional Competitiveness: Some Notions." Russian-European Centre for Economic Policy (RECEP), September2005. Soto, J.A. (1998) "Crecimiento y Convergencia Departamental: Una Aproximacibn de Panel a1 Caso Colombiano 1960-1995." Master's thesis inEconomics, Universidad de 10s Andes. Wilmsmeier, G. (2006) "Evoluci6n de las reformas en 10s puertos de Argentina, Chile y M6xico." Mimeo. U.S.Department ofAgriculture (2005) WorldHorticultural Trade & U.S. Export Opportunities: WorldAsparagus Situation & Outlook. Veeramani, C. and B. Goldar (2004) "Investment Climate and Total Factor Productivity in Manufacturing: Analysis of IndianStates.'' ICRIER Working Paperno. 127. Visser, E. (2004), "A Chilean Wine Cluster? Governance and Upgrading in the Phase o f Internationalization." Serie Desarrollo Productivo, No. 165.UnitedNations ECLAC. Williamson, 0.(1985) TheEconomic Institutions of Capitalism. NewYork: Free Press. (1991) "Economic Institutions: Spontaneous and Intentional Governance." Journal of Law, Economics, and Organization 7 (Special Issue), pp. 159-187. World Bank (2004a) "Tertiary Education in Colombia: Paving the Way for Reform." Washington, DC: World Bank. (2004b) "Public Training ReformIssues inColombia." (2004~)"Colombia: RecentEconomic Developments inInfrastructure (REDI): Balancing Social and Productive Needs for Infrastructure." Report No. 30379-CO. (2005) "Comparative Regional Development Initiatives: Lessons for Russia." Mimeo. 86 (2006a) "Infraestructura Logistica y de Calidad para la Competitividad de Colombia." ReporteNo. 35061-CO. (2006b) "Argentina: El Desafio de Reducir 10s Costos Logisticos ante el Crecimiento del Comercio Exterior." Informe No. 36606-AR. (2007a) "Brazil: S5o Paulo: Inputs for a Sustainable Competitive City Strategy." ReportNo. 37324-BR. (2007b) "Colombia Decentralization: Options and Incentives for Improving Efficiency." Mimeo. (2007~)"Peace Programmatic I:Reinsertion of Ex-Combatants in Colombia." Report No. 39222-CO. (2007d) "Peace Programmatic 11: Reparation for Especially Vulnerable Victims of the Armed Conflict inColombia." Mimeo. (2007e) "Economic Growth in Latin America and the Caribbean: A Microeconomic Perspective." Mimeo. (20079 "Measuring Global Connections." Mimeo. 87 88 ANNEX I:ECONOMETRICSUSINGINVESTMENTCLIMATE DATA44 1.Data Produced by the World Bank Group, the investment climate surveys (ICs) series of private enterprises i s an initiative which scope i s to explore the difficulties that the firms located in developing countries encounter in starting and running a business. More precisely, the surveys capture firms' experience in a range of areas related with the economic performance: financing, governance, corruption, crime, regulation, tax policy, labor relations, conflict resolution, infrastructure, supply and marketing, quality, technology, andtraining, among others. The investment climate factors were grouped in five categories to evaluate the impact of each group on the regional aggregate productivity. Inthe first group, infrastructure, are included all the variables related with customs clearance, power and water supply, telecommunications (including phone connection and information technologies) and transportation. In the second group, red tape, corruption and crime, are included all the IC factors regarding tax rates, conflicts resolution, crime, bureaucracy, informalities, corruption and regulations. The next group is finance and corporate governance, which includes factors related to governance, investments, informality inreporting of payments of sales andpurchasesfor tax purposes, access and cost of finance, and accounting (or auditing). The group quality, innovation and labor skills includes environmental regulations, quality certification, technology use, product and process innovation, research and development, quality of the workforce, and training. The last group, other control variables, is not a group of IC factors but a group of other firms' control characteristics - this group includes all the factors that are considered to have an important impact on the economic performance but are not IC factors, such as exports and imports, age, FDI, number of competitors, and firm size. The grouping of variables include 134 variables (31 infrastructure variables, 23 red tape, corruption and crime variables, 36finance and