77791 . The Implementation of Industrial Parks Some Lessons Learned in India Yannick Saleman Luke Simon Jordan . April 2013 . COMPETITIVE INDUSTRIES GLOBAL PRACTICE . . 1 Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. . Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750- 4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. 2 CURRENCY EQUIVALENTS (Exchange Rate as of February 15, 2013) Currency Unit = Indian Rupees (Rs.) 54 INR = US$1 1 Lakh (100,000) INR = US $1,850 1 Crore (100 Lakh) INR = US $0.185 million ABBREVIATIONS AND ACRONYMS APES Apparel Parks for Exports Scheme DIPP Department of Industrial Policy & Promotion DPR Detailed Project Report FPS Food Park Scheme IDC Industrial Development Corporation IIUS Industrial Infrastructure Upgradation Scheme IL&FS-CDI Infrastructure Leasing and Financial Services – Cluster Development Initiative MFPS Mega Food Park Scheme MoT Ministry of Textiles, Government of India PAC Project Approval Committee PMC Project Management Consultant PPP Public-Private Partnership PSC Project Scrutiny Committee SEZ Special Economic Zone SITP Scheme for Integrated Textile Parks SPV Special Purpose Vehicle TCIDS Textile Centers Infrastructure Development Scheme ToR Terms of Reference 3 1 ABSTRACT Industrial parks are as popular as they are controversial, in India and globally. At their best they align infrastructure provision and agglomeration economies to jolt industrial growth. More often, they generate negative spill-overs, provide handouts, sit empty, or simply do not get built. This paper disaggregates how parks are built and how they fail. It contextualizes parks in India, followed by a thick case study of an innovative scheme that appears to buck the trend. This performance is then explained by the way in which the scheme’s design and action fits India’s political economy. The paper concludes by considering how the analysis and the lessons learned might inform the design and implementation of industrial park programs and other public interventions, in India and elsewhere. TABLE OF CONTENTS INTRODUCTION .................................................................................................................................. 5 I. INDUSTRIAL PARKS: THEORY AND PRACTICE ....................................................................... 7 II. DESCRIPTION OF THE SITP MODEL IN INDIA ....................................................................... 11 III. EARLY RESULTS: PARK COMPLETION AND DEMAND GENERATION ........................... 16 IV. EARLY RESULTS: CLUSTER EFFECTS AND POLICY SPILLOVERS.................................. 22 V. EXPLAINING RESULTS: SITP DESIGN & THE POLITICAL ECONOMY OF IPs.................. 26 VI. REMAINING PITFALLS AND THE CHALLENGE OF SCALING UP..................................... 34 CONCLUSIONS: LESSONS AND FURTHER RESEARCH............................................................. 37 EXHIBITS ............................................................................................................................................ 39 1 The authors wish to thank our principal reviewers Alan Gelb (Senior Fellow, Center for Global Development), Ejaz Ghani (Lead Economist, World Bank), and Professor Mushtaq Khan (Professor of Economics, SOAS, University of London). In the drafting of the paper we received invaluable guidance from Ivan Rossignol (Chief Technical Specialist, Competitive Industries Global Practice) and Professor Charles Sabel (Columbia University). The paper also benefited from the valuable comments and inputs of Laurence Wilse-Samson (Columbia University) and Vincent Palmade, Stefano Negri, Megha Mukim and Douglas Zhihua Zeng at the World Bank. We are very grateful to Ms. Sunaina Tomar, IAS, Joint Secretary, as well as to Ms. Sangeeta Saxena, Deputy Economic Advisor, Ministry of Textiles, Government of India, for fruitful discussions of the scheme and for allowing us access to all required data. Special thanks go to Mr. Raman, Executive Director, Mr. Akshay Patel, Punita Bansal, Sharmistha Mohapatra, Riya Dhawan, Prashant Sood, Ketul Acharya, Dilip Gangadharan, M.G.Gnaneswaran and many others at IL&FS for their strong and continued cooperation and support for data provision and the organization of field visits. We are grateful to all the entrepreneurs met during those visits for their valuable time and enriching discussions. 4 INTRODUCTION The theory of industrial parks is seductive. It holds that governments must provide public goods, particularly infrastructure, for industrial firms to be competitive; co-located firms generate agglomeration economies; and so the public goods should be concentrated on areas of colocation. If the firms are not there yet, the provision of infrastructure will somehow induce them to be. If the area is demarcated, and the general economic environment difficult, exemptions can be made that will further entice firms and create competitiveness (in this case making the parks what is most generally termed “Special Economic Zones”, or SEZs).2 There are some cases where the practice has approached the theory. The most commonly cited cases are in East Asia, but parks also dot much of the developed world. Many question the true cost-benefit of such parks, though often with somewhat weak counterfactuals. In much of the developing world, though, the case is somewhat simpler. The practice is far enough from the theory for subtle arguments to be moot. Industrial park and SEZ programs are often rightly criticized for producing white elephants; eroding the tax base; creating vehicles for land speculation; delivering hand-outs to favoured firms; and funneling spending to favoured districts. That is if the parks are even completed in less than a decade. More than one of these failures has afflicted industrial park schemes in India over the last few decades. The instrument though has continued to grow in its use, if often rechristened as “cluster development” or using variations around the term “zones”. Many of the most high-profile programs have very mixed records, often delivering under-target, several years late and with low take-up. However, in the last few years a program has been developed and implemented in India, originated by the Ministry of Textiles, which seems to perform substantially better than the norm there. While results are still indicative, the scheme gets parks built; it guards against speculation and capture; the parks attract investment; and, in several cases, they show signs of generating real cluster effects. At the core of the scheme’s relative success are the details of its design, and the manner in which those details fit to the political economy of building parks in India. In particular the scheme keeps parks right-sized, neither too small nor too large; makes location and dimension decisions costly to those who make them (much more costly than forfeiting a consulting incentive fee); reduces the role of forecasts; places the burden of navigating local informal relationships on those best placed to do it; aligns roles and incentives; creates ‘soft’ infrastructure through the way it builds ‘hard’;3 and, from the outset, focuses the public sector and its hired consultants on catalyzing and monitoring groups of firms as owner-users, rather than initiating, designing or managing the parks. In the model, government grants are closely tied to areas where there are immediate gains from coordinated action: common infrastructure and facilities and sometimes more limited help with land. Firms which seek to gain access to support under the scheme need to cooperate on a number of immediate issues in order to access assistance. This builds agency that can lead to further externalities, whether through cooperation or “thick” markets such as for machinery and labor. This 2 The term “SEZ” can sometimes be used by policy-makers with a specific meaning as regards the regulatory and other (mostly tax-related) exemptions provided, but it is considered the most generic term in the literature for parks providing any type of regulatory exemptions, and therefore was used as such here. 3 “Hard” refers to physical infrastructure, such as roads, rail and power, and the operational systems required for it to operate. “Soft” refers to social infrastructure, such as organizations for collective action, firm capabilities, skills, hea lth services, cultural preservation, and so forth. 5 does not, however, work automatically: the effect requires a close fit between the incentives created and the political and entrepreneurial context. It required substantial early work by the lead Ministry and its advisors to identify firms capable of leading this process among their peers, though once a demonstration effect had been created they were able to switch to a heavy focus on monitoring. The study does not though begin to claim that the scheme is a panacea. It remains early days, even if the signs are encouraging. In particular, though the scheme seems to be generating forms of cluster organization, it could experiment with doing more to capitalize on these and generate agglomeration economies quicker. The first wave of parks, now reaching operational mass, might be fertile grounds for such repeat experiments. However, this is not a model which can be understood simply. Its performance is not uni-causal and its features are complex and deeply related to its context, in particular its local political economy, industry structures and firm capabilities. The scheme depends on the existence of a base of firms with the financial means to participate in park building (typically requiring $44,000 in equity) and more importantly with the management capabilities to at least initiate self-organization. It requires at least some “leader” firms, with the ability to recognize the value of common organization, and estimate that it outweighs the risks of free-riding and delays by “follower” firms, who themselves may join for more near-term instrumental benefits, such as reducing the transaction costs of obtaining sufficient land and scale in local infrastructural services. In turn, though, this creates a risk of such leader firms, if large or able to wield substantial influence, turning this organization into a vehicle of capture. The fragmented nature of the textile industry in India likely mitigates this risk, but this may not hold if the scheme is “pushed” too quickly into other sectors or contexts. Attempting a superficial replication is then likely to lead to one or more forms of failure, including creating the scope for corruption and collusion (and there are some signs that, even in India, the scheme is being “pushed” into sectors where it may not be appropriate). However, this does not mean there is no scope for learning, and it is our belief that there are a range of environments in which this scheme may have useful features. Nevertheless, this paper does not attempt to present a best practice for implementing industrial parks. It attempts to deepen our knowledge of how and why industrial parks fail, and how schemes can sometimes be designed in their detail, and implemented in practice, to avoid at least some types of failure. Perhaps most broadly, it attempts to delineate the type of questions which must be asked when designing the implementation of these and other industrial policy programs. These include: what are the precise steps involved in undertaking the intervention? What performance failures might arise as these are undertaken? Where might these steps, and these performance failures, intersect? Who is given responsibility for each step? How strong or weak are their incentives to avoid the performance failure? Are there others, capable of performing the step, with stronger incentives? What is the size of each discrete intervention, and how does it interact with the prevailing political settlement (how noticeable is it, and is that beneficial or harmful)? What ripple effects could there be on the long-term pressures for or against reform, by either strengthening or removing different voices? The remainder of the paper is structured as follows: the first section disaggregates the roles in building industrial parks, the types of failure witnessed in implementation, and relates these to each other. The second describes the model under consideration, and the third provides data on its results so far. The fourth considers tentative signs of agglomeration economies and positive and negative spillovers beyond the parks. Section five explains its performance by considering its alignment of formal and informal incentives and capabilities. Section six examines the challenges to scaling up and replicating the scheme. The paper then concludes with directions for further research and action. 