corporate governance variables, 23 variables of the quality, innovation and labor skills group, and 19 variables inthe other control variables The ICs also provides information on the productivity variables, says, output (sales), employment, intermediate materials, capital stock and labor cost. The ICs does not provide information on prices at the firm level, so the production function variables were deflated by using the ProducerPrice Indexes (PPI) of the Bank ofthe Republic of Colombia, base 2002. The Colombian ICs concentrates in a wide range of establishments mainly inthe food, apparel, chemical, textile, retail, and information technologies (IT) sectors. It covered 1,000 establishments in four metropolitan areas - Bogota (398 firms), Medellin(286 firms), Cali (172 firms) and Barranquilla (144 firms). The focus of this productivity exercise i s the manufacturing sector. By classifying establishments by their ISIC code, it used 686 manufacturing firms from 44 Based on the background paper Escribano, A., J.L. Guasch and J. Pena (2007) "Colombian Metropolitan Areas: InvestmentClimate Assessment on Product." Mimeo. 45 For a complete list of I C and control (C) variables as well as a description on how each one is measured see Escribano, et a1(2007). 89 the following sectors: (a) food, (b) apparel, (c) textile, (d) metal, machinery and equipment, (e) chemical and plastic, (0 non-metallic products, (g) wood and furniture, and (h) other manufacturing firms. Cleaning of data and regional classification The IC data is insome aspects troublesome data. Of the 686 manufacturing firms, only 338 were included in the analysis because these were the ones that had all the information needed to compute the productivity at the firm-level. There were 243 firms with at least one of the productivity variables missing and other 115 firms with outlier observations (outliers were observations with ratios of materials to sales and/or labor cost to sales greater than one). Using only 338 firms in the analysis implies losing representativenessand efficiency inthe regression analysis. In order to avoid this problem, a cleaning process that allows us to use in the analysis 570 establishments was developed. In this process, firms with all production function (or productivity) variables - sales, materials, capital stock and labor cost - were excluded. Outliers were converted to "missing" and the following steps were taken: (a) the missing values were replaced by the corresponding industry-region-size median of the variable, (b) if there is enough observations in some cells, they are replaced by the corresponding industry-size medians, (c) if still there i s not enough observations in those cells, the missing values are replaced by the region-size medians, and (d) if still necessary, the medians are computed only by size and/or by industryto replacethe missing values. The Table A.l summarizes the number of observations available before and after the cleaning process. It shows the distribution of the observations by industry and region, and by size and region respectively, before and after the cleaning process. In fact, the cleaning process does not alter the original representativeness of the data. After the cleaning process, there were 223 observations for Bogota, 104 for Cali, 240 for Medellin and 51 for Barranquilla. For the regional analysis, based on the classification in the number of observations available for each metropolitan area, three regions were considered: (a) Bogota, *b) Medellin, and (c) Cali and Barranquilla. Bogota and Medellin had enough observations to run independent regressions for each one of them. For Cali and Barranquilla, all observations were pooled so that the analysis could benefit from the law of the large numbers.To introduce some heterogeneity inthe result of the pool of Cali and Barranquilla, parameters of some IC and C variables were allowed to vary by regionthrough the use of interaction terms for some variables andregional dummies. Productivity (P) or multifactor productivity refers to the effects of any variable different from inputs - labor (L), intermediate materials (M), and capital (K) - affecting the production (Y) process. Since there is no single salient measure of productivity (or logpi), any empirical evaluation on the productivity impact of IC variables might critically depend on the way productivity is measured. Therefore, to get reliable empirical elasticities for policy analysis, Escribano and Guasch (2005) suggest searching for robust empirical results using several productivity measures. This is the approachadoptedinthis exercise. 90 Table A.l: Representativenessof production function variables before and after cleaning missing values and outliers Textiles Machinery & Nonmetallic Beforecleaning 1 0.15 0 0.00 1 0.15 0 0.00 2 0.29 products After cleaning 0 0.00 0 0.00 1 0.18 0 0.00 1 0.18 Wood and furniture manufacturing Small Medium Large Total Industry #Obs. Perc. #Obs. Perc. #Obs. Perc. #Obs. Perc. Bogota Beforecleaning 177 25.80 99 14.43 9 1.31 285 41.55 After cleaning 139 24.39 78 13.68 6 1.05 223 I 39.12 Cali Beforecleaning 65 9.48 38 5.54 1 0.15 104 15.16 After cleaninr! II 49 II 8.60 II 33 II 5.79 II 1 II 0.18 II 83 I 14.56 Medellin Beforecleaning 102 14.87 90 13.12 48 7.00 240 34.99 After cleaning 89 15.61 79 13.86 45 7.89 213 37.37 Barranquilla Beforecleaning 33 4.81 24 3.50 0 0.00 57 8.31 After cleaning 30 5.26 21 3.68 0 0.00 51 I 8.95 Total Beforecleaning 377 54.96 251 58 8.45 686 100.00 I I 36.59 After cleaning 1I 307 II 53.86 II 211 37.02 II 52 II 9.12 II 570 I 100.00 91 2. Estimation For policy recommendations, elasticities, or semi-elasticities of IC variables on productivity need to be robust (i-e., equal signs and of similar magnitudes) to the six productivity measures used. The alternative productivity measures used in this exercise, summarized in Table A.2,46 come from considering: (a) Different functional forms ofthe production functions (Cobb-Douglas and Translog); (b) Different set of assumptions (technology and market conditions) to get consistent estimators based on Solow`s residuals or OLS; (c) Different levels of aggregation inmeasuring input-output elasticities (at the industry level or at the aggregate country level). ' 1. Two Step 1.1Solow's Residual 1.1a RestrictedCoef. 2 (Pi,) measures Estimation 1.2b Unrestricted Coef. 2 (IC) elasticities 2. Single Step 2.1 Cobb-Douglas 2.l a RestrictedCoef. 2 (PiJ measures Estimation 2.2b Unrestricted Coef. 2 (IC) elasticities 2.2 Translog 3. l a RestrictedCoef. 2 (Pit) measures 3.2b Unrestricted Coef. 2 (IC) elasticities Total 6 (Pit) measures 6 (IC) elasticities The two steps estimation starts from the non-parametric approachbasedon cost-shares from Hall (1991) to obtain the Solow's residuals in logs under two different assumptions. First, cost shares are constant for all firms located in the same region (restricted Solow residual). Second, cost shares vary among industries inthe same region (unrestricted by industry Solow residual). Once the two productivity measures (logpi) in the first step were estimated, the IC elasticities and semi-elasticities from equation (A) inthe second step can be estimated for each one of the three regions: log 4 = IC, +C X ~ C, + Dj aP+ui + where IC and C are respectively the vectors of investment climate variables andcontrol variables listed in the Table A.1. In all the cross-section regressions, several sector-industry (Dj, j = 1, 2, ...,qD) dummyvariables were usedas well as aconstantterm (inter~ept)~~. 46The econometric methodology is describedinAnnex I. 47 In particular, four dummy variables for five sectors were included, using "Food" as the reference sector. "Chemicals and plastics," "Wood and furniture" and "Machinery and equipment" were grouped in the "Other manufacturing group" due to the small number of observations available for these groups. 92 The advantage of the Solow residuals i s that they require neither inputs (L, MyK) to be exogenous nor the input-output elasticities to be constant or homogeneous (Escribano and Guasch, 2005).Their drawback is that they requires having constant returns to scale (CRS) and at least competitive input markets. In the single step estimation approach, the parametric estimation was considered by ordinary least squares (OLS) of an extended production function in which to address the well-known problem of endogeneity of inputs, the approach proposed by Escribano and Guasch (2005) was used. That is, the usually unobservedfirm-specific fixed effects (whichare the main cause of the endogeneity of the inputs) were proxied by a long list of firm-specific observed fixed effects coming