6 I. INDUSTRIAL PARKS: THEORY AND PRACTICE The rationale for industrial parks has traditionally been twofold. First, the provision of functional infrastructure is much easier to plan in a geographically limited space, particularly for delivery- constrained governments. 4 Second, the concentration of firms can provide significant spillover effects both inside and outside the park5: information spillovers, including knowledge and technology; the specialization and division of labor among enterprises; the development of skilled labor markets; and the development of markets around the parks.6 The general type of an industrial park is a concentrated set of firms operating on a demarcated ‘hard' infrastructure platform. Beyond this, a number of different forms have been developed. Most involve additional regulatory features, often to bypass unsupportive business environments, facilitated by the “limited” area involved, in which case the umbrella term is “Special Economic Zones” (SEZs) . Such variations include: “tax” parks (which provide tax benefits); technology parks; “green” parks; “export processing zones”; and many others. In this paper, “industrial park” will be used in its most general sense, i.e., involving the provision of common infrastructure to a group of industrial firms in a demarcated area, while forms involving additional regulatory or other measures will be specifically denoted as such. In whichever variant, industrial parks in theory are a tool to develop a more general form of industrial organization: industrial “clusters”, a concentration of interconnected firms in a particular field. Such clusters can form “organically”, or can be the target of deliberate policies.7 The pursuit of active cluster development, like that of industrial parks themselves, can be a controversial policy goal. On the one hand, the growth of industrial clusters is frequently cited as a driver of China’s rapid growth. Many emerged spontaneously, but government (especially local government) also provided crucial support for their development. 8 On the other hand, cluster development, and industrial park and zone programs in particular, have also been subject to significant criticism. There have been controversies regarding land allocated for their construction, sometimes at large scale, especially when parks are long delayed or scarcely occupied, fueling accusations of land speculation. It has also proven difficult to demonstrate additionality, i.e. that the activity in parks would not have happened without their construction and the public money that has been spent to support it. This latter criticism has been strongest for policies that not only provide public money for park construction but forego revenues in the forms of taxes or resort to specific regulations such as looser labor laws to attract investment. Yet the criticisms seem to have had little to no impact on practice. Whether or not one or other side of the general debate is correct, park and zone programs continue to proliferate, and many continue to under-deliver. To take just a few examples, many parks in Central America have eroded the tax base there; in Africa, industrial parks are often known to stand mostly empty even after many years; and a number of industrial parks in Afghanistan supported by various donors either did not become 4 Marshall, A. (1920). Principles of Economics. London: Macmillan. 5 This review builds on previous writings by Douglas Zhihua Zeng (2006, 2010, 2011). 6 Sonobe, Tetushi, & Otsuka, K..(2006). Cluster-Based Industrial Development: An East Asian Model. New York: Palgrave Macmillan. 7 World Bank. (2010). Innovation Policy: A Guide for Developing Countries. Washington, DC: World Bank Institute; Porter, M. (1990). The Competitive Advantage of Nations. New York: Free Press; Schmitz, H. (1995). “Small Shoemakers and Fordist Giants: Tale of a Supercluster.” World Development 23 (1): 9–28.. 8 Ganne, B., & Lecler, Y., eds. (2009). Asian Industrial Clusters, Global Competitiveness and New Policy Initiatives. Singapore: World Scientific Publishing Co. Pte. Ltd. 7 operational or failed to attract demand. Yet a glance at pipelines of development programs across continents finds no slackening in the creation of new industrial park programs.9 Since the available evidence strongly suggests this will continue, there are likely more gains in understanding how and why park programs do or do not deliver than in attempting to frame a general law of their merits. These general characteristics apply strongly to the Indian context. For decades State-level industrial development corporations have been building industrial parks. At the Centre, multiple Ministries and Plan periods have framed new industrial park schemes. The most visible recent attempt was the SEZ program, launched in 2005, which stumbled in implementation.10 Many examples of successful parks have been built in India, through both Centre and State initiatives, but the overall record is decidedly mixed, and timelines tend to be long and lengthening (as will be described in more detail below). 11 Several factors are often posited to explain the failures of industrial park programs. Some are highly specific, such as the threat of conflict, but most are broad, for example, poor sector targeting; poor location choice; insufficient investment in infrastructure; poor implementation capacity and lack of authority; and lack of high-level support and stability.12 When these are surmounted, or sidestepped, relative failure is also attributed to factors such as a lack of support for technology acquisition, skills building and quality assurance, or other actions that would generate “cluster effects”.13 Such diagnostics focus on the absence of particular institutional forms, rather than the functions that should be served by the different stakeholders.14 This runs this risk of building in “strong priors about the nature of the problem and the appropriate fixes”,15 and can stand in the way of finding solutions in terms of performance. We therefore find it more useful to classify implementation failures of industrial parks into four main performance categories: (i) the parks do not get built; (ii) the parks are built but there is little demand from firms to locate and invest in them; (iii) the parks are built and generate demand, but with few “cluster effects”; (iv) the parks are successful but have a neutral or negative side-effects on investment climate outside the park (“negative spillovers” and “crowding out”).16 Two important points must be made on this classification. First, categories (i) and (ii) relate to a performance failures in terms of outputs, while (iii) and (iv) relate more to outcomes. The underlying causes affecting park building (i) and investment attraction into them (ii) are very different from those affecting the development of clustering externalities (iii) on the one hand and the provision of local public goods (iv) on the other. It is still early stage for the scheme examined here to provide a formal evaluation of the last two failures, and hence to make definitive statements about outcomes. As such, 9 For example, in a recent analysis of the WB CI Practice affiliate’s desire for knowledge by topic, the largest unmet demand for knowledge related to parks and zones. The authors are personally aware of park programs recently formulated or in development in countries ranging from Ethiopia to Brazil to Myanmar to Afghanistan. 10 World Bank (2012), “Learning from SEZs in India”, Unpublished. 11 Farole, T.,, and Akinci, G., eds. (2011). Special Economic Zones: Progress, Emerging Challenges, and Future Directions. Washington, DC: The World Bank. 12 Dinh, H.T. et al. (2012). Light Manufacturing in Africa: Targeted Policies to Enhance Private Investment and Create Jobs. Africa Development Forum Series. Agence Française de Développement and The World Bank. 13 Zeng, D.Z. (2006). Knowledge, Technology and Cluster-based Growth in Africa: Findings from Eleven Case Studies of Enterprise Clusters in Africa. Knowledge for Development (K4D) Program, World Bank Institute, The
 World Bank. 14 Blum, J., Manning, N. & Srivastava, V. (2012). Public Sector Management Reform: Toward a Problem-Solving Approach. Economic Premise, Number 100, December 2012. The World Bank. 15 Rodrik, D. (2008). The New Development Economics: We Shall Experiment, but How Shall We Learn? Cambridge, MA: John F. Kennedy School of Government, Harvard University. 16 What we term “cluster effects” and “negative spillovers” are explained in detail in section IV. 8 the study is principally focused on outputs: as these are necessary, if not sufficient for outcomes, and it is precisely the production of these outputs which has proven difficult and elusive (as discussed further below). Therefore the focus of this study is on evaluating how and why a scheme has managed to perform at getting parks built and attracting demand for them. However, the scheme’s potential in avoiding failure types (iii) and (iv) is explored, and areas of focus for a future evaluation identified. Second, the ordering above is time-wise. In particular, the opportunity cost of (i): not getting the parks built, and therefore probably not spending all the public money that was committed, could be lower than the opportunity cost of (iv): successful parks but with negative side-effects (such as tax erosion, or diminished pressure for reform, discussed below) if those side-effects are large enough. However, there is not an a priori ranking of the opportunity costs of the failures, because of the variation in the size and configuration of schemes. A very large “white elephant”, whether built and empty or absorbing human and political capital and financial commitments (if not disbursals) for years, could have a larger opportunity cost than a moderate park with minor external effects. These performance failures can be related to the process of building and operating industrial parks, the stakeholders involved and the functions they have been assigned. The steps in the process can be broadly enumerated as: (1) identification of park location; (2) identification of potential demand and overall dimensions; (3) procurement of land; (4) design and dimensioning (“master planning”) within the park; (5) financing and financial structuring and