from the IC information. Controlling for this large set of IC and C variables, it i s possible to get, under standard regularity conditions, consistent and unbiased least squares estimators of the parameters of the production function. In particular, two different functional forms of the production function were used (Cobb-Douglas and Tanslog) under two different assumptions on the input-output elasticities: equal input-output elasticities in each region (restricted case) and different input-output elasticities by industry and region (unrestricted case). Another econometric problem to be considered in the estimations of the elasticites and semi- elasticities of IC and C variables is the possible endogeneity of some of these variables. In the productivity equations, the traditional instrumental variable (IV) approach is difficult to implement, given there is information for only one year and, therefore, the natural instruments for the inputs, like those provided by their own lags, cannot be used. As an alternative correction for the endogeneity of the IC variables, the region-industry-size average at the firm-level IC variables (E)was usedinstead of the crude IC variables, which is a common solution inpanel data studies at the firm The endogeneityof the IC variables is a topic that has beendealt with in the recent literature on investment climate. Veeramani and Goldar (2004) estimate the impact of several IC indicators on TFP variable by variable using the industry-location averages as instruments to avoid the endogeneity problem. In the same line, Hallward, et al. (2003) to avoid multicollinearity problems due to the correlation among the IC indicators propose the use of industry-region averages in models with a reducednumber of explanatory variable^.^'. While this approach avoids problems of multicollinearity, it may introduce the omitted variables bias. At its has been pointed out, in this exercise the list of IC factors works as a proxy of the idiosyncratic differences among firms and, therefore, the omission of one group of variables may introduce biases and inconsistencies inthe estimation ofthe rest o fthe parameters ofthe model. Taking industry-region-size averages is also useful to mitigate the effect of missing individualIC observations at the firm-level, which represent one of the most important problems of IC surveys. It must be pointed out that, due to perfect multicollinearity, we can only use in the regressionsas many industry-region averages as the number of regions multipliedby the number of industries. Taking into account this issue and the number of missing values in most of the individual IC variables, in order to keep as many observations in the regressions as possible to 48Thistwo step estimationapproachhas an instrumentalvariables (2SLS) interpretation. 49 We do not believe on the plausibility of this solutions since it depends on the next identifying assumptionE(F!U) *,, = o,where &* are the residualsofthe shortregressionof logpion F,.This assumptionis difficult to holdifthere are industry-locationprocessescorrelatedwith ui*. 93 avoid losing efficiency, when the response rate o f the variables is large enough, it was decided to replace missing observations by the corresponding industry-region-size averages. Therefore, observations, efficiency and representativeness can be gained at the cost of introducing certain degree o fmeasurement errors insome variables.s0. The econometric methodology applied for the selection o f the I C and C variables goes from the general to the speczjk. The otherwise omitted variables problem that was found, starting from a too simple model, generates biased and inconsistent parameter estimates. The selection o f variables starts with a wide set compounded by up to 80 variables (depending on the region). It was avoided using at the same time variables providing the same information and likely to be correlated, mitigating the problem o f multicollinearity. Then the less significant variables one by one were removed from regressions the final set of variables all significant in at least one o f the regressions andwith parameters varyingwithin a reasonable range o f values was obtained. Tables A.3 to A.6 includes the sets o f I C and C variables that were significant in at least one of the 6 productivity specifications estimated for each metropolitan area. As the rest of the analysis will focus on only the two steps restricted Solow residual elasticities and semi-elasticities, the summary of the parameters obtained by usingthis productivity measure as dependent variable is inthe Table A.7. The measurement error introduces a downward bias in the parameters that will depend on the ratio between the variances of the variable and the measurement error, since the errors are constant within regions, sizes and industries there are reasons to believe that their variances will be small. 94 Table A.3: IC elasticitiesand semi-elasticitieswith respectto productivity (Bogota) Infrastructures innovation and Other control Notes: * significant at 10%;** significant at 5%; *** significant at 1%. Each regression includes a set of industry dummies and a constant term (a) Variables instrumentedwith the industry-region-sizeaverage. (b) Variables approximatedwith aproxy (only missingvalues replacedby the industry-region- size average). 95 Table A.4: IC elasticities and semi-elasticitieswith respectto productivity(Medellin) Two steps estimation Single step estimation Blocks of I C variables Solow residual Cobb-Douglas Translog Explanatory I C variables RestrictedI Unrestr. RestrictedI Unrestr. I RestrictedI Unrestr. Infrastructures Largest shareholder 0.004* 0.003' 0.004** 0.004** 0.004** I 0.004** Finance and Working caDital:farnilv/fiiends II-0.007** II -0.008**II -0.007** -0.007** -0.006** I -0.007** corporate governance I Owner ofthe lands I 0.003' I 0.003* I 0.003** 0.002* 0.003** I 0.002' I Dummyfor loanwith collateral (b) I 0.115 0.128 0.134 0.159 0.199' 0.155 QuaIity, IIDummyfor foreign technology (b) I 0.275 II 0.273 II 0.323* IIIII 0.288 IIIII 0.278 1I 0.252 innovation and labor skills Staff- universityeducation (b) 0.970* 0.976* 0.738** 0.748" 0.643** 0.479 Other control variables 1 Exportingexperience (b) 1 0.032 I 0.036 10.144*** 1 0.116** I 0.121" 1 0.081 I Dummyfor more than 5 competitors 0.210* 0.201* 0.11 0.155 0.092 0.148 Observations 213 213 213 213 213 213 R-squared 0.31 0.34 0.87 0.88 0.88 0.9 96 Two steps estimation Single step estimation Blocks of I C variables Solow residual Cobb-Douglas Translog Exolanatorv I C variables Restricted I Unrestr. RestrictedI Unrestr. I Restricted I Unrestr. Losses due to power outages (b) -0 017** -0 015' -0 015* -0 015'. -0 015** -0 019** Infrastructures Average durationofwater outages (b) -0 285*** -0 308*** -0 229.. -0 183'. -0 214** -0 116 Transport cost (a) -0 027** -0 027** -0 035*** -0 037*** -0 035*** -0 021 IWorkforcereportedto taxes (a) I 0.026** I 0.029** I0.035*** I 0.038*** I 0.051*** I 0.040* I variables Dummyfor Barranauilla I -0.344 I -0.332 I-0.511** I-0.570*** I -0.750*** I -0.582 Observations 134 134 134 134 134 134 R-squared 0.48 0.46 0.85 0.88 0.88 0.93 97 Two steps estimation Single step estimation Blocks of I C variables Solow residual Cobb-Douglas Translog Exolanatorv I C variables RestrictedI Unrestr. RestrictedI Unrestr. I RestrictedI Unrestr. 1 I . I I I I 1 I I Lossesdue to power outages (b) I I -0.017** I I -0.015* 1 I -0.015* I I -0.015** I I -0.015** -0.019** Infrastructures Wait for anelectric connection (a) -0.938*** -1.021*** -1.047*** -0.998*** -1.084*** -0.3 I I I I I I Average duration ofwater outages (b) -0.285*** -0.308*** -0.229** -0.183** -0.214** -0.116 Transportcost (a) I -0.027** I -0.027** I -0.035*** I -0.037*** I -0.035*** -0.021 I Workforcereportedto taxes (a) 0.026** 0.029** 0.035*** 0.038*** 0.051*** 0.040* Dummyfor security II 0.478*** II 0.407** II 0.410** 1I 0.322** 1I 0.283* IIIII 0.241 Red tape, corruption and Dummyfor crime -0.206** -0.191* -0.193* -0.195' -0.245** -0.183 crime Manager's time spent inbur. iss. (b) 0.005*** 0.004** 0.004*** 0.004** 0.006*** 0.005* Dummyfor payments to obtain a contractwith the government (b) -0.415** -0.408** -0.482** -0.429** -0.460** -0.563** Finance and Initial investment:private banks (b) 10.006*** I 0.006*** I 0.005*** I 0.005*** I 0.005*** I 0.007*** corporate governance Working capital: family/friends 0.007*** 0.007*** 0.005** 0.006*** 0.006** 0.009*** Quality, Dummy for quality certification 0.744** 0.693** 0.616** 0.511** 0.598** 0.413 innovation and lnhnr skills Outsourcing III 0.006* III 0.006 III 0.005* III 0.004 III 0.006** III 0.005 Exportingexperience(b) 0.261** 