planning; (6) procurement of infrastructure building; (7) construction of infrastructure; (8) operation & maintenance; and (9) monitoring and evaluation. These roles have been shared in many different ways between national, sub-national and local levels of government, and as many as four types of private sector firms: consultants hired by the government for demand identification, and/or designing and financing of the parks; “developers” who typically prepare the master plan for the parks; hired “constructors” and “operators” who construct the parks and manage their operation and maintenance (i.e., who follow more than make decisions); and “users”, i.e. the private entrepreneurs occupying the park themselves. In this light, it is of limited use to call an industrial park delivery model “private sector driven” or “public -private” as soon as the private sector is “involved”. With a large number of process steps and roles, the critical questions are: who does the “private sector” refer to (real estate developers, consultants, users)? And, what steps will they undertake? In this light, the pertinent question for a park scheme is how the choice of a certain part of the public or private sector to implement these steps increases or decreases the possibility for such failures. The phrases “private sector” or “PPP” need to be used carefully. Often, this implies private sector involvement primarily in the later steps of internal dimensioning (4); infrastructure construction (7) and park operation & maintenance (8), whereas the more crucial location selection (1), and demand identification and overall dimensioning (2) are predominantly, if not exclusively taken on by a public entity. Even if this entity is supported by a technical consultancy, client-consultant agency and conflict of interest problems mean such steps may be de facto public and very often political. Moreover, the “private sector” is heterogenous. The incentives and costs facing a consultant, a developer and the users of the park are very different, as is their situation – formal and informal – within the local political economy. It is important to caution against the prior assumption that “the private sector” will be better at all steps. Whether this is so, and who is meant by the “private sector”, will depend on specific capabilities and incentives. The design of industrial park programs should consider each of these steps, and seek to understand, in the light of the local political economy, which 9 actors would be better or worse suited to which roles. Among the conjectures we will explore below is that such “role allocation” is the most important parameter for park implementation. To illustrate, Table 1 provides an indication of how the different functions or steps enumerated above might hypothetically map with the different types of failures mentioned earlier in some instances. Table 2 gives an example of the mapping of actors with implementation steps in a typical “public- private partnership” model, showing the limited role played by the users/entrepreneurs, which will become a central point of the discussion below. Table 1: Steps in industrial park development and potential failures Function vs. (i) Park does not (ii) Not enough (iii) Limited cluster (iv) Negative Failures get built demand effects spillovers (1) Location X X selection (2) Demand X X identification (3) Land X X X procurement (4) Design & X X X X dimensioning (5) Financing and X X structuring (6) Infrastructure X procurement (7) Infrastructure X construction (8) Operation & X X maintenance (9) Monitoring & X X evaluation (M&E) Table 2: Example mapping of stakeholders and responsibilities Steps vs. Government / Private Private Pvt construction / Park users Stakeholders public agency consultancy developer operation firm (firms) (1) Location X selection (2) Demand X X identification (3) Land X procurement (4) Design & X X X dimensioning (5) Financing and X X structuring (6) Infrastructure X procurement (7) Infrastructure X X construction (8) Operation & X X maintenance (9) M&E X 10 II. DESCRIPTION OF THE SITP MODEL IN INDIA The Scheme for Integrated Textile Parks (SITP) was inaugurated in 2005 by the Ministry of Textiles (MoT) of the Government of India (GoI). The objective was to create “world class infrastructure” for the industry and create “new textile parks of international standards at potential growth centres” to realize the full potential of the Indian textile sector. The SITP’s most prominent innovation is the far greater, and far earlier, role it gives to the users of the park. In contrast to almost all other schemes the authors are aware of, the roles of the Centre and State are diminished, consultants are used quite differently than in common practice, and third-party developers are conspicuous by their absence. The former two concentrate on organizing, supporting and monitoring groups of firms, who must then navigate the formal and informal requirements for getting the park built. In terms of specific roles, the MoT aimed to provide a robust implementation and funding framework, but then to delegate actual implementation and focus on monitoring. The entrepreneurs who will use the park are first required to organize themselves by forming a Special Purpose Vehicle (SPV). The SPV then selects a specialized Project Management Consultant (PMC) from a list empanelled by the MoT and under a standardized Terms of Reference (ToR) supplied by the Ministry, to prepare a Detailed Project Report (DPR) for the MoT.17 How the choice from the list is made is left to the SPV members’ discretion. The SPV is then provided with a partial grant for the construction and operation of the park, for which they are also required to contract a PMC, though at this point it does not have to be chosen from the MoT panel. In practice, this PMC (the “SPV-PMC”), chosen for the execution of the park, is usually the same as the “MoT-PMC” which prepared the DPR and later provides advisory services to the MoT.18 For simplicity, we will therefore restrict our attention to this case in the remainder. In more detail, the policy makes the entrepreneurs the drivers and ultimate decision-makers of the entire initiative for the creation and functioning of the park, but with support from the PMC across the steps of industrial park development. The PMCs are required to possess “considerable infrastructure development experience”, for “speedy implementation of the project”. The bar to be an empanelled PMC is high: as of 2012, only seven PMCs had been empanelled, across India.19 Project structuring, feasibility study and costing of all aspects of the project are done by the PMC, according to decisions made by the SPV, which has the responsibility to define its needs and make final decisions on location, project design and costing. When complete, the PMC transmits the DPR for appraisal to a Project Scrutiny Committee (PSC) headed by the Joint Secretary of the Ministry of Textiles. If passed, the project is submitted for approval to a Project Approval Committee (PAC), headed by the Minister of Textiles. If the project is approved, the Central Government then contributes a significant share of the funds for common infrastructure, and the SPV finances the rest. This contribution is up to 40% of the total or 40 crores (~$8m), whichever is lower, as “grant-in-aid”, i.e. this does not constitute equity in the SPV. SPVs have the possibility to solicit States and state/locally-affiliated bodies such as Industrial 17 Empanelled PMCs as of 27 March 2012 were Infrastructure Leasing and Financial Services Limited – Cluster Development Initiative (IL&FS-CDI), ICICI Winfra; Industrial Consultants; CS Architects; Kushal Global-Pearl Academy of Fashion;.Technopark Advisors; and Magus Consulting 18 IL&FS-CDI, the largest PMC on the MoT panel by number of parks managed (80%), says they have played both advisory and execution roles in 80% of the parks they manage, and that the ratio was similar for parks managed by other PMCs. 19 The policy is now being revised and a new panel will be chosen. 11 Development Corporations (IDCs) for additional contributions to finance their 60% share, for example under State-funded schemes. State/local contributions can also be under the form of subsidized land or other benefits. PMCs may also invest in the SPV’s equity, if interested and requested by the entrepreneurs.20 However these additional sources of financing are subject to the SPV members contributing at least 51% of the equity, so that they retain managerial control. There is a limit of 20% of ownership for any single private entity within the cluster, so that there is no legally dominant firm, 21 and the land area occupied in the park must also be proportional to the equity contribution. The MoT initially provided a guideline of approximately 50 members per park, but has since allowed some flexibility by providing a matrix of the additional investment expected per entrepreneur when the number is lower. Figure 1 shows a diagram of the basic SITP scheme structure and grant financing, while Figure 2 highlights potential state and other contributions. Figure 1: SITP design and grant structure SITP Design & Basic Grant Structure SPV receives 40% of project costs as grants, so up to 60% of costs are born by SPV members Assists in identification and procurement COMMON INFRASTRUCTURE Clearances State Internal roads, storm water drainage, power, common facilities... (Gvt, IDC...) (excludes land & machinery; sometimes buildings) Monitors Implementation Most cases SITP only: implements Uses funds to buy, own, operate and maintain PMC Helps Establish, Advises SPV Purchase LAND Usually same SPV formation, Project Structuring, Engineering, created by industry Usually fully privately for SPV & MoT freedom of location, Financing, Management association Invest Fee Grants Designates (40% of cost) (60% of cost) & Monitors Monthly Charges O&M, sinking fund, debt Choice among panel, after reserve account... Entrepreneurs Invest Individual Expression Of Interest (EOI) MoT Led & coordinated by a Assets Dedicated JS level “promoter” e.g. machinery Needs assessment and capacity building 0 20 IL&FS has participated in the equity in two parks, the Palladam Hi-Tech Weaving park in Tamil Nadu, and the Pochampally Handloom Park in Andhra Pradesh. 21 20% is the legal threshold for giving key decision powers to an equity holder, although the threshold was not explicitly mentioned in the policy document and it was not clear if it could be circumvented by the use of two legal entities of the same group. 