0.286** 0.218" 0.148 0.16 0.021 Other control Dummy for local mOIIOpOly -0.236 -0.177 -0.188 -0.317' -0.148 -0.04 variables Dummy morethan 5 competitors -0.152 -0.135 -0.091 -0.142 -0.191 -0.342* Dummy for Barranquilla -0.344 -0.332 -0.511** -0.570*** -0.750*** -0.582 Observations 134 134 134 134 134 134 R-squared 0.48 0.46 0.85 0.88 0.88 0.93 98 Table A.7: Summary of IC elasticities and semi-elasticitieswith respectto productivityobtainedfrom the two step estimationandusingth restrictedSolow residualas dependentvariable OfIC variables I Explanatory I C variables I Bogota(1) I Medellin (2) I Cali(3) I Barranquilla (4) I Colombia (5) I Infrastructures Red tape, corruption and crime Paymentsto obtain a contract with the government -0.093 -0.072 Paymentsto speed up bureaucracy -0.009 Largest shareholder 0.003 Initial investment:privatebanks 0.006 0.006 Initial investment:public banks 0.018 Finance and Salespaid after delivery -0.004 -0.003 corporate Working capital: family/friends -0.002 -0.008 0.007 0.007 -0.003 governance Owner ofthe lands 0.003 Checkingor saving account 0.231 Dummyfor loan with collateral 0.112 Quality, innovation and labor skills Other control Notes I)Regressionfor the firms locatedin the metropolitans areas of Bogota and Medellin with a regiondummy for Bogota. (2) Regressionfor the firms locate in the metropolitanarea of Cali. (3) &(4) Joint regression with interactionterms for the firms located in the metropolitanareas of Barranquillaand Medel I. (3) Regressionfor the entire sample (all firms). 99 IC evaluation on the Olley and Pakes decomposition: percentage contributions and simulations The I C impacts boththe productivity o fthe average Colombian firm andthe allocative efficiency (i.e., the efficient reallocation o f results from less productive firms to more productive ones). It is a well-known issue that competitive markets reassign resources efficiently. Let us suppose an economy in which all the markets work in an efficient way (in such a world, more productive firms are more competitive and the competitiveness in the markets o f both goods and inputs makes those firms to gain market share from the less productive ones), depending the gains only on the strategic interaction and the competition among firms. Nevertheless, a turbulent I C introduces distortions to markets and, as a result, to the efficiency o f the economy as a whole. Our belief is that the allocative efficiency term has to reflect these imperfections. Inthe second stage ofthe analysis, our aim is to take advantage o fthe robustness ofthe IC and C coefficients estimated and focus on only one set parameters (those coming from the two step estimators with restricted Solow residuals), to, by following Escribano, et. al. (2007), explore the relationships between I C and C variables and each one o f the terms o f the Olley and Pakes (O&P) decomposition. To do it, the linear properties o f the next O&P decomposition for each one o fthe regions considered with the variables expressed inlogs are exploited: log P, = loge + N, ~ o v ( S ~ p ~ ,P,,i) l o g By means of this decomposition, the aggregate log-productivity of the region "q" (logP,) is explained, as the sum o f the log-productivity o f the average firm o f that region and the covariance between the share o f sales in logs and the productivity in logs (Le., the allocative efficiency o f region "q "). The additive properties o f the logarithms o f the variables allow for the decomposition of log-productivity according to equation (A) and, after some straightforward algebra, from (B) the next expression for the aggregate log-productivity can be reached as a weighted sum o f the average values o f the IC, C and D variables the intercept and the residuals o f equation (A) and the covariances between the share o f log-sales, and IC, C and D variables andthe residuals: where the set o f parameters come from the two steps estimation with the restricted Solow residual as dependent variable. Equation (C) i s an exact relation and the reliability o f the equality only depends on the correct parameterization o fthe productivity equation (A). From equation (C), each I C and C variable may affect the aggregate log-productivity through: their averages and the covariance with respect to the share o f sales. This complements the information provided by the marginal effects - suppose that an I C variable with a low impact in terms of marginal effects is suffered by most o f the firms in a given region, the impact o f such a variable interms o fthe average firm will be dramatically increased. 