12 Figure 2: State and other bodies’ contributions State/Local and Other contributions State/Local & other institutions can contribute subject to SPV retaining 51% of equity share Assists (identification & procurement) Clearances & COMMON INFRASTRUCTURE add-ons State/IDC Monitors Implementation Assists with or Buys, Own, Acquires Operate, Maintain Helps Establish, Advises Identifies & PMC Invests SPV LAND Procures E.g. IL&FS through PE fund Invest Designates (min 51% of Grants equity share) & Monitors Individual MoT State/IDC Entrepreneurs Assets Needs assessment and capacity building 0 To raise funds for the SPV’s share of costs, the PMC supports the SPV to structure and obtain financing, most often securing term loans for 40% of the costs by using common assets as collateral. Financial innovations may be used to facilitate securing the loan, such as including a 6-month debt repayment fund, 22 or having reimbursements made individually, despite the debt being raised in common through the SPV, to limit free riding23. This leaves approximately 20% of the costs as the equity portion of funding, to come in the most part as equity from the parks’ users. Project costs, as described by one of the PMCs, include: land purchase and site development 24 ; common infrastructure (roads, water supply, storm water drain, power provision, etc.); factory buildings; common facilities (lab, training center, etc.); engineering fees (for engineering design); and pre-operative costs (depending on components, and usually around 5% of costs). Factory buildings are eligible for grants only if built and owned by the SPV, and not by individual firms in the cluster. The latter option is most common when firms in the cluster are heterogeneous. If firms are more homogenous, there can be large cost savings and added flexibility by building identical factories. This might have implications for the fitness of the model to different types of industries, due to the level of homogeneity in firms’ processes. For example spinning or weaving parks would typically be homogenous, as opposed to garmenting parks, and the conjecture is that the scheme structure might favour the former type, in terms of project costs and/or completion time. We do not have enough data 22 The debt repayment fund is effectively a debt service reserve account that allows the SPV to sustain up to 6 months of non-repayment of the loan, for example in the case of a liquidity crunch, before defaulting on the loan (which would trigger the seizure of assets by the lending institution). 23 This innovation was described by IL&FS as originating from them and being unique among empanelled PMCs. 24 Land is not eligible for the Central Government grant, but instead counted as equity contribution from the SPV, who is responsible for finding and buying it. 13 at this point to make such an analysis, but flag it for further research. Machinery is excluded from the costing of the project for the provision of grants, and is to be funded by individual units. During operations, the SPV charges its members monthly to recover the money from investments made and repay the debt. The level and structure of these charges are often designed by the PMC during project initiation. The largest PMC usually includes in the fees: maintenance charges; utility charges; infrastructure charges proportional to the area occupied; contributions to a sinking fund (in case of major repairs); and financial payments 25 . Figure 3a and 3b show diagrams of the debt financing mechanisms used and possibilities of financial structuring. Note that these financial structures are at the discretion of the users, PMCs and financial institutions. New financing models for the costs not funded by the grants do not require MoT sanction, encouraging innovation. Finally, project monitoring and evaluation are undertaken by the MoT, through the PMCs. The PMC is required to devise a “suitable monitoring and evaluation system” and must furnish “monthly reports/returns to the MoT” (see Annexes 1a and 1b for samples). The PMCs are paid a fee by the Ministry for their work (we consider the fee structure in Section V). Figure 3a: Typical debt financing structure 25 For example IL&FS suggests firms contribute around 10% of the monthly payment to a Common Debt Repayment Service Fund (CDRSF), to meet dues in case of default of a member. This solidarity mechanism typically allows the structure to survive temporary liquidity issues of a member, for example because of delays in getting payments from clients. The member is asked to replenish the fund when the situation normalizes. These innovations would be conditional of MoT review and approval at the Project Scrutiny stage. 14 Figure 3b: Alternative debt repayment mechanisms (innovation of PMC IL&FS) Comparing this model to the implementation framework developed above, the SPV, with the PMC, is involved in or has exclusive execution responsibility for all of the steps (1) through (8). The crucial decisions are costly to those who make them, who are also those with the thickest information about the local conditions needed to site the parks. Location selection (1) and land procurement (3) are supported by the PMC but decided and executed by the entrepreneurs themselves. By participating in the SPV the entrepreneurs make a costly signal of demand, providing robust demand identification (2), even if, as does occur, the government or the PMC initiate the process by approaching the entrepreneurs. Demand is therefore more or less actual, rather than potential or projected, before subsequent steps are triggered. Dimensioning (4), financing (5) and procurement are the responsibility of the PMC. Infrastructure provision (7) and O&M (8) are executed by specialized contractors, chosen by the SPV (though again with assistance from the PMC). Figure 4 illustrates a typical PMC’s involvement in the park development process. Two initial notes are important here on the capabilities of the firms involved, and on what the scheme offers to them. First, while the aggregate investment requirement from participating firms may seem large, it can and often is brought within range of a wide number of firms through the financing arrangements. Of the Rs 2 Cr (~$400k) average cost per firm, Rs 0.8 Cr (~$160k) are financed through grants, most often an additional Rs 0.8 Cr through bank loans and in many cases Rs 18 lakhs (~$36k) through state grants, leaving a still sizable but less stringent Rs 22 lakhs ($44k) to be paid directly as equity per firm. While substantial, this should be within range of a small to medium firm, for example in weaving being approximately the same cost as one to two modern autolooms. A more difficult requirement is likely to be that for the management capabilities to create and manage the SPV, including the ability to identify fellow firms and motivate them to join. This will be treated in more detail below, but the critical role here seems to be that of a “lead” firm (or firms), which see early the potential of the scheme for creating collective benefit, and which then identify other 15 “follower” firms to participate. These latter must be convinced to delay set-up or expansion plans and take part in a collective enterprise requiring a high degree of trust and a costly commitment of capital. Not only must such firms exist, but the scheme must offer to solve some problem they face which they find difficult to solve alone. While it is difficult to make definitive conclusions on this score, our conjecture is that the “lead” firms are motivated by the promise of cluster effects, and the “followers” of solving the small-scale infrastructure service and land purchase market failures endemic in India. This will be discussed at greater length below. Last, we make note of what the SITP does not offer. The Center provides no assistance in procuring land, while States may do so, but have tended not to, and have tended not to be asked (a topic discussed at more length below). SITP parks have no exceptional regulatory provisions and no ongoing fiscal subsidies to the firms as individual units or to the SPV. It funds only 40% of the cost of the parks’ infrastructure. In money and regulation, it is less generous to firms than most industrial park schemes, which should be borne in mind as we turn to consider its results. Figure 4: PMC project development process snapshot26 III. EARLY RESULTS: PARK COMPLETION AND DEMAND GENERATION Relative success in getting parks built To analyze the SITP’s results to date, data was obtained primarily from Infrastructure Leasing and Financial Services, Cluster Development Initiative (IL&FS-CDI, here simply referred to as IL&FS), following approval from the MoT27. IL&FS is the most active PMC28, with 32 projects currently 26 Modified from an IL&FS document 27 A meeting with the Ministry of Textiles confirmed that using IL&FS data on their parks would be most efficient to conduct an analysis of the scheme. 28 IL&FS is a limited company headquartered in Mumbai and with equity participations from Life insurance Corporation of India (25.9%), ORIX Corporation Japan (23.6%), Abu Dhabi Investment Authority (11.3%) and Housing Development Finance Corporation (10.7%), as well as the IL&FS Employees’ Welfare Trust (10.9%) and other smaller participations including from CBI and SBI. It started as a leasing company, then expanded into doing infrastructure PPPs. The creation of IL&FS-CDI, initially as an integral part of IL&FS and then as a separate (but 100%-owned) subsidiary, was triggered by the launch of the SITP scheme in 2005. 16 under management, of the 40 approved in the first round of the scheme. The data was complemented with field visits and interviews conducted at multiple sites in two states with among the largest number of parks (Gujarat and Tamil Nadu), as well as meetings with relevant state authorities and industry associations when possible (further details of field visits are provided in Annex 2). The data provides a strong indication that type (i) implementation failures (parks not being built, or not in a timely manner) are largely being avoided under SITP. The scheme began in 2005 with a target of 25 parks, later extended to 40 parks because of high demand. By 2011, all 40 parks had been approved and had at least begun construction. By 2013, almost all were operational or close to being. Specifically, of the 40 parks, the breakdown of advancement as of October 2012 was as follows: 11 fully completed; 13 started commercial production; and 16 parks at various stages of construction. The list of parks and their basic characteristics is given in Exhibit 1 at the end of this paper. The time- wise approval of parks under SITP is as per Figure 5 below. Figure 5: time-wise approval of textile parks under SITP Year of Approval Number of Parks FY 2005-2006 07 FY 2006-2007 17 FY 2007-2008 03 FY 2008-2009 12 FY 2009-2010 01 TOTAL 40 In terms of spatial location, the first wave of parks have principally located in the western and southern states, particularly Gujarat, Maharashtra, Tamil Nadu and Andhra Pradesh. Few located in major cities, but most were set up in secondary cities with strong industrial clusters (such as Surat in Gujarat, or in the textile belt around Coimbatore in Tamil Nadu). This accords with recent studies finding a shift of industrialization in India to secondary cities, a development which is likely to have beneficial effects for the inclusiveness of growth.29 Among IL&FS parks for which we have data there is considerable variance in time-to-completion between projects. Most were completed or are expected to be completed within an average (and median) of around 6 years from approval time. The fastest has been completed in 3.9 years and the longest is expected to be completed in 9.1 years, and around 60% of parks have been completed or are expected to be completed in between 4.5 to 7 years. Moreover, 2 of the 32 projects managed by IL&FS have been cancelled. The cancellations occurred while the projects had disbursed only 30% of the planned MoT grants (corresponding to milestone two of five in a typical park timeline), and these projects had been approved during the most active approval year, in FY 2006-2007. Two other projects were cancelled among the 8 non-IL&FS parks, both after 10% of committed grants had been disbursed.30 But even adding those, this translates into a loss rate of less than 3% so far for what might be viewed as public grant “venture capital”. 29 Desmet, K., Ghani, E., O’Connell, S., & Rossi-Hansberg, E. (2012). The Spatial Development of India. Policy Research Working Paper. The World Bank. 30 One was managed by SREI Capital Markets Ltd., the other by CS Architects Pvt. Ltd. Both were cancelled 17 Figure 6a: Map of existing parks, approved between 2005 and 2010 Figure 6b: Map of newly approved parks (2011-2012), to be commenced in the 12th Plan 18 In interviews the PMC attributed variation to the complexity of the project and the speed of obtaining clearances and financing. Environmental clearances in particular were said to lengthen completion time by approximately a year. The data itself shows a small positive correlation of average completion time with total cost, but no significant relationship was found with the size of businesses in the park (proxied by average land area per entrepreneur) nor with the size of parks (in terms of land area). There was some variance across states, with Gujarat, Karnataka and Andhra Pradesh faring best, and West Bengal worst. However, the statistical evidence is relatively weak given the small number of parks. Moreover, variations within states were at least as wide as those between states. For example in Tamil Nadu average completion time was close to the national average (6.7 years versus 6.4 years in the 32 IL&FS parks), but ranged from 4 to 9 years within the state. However controlling for states did not yield any additional significant results in terms of correlation between completion time and park attributes such as complexity (project cost), size (in acres) or the average size of businesses in the parks. Exhibit 2 summarizes this time-to-delivery data. While these gestation periods may seem long in some contexts, for industrial parks in India this pace of delivery is at the least respectable. There is a common perception that the speed of implementation, of “getting things done”, is a constraint on India’s growth, and long project timelines are common for general as well as industrial infrastructure.31 Moreover, there is some evidence that the early years of the program represented a natural learning period, as stakeholders adjusted to an innovative policy and its arrangements. With the stakeholder system having achieved learning by doing, there are some indications momentum is building. The policy will be extended commencing this year, and 45 Detailed Project Reports (DPRs) are already on the Ministry’s table. The strongest evidence for the SITP’s ability to avoid type (i) and type (ii) failures, though, come from a comparison with o ther industrial park schemes attempted in India in the last few years. First, and probably the highest profile such attempt, was the SEZ scheme. This was launched to much fanfare in 2005, with the passage of the SEZ Act. Being enacted by national legislation, with wide coverage, and widely supported by public land acquisition, it attracted vastly greater interest than the SITP: 580 SEZs have been approved, but only 380 notified (i.e., land has been acquired), and of those only 124 were operational at the time of writing, a conversion ratio of 21%. The vast majority of the operational SEZs were IT office parks. In manufacturing, only 44 non-IT SEZs were in operation in mid-2011. This is roughly comparable to SITP parks, despite the significant fiscal and regulatory advantages offered by SEZs, the much wider sectorial scope of the policy and the broad support it has received from both states and center (including again through cheap government land).32 Moreover, while SEZ approvals have been declining rapidly, and private sector interest waning, interest in SITP parks is only growing, despite the absence of tax breaks or similar instruments. Exhibit 3 summarizes comparative figures between realizations under the SEZ Act of 2005 and under SITP. Previous textile cluster schemes were also less successful than SITP, despite offering sometimes much higher levels of support. In those, the states or their agencies were responsible for most implementation steps, from location selection (1) to dimensioning (4), and from infrastructure provision (6) to park operation and maintenance (8), with monitoring (9) remaining mostly with the central government. TCIDS33, which offered assistance of 50% of costs capped at Rs. 20 crores, saw 31 12th Five-Year Plan, Government of India. 32 World Bank (2012) 33 Textile Centers Infrastructure Development Scheme 19 some projects being completed, i.e., somewhat avoided type (i) failure, but most were small and brownfield, and many were perceived to be subject to a type (ii) failure, i.e., lack of demand. A more recent scheme was APES (Apparel Parks for Exports Scheme). This targeted green field and larger parks. Location selection (1) was shared between state and center, dimensioning was pre- decided at between 150 to 250 acres (4), and land provision was the explicit responsibility of state governments (3). Despite offering grants for 75% of infrastructure costs (vs. 40% under SITP), APES saw only 2 projects approved: one in Tirupur, Tamil Nadu, where the industry association de facto took over implementation;34 and one in Ludhiana, Punjab, which took more than 10 years to develop. A similar pattern of results may be identified in other sectors. For example, the Ministry of Food Processing (MoFP) also had a cluster scheme, the Food Park Scheme (FPS), which was a centrally sponsored scheme, i.e., funding was flowing to the states that would then fund the scheme locally. The result was that financing (step 5) became even more problematic than when the Center was directly playing this role, leading to the failure to build parks at all (type (i) failure, and therefore all other types of failures). In contrast, another central scheme for industrial parks that has made similar design choices as the SITP has been widely perceived as effective at avoiding type (i) and type (ii) failures. Like SITP, the Industrial Infrastructure Upgradation Scheme (IIUS) requires compulsory formation of an SPV to drive the process and appointment of the equivalent of a PMC for all-encompassing support. IIUS can only be used for existing parks, i.e. brownfield projects, so the challenge may be lesser. In addition the IIUS only took on SITP-like features in 2009, so the case has been studied in less detail. However, the IIUS applies to any sector, not only textiles, which provides some support to the hypothesis that role selection is crucial in avoiding type (i) and type (ii) failures, whereas the sector to which this applies is much less relevant (with caveats explained below). Annex 3 provides more details on IIUS. Early indications in firm demand, investment and jobs Returning to the SITP, the data indicates that parks are not only built but are realizing investment. Long-run target investment in the 40 parks is approximately Rs. 20,000 crores, or $4 billion.35 Among IL&FS parks36, current investment has already reached 38% of the final target. In absolute terms, current investments represent 4 times grant disbursals, which is twice the policy’s objective. Progress towards targets is not related to the average size of plots in the parks. This indicates the scheme has managed to spur extra private investment, beyond park infrastructure, across the SME size spectrum. Further, progress towards investment targets does seem positively related to additional grant disbursal, indicating that private and public investment are being sequenced together. Seven parks already show a ratio of private investment to public grants in line with the final targets. Exhibit 4 provides detailed investment data and statistics, while Exhibit 5 shows the relationships between investment figures and ratios with the percentage of government grant released, park project costs, average plot size and target investment/grant ratios. 34 This park was part of the field visits conducted as part of the study, and several entrepreneurs were interviewed. Details are provided in Annex 2. 35 This corresponds to around $2 million per entrepreneur per park on average, which might seem an ambitious target and if reached would mean the SMEs in the park would eventually graduate from being officially classified as “medium”. Reaching half that target would still be a considerable achievement and does not seem out of line with investment rates observed during field visits. 36 Excluding the two parks cancelled among IL&FS ones, the sample analysed here 20 Job creation and production will lag investment, and so the data can give only limited indications at this stage. The program’s target is 300,000 jobs across all parks37, as well as 350,000 indirect jobs. This translates into a target of 230,000 direct jobs for the subset of parks managed by IL&FS. In those parks 42,000 jobs have already been created, or 19% of the end-state target, which is promising given most of the parks are only now commencing production. The crucial question remains whether (or how many) investments would have been made (and jobs created) in the absence of the policy and the public money spent, in other words, the question of “additionality”.38 This might be posed locally, i.e., whether local entrepreneurs would have expanded anyway, or nationally, e.g., if investments which appeared new locally resulted in fact from cross- state raiding of firms. Unfortunately, the data available is insufficient to serve as a means to probe this question at the moment. The authors are not aware of an analytical technique that would yield a rigorous answer, even with a more complete dataset. As such, anecdotal evidence can, at this stage, only provide indications. In that regard, discussions with IL&FS and entrepreneurs suggest however that around two thirds of capacity in the park is indeed additional. While there would be an incentive for them to report this, individual entrepreneurs indicated, under repeated probing, that they would not have expanded or created their firms without the scheme. Some were already active in upstream parts of the value chain, such as machine sales, and entered the downstream segments through a park they came to know. Others had existing sites which had reached capacity and for which they said they had been struggling to find suitable expansion space. The machines visually observed in the park were new or of recent vintage, from high-end capital goods producers in Europe and the Far East. This was noteworthy in the weaving segment in particular, since textile weaving in India is little mechanized versus comparator countries, an acknowledged constraint on the sector’s growth. Nevertheless, we are conscious this does not amount to proof that new investment would not have happened anyway, though it should be noted much of the investment, in both parks and machinery, took place in a period of falling macroeconomic investment rates, particularly in manufacturing. Private investment growth fell 6.2 percentage points and manufacturing growth 11.8 percentage points in the period39. In the same timeframe the textiles sector was widely reported to be undergoing difficulties, with reports of rising defaults and struggling exports. In terms of public-private investment ratios, early indications are that the grants are relatively efficient. Rs. 7.4 lakhs (~ USD 15k) have been disbursed so far per job created in IL&FS parks (though no jobs have been created yet in 7 of these 30 parks), and, if the final targets are met, the final cost would be Rs. 1.1 lakhs (~ USD 2k) per job. Similarly, if the targets are achieved labor productivity in the IL&FS parks would be Rs. 20.3 lakhs (~ USD 40k), compared with estimates of approximately Rs. 1.2 lakhs for the Indian textile & garment sector in 2005-2006, the most recent figures available.40 While, as mentioned above, the parks seem on track to meet their jobs targets, the current absence of production data prevents an analysis of whether the productivity targets are similarly in reach. One interesting observation is the especially high productivity target for Gujarat 37 The approximation remains the same whether the 2 IL&FS parks cancelled are not accounted for or not 38 Formally, transfers of capacity into the parks are forbidden. 