100 A variable with a negative marginal effect on the average productivity (2q,,c) may have either a positive or a negative effect on the efficiency term. If the covariance o f that variable and the market share is positive, then the more proportion o f sales is in hands o f establishments with high levels of a variable that harms the productivity and, therefore, efficiency decreases. In contrast, a negative covariance means that those establishments with the higher levels of the variable have the lower markets shares and therefore the efficiency increases, or in other words, establishments with more problems with that variable are less productive, but at the same time those firms have lower market shares, which increase efficiency. By operating in (C), the next expression can be obtained, which brings up the possibility o f directly comparing the impacts o f each I C and C variable relative to the aggregate productivity: +N,~'q,C60v(s~~Y,C,,,) ~ ' q , D s 6 0D,)(+s N,~60v(s::~,ii,",,)] + N q ~ ~ Y , There are several advantages o f usingthe equation (D) instead o f equation (C). Contributions can be compared by isolating the impact o f I C variables from the impact o f the industry dummies, the intercept and the residual. Furthermore, it may be possible to know what portion of the aggregate productivity is explainedby the variability o f the I C and C variables and how much i s due to the constant term common to all firms. By expressing the impacts in terms o f relative contributions, the problems o f the measurement errors in the production function variables that are common to all firms within the same region may be mitigated. The results of equation (D) for each region are presentedin Tables A.8 to A.11. Finally, the relative contribution o f each group o f I C and C variables to the terms of the O&P decomposition i s computed. 101 Table A.8: PercentageContributionof IC andC Variablesto the Olley andPakesDecompositionofthe Aggregate ProductivityinLogs (Bogota) Average durationof power outages -2.14 -2.14 0.00 Infrastructures Water from public sources 28.50 28.43 0.07 Wait for awater supply -0.81 -0.77 -0.03 Sales reportedtot taxes -59.85 -59.75 -0.10 Red tape, corruption and Dummyfor conflicts incourts 2.88 2.76 0.11 crime Number of inspections 3.22 3.05 0.17 Pavmentsto obtain a contractwith the government -6.99 -7.03 0.05 Initial investment:public banks I 0.80 1 0.77 1 0.03 I Finance and corporate Sales paidafter delivery -13.01 -12.97 -0.04 governance Working capital: family/friends -0.36 -0.37 0.02 Value ofthe collateral 1 -2.76 I -2.74 1 -0.03 I Dummyfor environmental programs -7.70 -7.67 -0.03 Quality, innovation and labor Staff female workers - -5.65 -5.65 0.00 skills Staff- universityeducation 0.40 0.41 -0.01 Training to productionworkers 26.87 26.90 -0.03 Dummy for limited company I 10.59 I 10.47 I 0.12 I Other control variables Dummy for FDI 0.59 0.51 0.08 Dummyfor local monopoly -0.29 -0.31 0.02 Food I -2.73 I -2.78 I 0.05 I Apparels 2.50 2.45 0.05 Industry dummies Textiles 0.81 0.83 -0.02 Chemicals and plastics 5.41 5.33 0.08 Constant 119.02 119.02 0.00 Residual 0.71 0.00 0.71 I Total 100.00 98.74 1.26 Notes: The productivity measurei 102 Table A.9: PercentageContribution of IC and C Variablesto the Olley andPakes Decomposition ofthe Aggregate Productivity inLogs (Medellin) Aggregate Average productivity productivity Eficiency Duration of Dower outages bv month -~ -17.30 -17.43 0.13 I Electricity from a generator 1.63 1.29 0.33 Infrastructures Wait for an electric supply -7.92 -7.63 -0.29 Water from public sources 35.73 35.75 -0.02 Dummy own transport -5.82 -5.87 0.05 Shipmentlosses, domestic -1.50 -1.45 -0.05 Sales reportedtot axes -16.81 -16.73 -0.08 Red tape, corruption and Workforce reportedto taxes 19.84 19.73 0.11 crime Dummv for securitv 4.62 4.56 0.06 Lossesdue to crime -4.71 -4.81 0.10 Largest shareholder 8.36 8.33 0.03 Finance and corporate Working capital: family/friends -1.80 -1.93 0.13 governance Owner ofthe lands 4.37 4.08 0.29 Dummy for loanwith collateral 1.60 1.58 0.02 Quality, innovation and labor