39 Authors calculations from World Bank and OECD national accounts data for the period from fiscal year 2005/2006 to fiscal year 2011/20112. Private investment is measured as private gross capital formation in current rupees (the data was not available in constant rupees), while manufacturing growth is measured in constant rupees with base year 2004/2005. 40 Economic Services Group, National Productivity Council (2010). Productivity & Competitiveness of Indian Manufacturing Sector: Textiles & Garments. Report submitted to the National Manufacturing Competitiveness Council, Government of India. 21 parks (Rs. 42 lakhs per job on average, more than twice the average target), in line with high capital intensity of Gujarat’s economy, and the comparatively low number for Andhra Pradesh (Rs. 8 lakhs per job) which accounts for a large number of the jobs to be created, but which also comprises the two largest park projects. Exhibit 6 summarizes employment, production, grant efficiency and related data for IL&FS parks. Of course, translating current trends to meeting targets requires sustaining demand beyond the first wave of entrepreneurs’ investments. While early demand generation required substantial marketing by the Government and PMCs, the apparent success of the scheme has had a strong demonstration effect and the scheme now faces excess demand, as evidenced by the number of EoIs and DPRs for the next phase of the scheme. Moreover, while the early parks were concentrated in States that have historically had more developed textiles industries, the next phase looks set to achieve a broader coverage (see Figure 6a and Figure 6b above), though the impact of this may be reduced if it esults in cross-state raiding. Overall, though, allowing the scheme to succeed first in states with initial advantages has facilitated rather than held back the later spread of the scheme to lower income areas. IV. EARLY RESULTS: CLUSTER EFFECTS AND POLICY SPILLOVERS The previous section analyzed data on SITP performance to date in avoiding failures of types (i) and (ii), in the failure framework presented above. The present section presents the suggestive but consistent evidence gathered in the field and inferred from available data on whether the scheme is avoiding pitfall (iii) (lack of cluster effects or agglomeration economies). It also discusses the potential for, and safeguards against pitfall (iv), negative spillovers or side-effects. Agglomeration economies There have been suggestive indications of scale efficiencies in “hard” inputs, though across some parks and not others. The first of these relates to environmental compliance. Since the cost of Common Effluent Treatment Plants (CETP) can be included in project costing, some entrepreneurs indicated that, together with scale economies, it had altered the balance between the cost of complying with pollution regulations, versus the risk-adjusted cost of being caught.41 The several CETPs visited in SITP parks in Gujarat and Tamil Nadu were all of high standards, with advanced effluent monitoring and testing facilities, surveillance cameras on all critical effluent circulation and discharge areas and usually a complementary solid waste treatment facility. This could be contrasted with the Ministry of Environment and Forest (MOEF)’s CETP scheme, which also provides grants for CETPs, but on a more traditional model requiring multiple State and Central clearances and interventions.42 As such, it has had relatively limited uptake and has shown limitations in encouraging the building and maintenance of modern, efficient facilities.43 Secondly, electricity access, a perennial issue for manufacturers in most parts of India, was more easily alleviated by interacting with the State government through the SPV structure, and with the 41 Although this was not true for dyeing and processing in Tamil Nadu where the requirement for expensive Zero Liquid Discharge (ZLD) still made it prohibitive for the most polluting industries such as processing to comply, even within a park. 42 The IIUS, mentioned earlier, can also be used to fund CETPs, though only in existing industrial parks, and now uses an SITP-like design. 43 The conclusions of the most recent report of the Central Pollution Control Board (CPCB) of the MoEF can be found at: http://www.cpcb.nic.in/statusCETP.php. The MoEF has acknowledged the deficiencies and the CETP scheme is now being modified for the 12th Five-Year Plan to require more SITP-like features such as constitution of an SPV for the users, though it retains a large role for States, especially on funding and land acquisition. The revised policy can be found at: http://envfor.nic.in/downloads/public-information/revised-cetp.pdf 22 cumulative “weight” of entrepreneurs in the parks in terms of employment and taxes. This was observed, for example, in Tamil Nadu, where electricity shortfalls were relatively more acute than in other states. There all of the SITP parks visited had managed to negotiate with the local distribution agencies a dedicated power distribution station and specific power agreements (however, this does raise a risk of negative distortions, which will be discussed further below). In other areas the support of PMCs was again important in terms of providing technical advice to SPVs, for example those which decided to set up their own captive power plant (as in two of the SITP parks in Gujarat). Turning to the effects of park implementation on “soft” infrastructure, there is some evidence that cohesion between entrepreneurs has been nurtured in the parks, with particular effects for self- resilience. The principal reasons parks could still fail after being built and filled include them not being maintained properly; entrepreneurs finding a way to pocket the grant money and leave; or a lack of flexibility for transfer of ownership in case an entrepreneur leaves. The first is unlikely given the SPV’s structure. On that, and on the second, interviews suggested that the significant amount of money already invested by each entrepreneur in the park (through the SPV and individual investments already made) was a sufficient incentive to keep effort levels high and prevent fleeing. In interviews it was also consistently mentioned that entrepreneurs facing difficulties benefited from joint support, from the indirect effects of the park’s financial structure, and as importantly from mentorship and other forms of assistance from more capable members. While there might be a risk of this allowing inefficient firms to survive, the instruments seemed more targeted at temporary, liquidity-type issues (such as in the case of late payment of a large order). Further, the other members would have a strong incentive not to maintain low performing firms in the park, due to the damage it could do to the reputation, cohesion and financial health of the SPV. Barriers to exit might then be problematic, but several parks reported that they had some form of flexibility to sell the shells of entrepreneurs and to transfer SPV memberships (and one had already done so). Access to, and cost of finance were said to have improved thanks to the pooling and credibility effect of the SPV structure, and the support received from PMCs on designing business plans and structuring the financing.44 Though bank financing still remained a difficult process, it was perceived that the scheme had made it possible where it was not before, and the support from PMCs was seen as particularly important for small to medium size firm entrepreneurs who lacked the adequate financial skills and wanted to “focus on their business”. Further, entrepreneurs in all the parks visited said grouping with peers had given them the confidence to invest more for expansion and upgrade. As mentioned above, machines in the parks visited were mostly recent or new ones, and even in the most modest units of the medium-sized Palladam weaving park in Tamil Nadu were highly automatized, high-quality ones originating mostly from Europe or Japan. From the interviews conducted, this investment behavior seemed likely due to the “risk- pooling” effect of regrouping in the parks, e.g. the ability to share or delegate job work or even sell or buy machines from “neighbours”. The sharing of knowledge among entrepreneurs was also evident, and might have worked among workers too through the presence of common training facilities. In some cases this has also prompted competitive poaching, with workers more easily reallocated to more efficient users, though other parks have attempted to prevent this.45 Branding may also have 44 See again how IL&FS helped resolve the issue of common repayment of loans, as mentioned in section III and illustrated in Figure 3b. 45 A 2011 study by Grant Thortnton for FICCI mentions the case of one park member offering higher wages than others and disturbing less profitable members. To avoid conflict, in some Parks, members agreed upon a fixed salary to be offered to a new entrant for first six months and not to poach each others workers. 23 been made easier by the grouping into parks46, and with improved logistics has helped access new markets, including internationally, especially in parks focusing on one specific part of the value chain (such as yarn in Surat, Gujarat, and weaving in Palladam, Tamil Nadu). However, these effects in particular have been highly variable. In some parks the entrepreneurs were proactive in pursuing initiatives such as common purchasing, common order bidding, and common marketing. In others, the entrepreneurs were more passive or reactive, and such initiatives were less advanced, in several parks being an expressed wish which had not been acted upon. Overall, there is uneven but substantial evidence that the scheme is generating agglomeration economies, though heterogeneously across parks. We postulate that some of these economies are driving the high investment/grant ratio observed, through increased risk-taking and modern machinery acquisition. If valid, agglomeration economies should also result in firms in the scheme substantially outperforming the rest of the sector in productivity, i.e., would result in meeting or at least approaching the scheme’s very ambitious productivity targets. However, the unevenness of these effects across parks indicates some scope to do more: the scheme, through the SPV and its tasks, brings disorganized firms into a coherent group, creating a “soft” platform which can be built upon (more so than when firms simply take up space in a park built by the state or by a developer). We will return to this theme below. Negative side-effects on investment climate outside the park (“negative spillovers” and “crowding out”) There remain the risks of type (iv) failures, negative externalities. The design of the SITP consciously attempts to avoid its parks becoming instruments of “fiscal wars” that result in a “race to the bottom” among neighboring States, the most common and easiest spotted form of negative spillover. There is, though, another kind of negative spillover, best understood through the interplay of “voice” and “exit” by firms, with respect to the business environment.47 The “exit” of industries from a general, difficult economic environment into the “privileged”, but limited space of parks may tend to reduce their willingness to raise their “voice” (a costly action) to demand broad-based improvements (which would now only bring limited benefits). The effect may be strengthened if the first firms to take advantage of the opportunity to exit are the most capable of seeing and using government programs, and therefore perhaps the very firms that would be most effective in lobbying. If authorities are not naturally very sensitive to firms’ concerns, the result could be the stalling of broad-based reforms to improve the investment climate. This risk may be mitigated if firms can move easily between many parks; and authorities are sensitive to such movement; and firms in the park still depend on firms outside them. Then firms inside and outside the park may move if the environment around them is weak, forcing a response. But if any of these conditions do not hold, for example if the firms in the park source from each other or through dedicated imports, then the result may have been simply to remove the relatively more effective advocates of reform from the political landscape. Therefore the design of industrial park programs, including variants such as regulatory and tax zones, should be evaluated including these potential negative spillover effects. This may provide a word of 46 These initiatives can benefit from the collaboration established with PMCs. For example, IL&FS has helped the SPVs it advises choose a specific brand name and logo for their park. 47 Hirschman, A. O. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Harvard University Press. 24 caution regarding such zones: they provide a much higher degree of “exit”, since they typically offer, in addition to privileged infrastructure, lighter regulations across the board and large fiscal incentives. On the other hand, when successfully implemented and scaled up, such as in China, such zones might offer the possibility of moving between them, creating a “race to the top”. To our knowledge, this conjecture has not been the subject of a specific study, but finds support in a range of related work, for example on the pressures to learn how to use SEZs in the early reform period of the 1980s, and on the ‘point’ to ‘surface’ model of experimental governance, including its application to ‘science and technology’ parks.48 However, this conjecture depends on the way in which sub-national authorities there have been made sensitive to the exit from their province of high capability firms, and so have sought to use such zones to learn, rather than to provide an easy means to placate louder voices without undergoing more substantial reform efforts. This in turn depends on local governments’ heavy fiscal dependence on their local industrial bases.49 In other words, because such governments fear exit, and the parks are not sticky enough to preclude it, local governments are strongly incentivized to both make the parks effective and to learn from them. This may be contrasted with India, where one may speculate that the large-scale exit of India’s IT firms into SEZs over the previous three years has contributed to their diminishing pressure for broader reform. The authors are not aware of specific research that rigorously tests this hypothesis, though it has found anecdotal support in conversations with policymakers in India, and may be a fruitful area for future research. In this regard, the SITP again provides some advantages, as it provides no regulatory exceptions. It promises to create a large number of parks, increasing the possibility of exit, though the SPV may create stickiness for one firm in its original park, at least in early years. Most parks being of moderate size (a point to which we return further below), their members are still dependent on the surrounding environment, and do not have an exit even from “trunk” infrastructure. 50 Financial support is significant, but still leaves most of the financing to be raised by the entrepreneurs themselves (whether it is through further grants, loans or equity), so they do not have full exit from the financial sector either. However, there may be a countervailing effect, which creates a long-term risk for the program’s impact. This risk is that firms in the parks become highly organized and efficient at obtaining preferential treatment, effectively turning the parks into instruments of lobbying and capture. There is even a risk that this is occurring, for example, in the Tamil Nadu parks’ ability to secure electricity. If in doing so they deprive more productive users of electricity, this would be a substantial negative externality. While this cannot be tested yet, it bears watching in the future, particularly as more parks become operational. More generally, as described above, the SPVs are a vehicle for collective action, often formed and lead by highly capable firms. They could become too effective at local lobbying, resulting in capture 48 On the early reform period, see Vogel, E. F. (2011). Deng Xiaoping and the Transformation of China. Cambridge, MA: Belknap Press of Harvard University Press. For ‘point’ to ‘surface’, see Heilmann, S. (2008), “From local experiments to national policy: The origins of China's distinctive policy process”, The China Journal, 59:1 -30. 49 Byrd, W. A., & Lin, Q. (1990). China's rural industry: Structure, development, and reform. Oxford University Press. 50 Large, linear infrastructure, such as transmission lines, national highways, and freight railway. The presence of quality highways was explicitly mentioned as key to the viability of the projects in many cases, since it allowed locating parks far enough from the city to find land at a reasonable price, while allowing easy commute to it so as to benefit from agglomeration economies (including proximity to markets and to qualified labor). This has been justly pointed by one of our reviewers as encouraging the development of secondary cities, and therefore also putting additional pressure for planning their development. 25 of local schemes and increasing distortions in favor of the parks. On the one hand this risk is inherent in all successful cluster development programs; on the other, the small size of the firms involved in this scheme may mitigate it. This in turn depends somewhat on the structure of the textile industry, and the size of the grants for the parks (treated further below). It is however another risk that needs watching, especially when adapting the scheme to other industries and contexts. For example, very large parks, or parks with very large firms, would create particular concern in this regard. Overall, the SITP encourages entrepreneurs to build a new mechanism for voice, through organization into the SPV, and their connection with the PMC. It is far too early to tell if these will be effective in generating positive rather than negative spillovers. On the whole, however, the balance of risks seems neutral to positive, as it does for type (iii) failures, and more favorable than more common schemes. V. EXPLAINING RESULTS: SITP DESIGN & THE POLITICAL ECONOMY OF IPs In sum, the early data indicates the SITP has managed to avoid type (i) and type (ii) implementation failures (getting parks built and occupied), while providing some positive early signs on avoiding type (iii) and type (iv), generating cluster effects and avoiding negative spillovers. In this section we will attempt to understand this performance in terms of the mapping of implementation responsibilities (1 to 9) to the various stakeholders, and how that interacts with the political economy in which the parks are built. Our principal contention is that the key achievements of the SITP design are to fit those roles to the incentive structure of that political economy so as to leverage existing but underused capacity (firms using the parks) as well as new capacities (PMCs), and to allow necessary flexibility while limiting the opportunities for collusion. This design created a latent potential which had to be triggered to become actual, through demonstration effects. That in turn was provided by the MoT and the first PMC’s focus on organizing capable firms, rather than organizing parks. Engagement process and entrepreneurs’ cohesion The initial phase of the scheme involved intense prospection on the part of the MoT and IL&FS-CDI, which had been chosen as sole PMC at the time. This engagement process was especially critical as the scheme was to rely heavily on the entrepreneurs and their ability to drive scheme implementation, from the very start. The objective was therefore to identify the entrepreneurs that had most demand and at the same time were most capable of organizing themselves to achieve that goal. The idea was to generate enough awareness in key dissemination points, and then let self-selection operate as willing and capable entrepreneurs would have an incentive to regroup and follow the most capable “leader”, while keeping less capable ones at bay. IL&FS explained that the most important quality of that entrepreneur was the ability to generate cohesion within the prospective SPV members. In practice the procedure was as follows. First, IL&FS and the Ministry would identify in which states the main “organic” clusters in the relevant sector were located (e.g. textiles for SITP), and within those states which were the principal industry associations and chambers of commerce. Next, they involved the local offices of the MoT and the state government’s Ministry of Industry. Together they would focus on key dissemination points (industry associations, chambers of commerce etc.) and make presentations of the scheme three or four times to different groups. There would always be a Project Manager (of an existing SITP park if any, or a prospective one) at presentations, who could later bring interested entrepreneurs to project sites to see the infrastructure and talk to entrepreneurs, if the park was already built. Finally the PMC would offer capacity building on what it takes to make a successful park to those among the prospective entrepreneurs who confirmed their interest. This capacity building would focus on the processes of organizing and coordinating within the cluster. 