Dummy for foreign t e c ~ o b Y 1.03 0.93 0.09 skills Staff- universityeducation 1.43 1.32 0.11 Age -27.22 -26.69 -0.54 Other control variables I Exporting experience I 1.22 I 1.13 I 0.10 I IIDummy for more than 5 competitors 6.17 6.24 -0.07 Food -6.03 -5.86 -0.17 Industry dummies Apparels -6.93 -6.86 -0.06 Textiles -10.85 -11.11 0.26 Other -1.12 -1.09 -0.03 Constant 121.37 121.37 0.00 Residual 0.63 0.00 0.63 Total 100.00 98.86 1.14 103 Table A.lO: Percentage Contribution of IC and C Variables to the Olley andPakesDecomposition ofthe Aggregate Productivity inLogs (Cali) Aggregate Average productivity productivity Efficiency Lossesdue to power outages -0.22 -0.22 0.00 Infrastructures Average durationof water outages -0.86 -0.79 -0.07 Transport cost -5 1.78 -51.89 0.11 I Workforcereoortedto taxes 116.16 115.68 0.47 I Dummyfor security I -7.59 I -7.29 I -0.30 I Red tape, corruption and Dummyfor crime -2,04 -2.01 -0.02 crime Manager's time spent in bur. Issues 6.30 6.27 0.03 Dummyfor payments to obtain acontractwith the government -2.11 -2.17 0.06 Finance and corporate Initial investment:private banks 4.40 4.32 0.07 governance Working capital: family/friends 2.61 2.65 -0.05 Quality, innovation and labor Dummy for quality certification 2.03 1.94 0.09 skills Outsourcing 2.58 2.49 0.09 Exportingexperience 1.35 1.28 0.08 Other control variables Dummy for local monopoly -0.81 -0.77 -0.04 Dummy for Bogota -18.61 -18.61 0.00 Food 2.24 2.30 -0.06 Industry dummies Apparels 1.79 1.74 0.05 IITextiles -3.64 -3.66 0.02 Other 8.27 7.85 0.42 Constant 39.54 39.54 0.00 I I Residual I 0.40 I 0.00 I 0.40 I Total 100.00 98.65 1.35 104 Table A.ll: Percentage Contributionof IC and C Variables to the Olley and PakesDecompositionofthe Aggregate ProductivityinLogs (Barranquilla) I Aggregate I Average I Efficiency I Losses due to power outages -2.31 -2.30 0.00 Wait for anelectric connection -19.79 -19.68 -0.11 Infrastructures Averagedurationof water outages -0.42 -0.43 0.01 Transport cost -44.49 -45.00 0.50 Workforcereportedto taxes 105.17 105.10 0.07 Dummy for security I 16.64 I 16.69 I -0.05 I Red tape, corruption and crime Dummy for crime -4.00 -3.99 0.00 Manager's time spent inbur. Issues 2.36 2.34 0.02 Dummy for payments to obtain acontractwith the government -1.94 -1.99 0.06 Finance and corporate Initial investment:private banks 1.94 1.94 0.00 governance Working capital: familyhiends 1.46 1.50 -0.04 Quality, innovation and labor Dummyfor quality certification 2.20 2.16 0.04 skills Outsourcing 1.41 1.43 -0.02 I Exportingexperience I 3.82 I 3.63 I 0.19 I Other control variables Dummyfor local monopoly -0.44 -0.46 0.02 III Apparels Dummymore than 5 competitors -3.22 -3.24 0.01 Dummyfor Bogota 0.00 0.00 0.00 Food 2.72 2.76 -0.03 I 1.58 I 1.62 I -0.04 1 Industry dummies Textiles -4.07 -3.96 -0.11 Other 4.77 5.00 -0.23 Constant 36.12 36.12 0.00 Residual 0.46 0.00 0.46 Total 100.00 99.25 0.75 From now on, the linear properties o f the logarithm form o f the O&P decomposition have been exploited. It could be argued that the O&P decomposition in levels captures non-linear relations between market shares and productivity. To know to what extent these non-linear terms are affecting this relation, we propose to perform simulations in the I C and C variables, comparing the results with the ones obtained from the decomposition in logs. The simulations are done variable by variable (i.e., a scenario inwhich the levels o f the I C variable are 20 percent better in all the firms i s proposed)51 and the rates o f change o f the aggregate productivity, average productivity and o f the efficiency caused by such improvement relative to the initial situation are computed. We follow the same procedure for all the I C and C variables and for comparative purposes we compute the relative impact caused by the I C indicators group by ''Dollar et a1 (2003) proposedto compute how the productivity changes if countries with poor investment climate adopt the levels ofthe investmentclimate indicatorsof China. 52Ifthe variable has a positive signthe improvement implies higher levels of the variable, vice versa ifthe sign is negative. An improvement in the dummy variables means increasing (or reducing) the proportion of f m s taking value one (or zero ifthe effect is negative) by 20 percent, those f m s are randomlyselected. 105 .