26 This was clearly a time intensive process. Here the PMC was vital. Since the PMC would only receive fees if entrepreneurs took up the scheme, and only if those were capable enough to successfully organize others, it had a strong incentive to find capable entrepreneurs and generate interest. Since the first PMC, IL&FS, had a long history of working with both Centre and State, and is partly government owned, it was well placed to be a trusted partner in the initial process, and to work the informal relationships needed among the various levels of government. At the same time, the importance of the textile industry in jobs and investment, as well as the need to find new solutions after prior schemes, likely played a role in the MoT’s drive to initiate the scheme, and State departments to cooperate in its launch. Once the scheme was visibly yielding results, excess demand in new parks has allowed the MoT to later avoid most of these efforts, except in some new, less advanced target states. Such excess demand then provides some confidence in the selectivity of the engagement process with entrepreneurs, again the most crucial aspect of the scheme. On the other hand, it might be feared that the first entrepreneurs to take up the scheme were also the most capable, and therefore that SPV quality could deteriorate as the scheme expands. This will merit close monitoring.51 The cluster effects observed in the previous section (the suggestive avoidance of pitfall (iii)) probably find their strongest spur in the initial cohesion-building exercise. This involves not just the initial formation of the SPV and the initiative to apply for the scheme, but also the practical and difficult process of selecting and purchasing land as a joint effort, including a monetary one, as the first grant installment is made only after monies have been paid to purchase land and title obtained. This high “initiation cost” may have a positive effect on the willingness of entrepreneurs to keep organized to go through the rest of the difficulties of the park development process, possibly limiting free-riding.52 Overall, though, it should be asked why firms would undergo the process in the first place. A general answer would point simply to the local failures of infrastructure provision in India, so that firms’ alternatives to set up may suffer high congestion costs. There is evidence that such costs are pushing firms in India to move out of urban areas earlier in the development process than in comparator countries,53 and in interviews this was often cited as a primary motivator for locating in a park. On the other hand, the parks only partly solve this constraint, given their continuing dependence on trunk infrastructure. As such, other factors must be at work. One conjecture is that the firms see the SPVs as ways to reduce the individual transaction costs of obtaining land, by pooling the formal and informal risks and costs of doing so. Several firms did report this as a benefit of the scheme which they valued. On the other hand, if this were paramount, we would expect to see fewer parks the better functioning were local land markets, and vice versa. In contrast, Gujarat, widely seen as having the least difficult industrial land market in India, has the most parks; and West Bengal, with a notoriously difficult environment for obtaining 51 Another aspect to watch is the impact of location decisions on urbanization. There seems to be some evidence that industrial parks are spreading out to other cities, mostly secondary cities or other states, but not to any city or any other state. For example many projects were launched in Gujarat in the first phase, and much fewer in the current phase. The tension between agglomeration benefits and congestion costs (in particular land costs) for location choices has in fact been explicitly voiced by most of the entrepreneurs interviewed, in parks but also outside of them. For further discussion of this aspect of urbanization, see Ghani, et al. (2012). 52 Hirschmann (1970). 53 See, as well as Ghani et al (2012) above, an extensive literature including: Lall, S. V., Shalizi, Z., & Deichmann, U. (2004). Agglomeration economies and productivity in Indian industry. Journal of Development Economics, 73(2), 643-673, as well as the in-progress “India Urbanization Review: Nurturing Metropolitan Economies and Connecting Peri -urban Areas”, World Bank, New Delhi. 27 and converting land, has only one park, and that has taken the longest of all in the first wave to become operational. A more general comparison between states’ ratings on proxy indicators in Sub - National Doing Business indicators for India, from 2009, and the number of parks, reveals a similar pattern: there are more, rather than less parks, in States with better functioning land markets. So, while the limited data does not allow establishing significant relationships, it seems parks have been developed in greater numbers and fastest where there was a higher density of capable firms (Maharashtra, Tamil Nadu, Gujarat). On the other hand, there may be hidden threshold effects, so that the scheme is not effective at reducing costs in highly dysfunctional land markets, but is effective in moderately dysfunctional ones. More generally, we would return to the fact, mentioned above, that the participating firms are not homogenous, but may be distinguished between “lead” and “follower” firms. The scheme likely answers different needs for these different categories. Principally, it was clear to us that these “leaders” (and usually larger firms) had been motivated ex ante by the perceived advantages in clustering, and relatively less so by the subsidy and access to quality infrastructure (which they were more likely to be able to afford on their own). The “followers” that then joined were often smaller firms, relatively more attracted by the possibility to benefit from the initiative taken by the leaders, as well as the possibility to avail high quality infrastructure and obtain land for their own business at a relatively low cost. Roles and responsibilities We now turn to the incentive structure of the scheme. Its design creates a set of interdependences between the GoI, the MoT, states, PMCs and entrepreneurs (through SPVs). These are reinforced by institutional mechanisms such as regular monitoring reports sent by the PMC to the MoT; quarterly reviews of the overall scheme by an Apex inter-ministerial committee at the GoI level; and the appointment of directors on the SPV board by the MoT (through its local representatives), the PMC, and the State. A simplified diagram of this structure is represented in Figure 7 below. We consider, in turn, how this structure fits the capabilities and incentives of each actor to their respective role. Figure 7: Simplified diagram of SITP incentive structure Basic Incentive Structure The overall incentive structure fits the political economy of relationships between stakeholders Quarterly Scheme Reviews APEX Inter-ministerial Appoints, approves Ministry of Textiles Committee Sends monthly progress DPR, pays fee; high- reports; relays SPV level decisions from complaints monitoring reports Checks bottlenecks in clearances when Reports major issues Appoints a director raised by PMC/SPV (through local MoT rep) on the board PMC State Reports daily Lobbies for additional issues support (connecting roads…) SPV Ground monitoring Provides Clearances (daily contact, weekly park visits) Appoints a director on the board Appoints a director on the board 28 First, the central government is the main grant provider. In theory, despite not holding an absolute majority stake in the SPVs, it might be tempted to interfere in other aspects of project implementation. In some cases elsewhere, for example, political considerations have been known to induce sub- optimal location selection. However, by virtue of India’s size and diversity, bureaucrats in central Ministries have fewer incentives for such micro-targeting than State officials. The small size of the parks and the grants, in absolute terms, reinforce this bias, lowering incentives for location interference and increasing incentives for performance across all the parks. In terms of capabilities, a central Ministry rarely has the capacity (both in terms of local knowledge and manpower) to monitor at a micro level, so the devolution to a specialized agency (the PMC) allows the MoT to both focus on its core strategic mission and obtain better micro data. In this regard, the crucial instrument in the SITP policy are the clear Terms of Reference (ToRs) for PMCs, who must send monthly progress reports, with a uniform, simple format that helps the ministry take rapid decisions. Strict rules and methods for grant disbursements limit the abuse of funds in theory (though we discuss this at further length below). Annexes 1a, 1b show the format of progress reports and Annex 4 the utilization certificates that condition the release of installments of grant money, as specified in the grant sanction orders (a sample of which is given in Annex 5). Turning to the states, they can (and typically do) also participate in grant funding, but the limit on their participation in SPV funding is adapted to their lower funding capacity (compared to the Center) and somewhat limits grant wars between states.54 Therefore they have a limited direct role, in step (5), but together with their ability to appoint a director on the board of the SPV this often proves enough to induce them to be engaged with the projects. This is critical as states are ultimately responsible for providing necessary clearances, in particular for land (principally, conversion of agricultural to industrial land). On the other hand, the relatively low amount of the State grant, such that the project often remains viable even in its absence, means State politicians and bureaucrats will most often lack the leverage to distort location decisions. In addition, the SITP is relatively well adapted to the political economy of State-Center relationships. Since State and Centre contributions are separate, not subject to one another (in contrast, for example, to the Food Park Scheme mentioned earlier), they become less likely to be subject to disputes resulting in delays. The burden of engagement and coordination with various State agencies is placed not on the central Ministry, but on local entrepreneurs, organized through the SPV and supported by the PMC’s experience, skills and contacts. Local entrepreneurs will in most cases be substantially better positioned to navigate the political economy of their State than a central Ministry, and it is they who are tasked with this by the scheme. As a last resort, recourse to the center remains possible for entrepreneurs in case of major issues such as long delays in approval, which does maintain some pressure on states to perform. In short, rather than resting on State-Centre coordination for the multiple routine processes needed to build the park, the scheme leverages the latent capacity of local entrepreneurs, but provides them implicit leverage through, in extremis, recourse to the Centre. As a result, the scheme, though having a uniform national design, adapts itself in its implementation to wide variations in local conditions. For example, the political incentives to develop industry in Gujarat are very strong, and GIDC 55 is recognized as one of the most efficient state industrial 54 As mentioned earlier, combined contributions of states and other institutions to the SPV capital must be such that entrepreneurs retain at least 51% share of the equity part. In the typical case of a 40% (and in any case less than Rs. 40 crores) central government grant and 40% term loan, this means a limit of 9% of project costs (and less than Rs. 9 crores). 55 Gujarat Industrial Development Corporation 29 development corporations, while in Maharashtra the environment is often perceived as much less favorable on both counts. However, almost as many parks have been built or are being built in both, with limited and offsetting gains to each in time to completion, investment and job creation (cf. Exhibits 2, 4 and 5 for detailed figures). We would note however that both states do have a substantial base of capable local entrepreneurs. The scheme design seems then to shift the variable for success from a State’s political economy or the quality of coordination between it and the Centre, to the capabilities of local entrepreneurs. Those entrepreneurs’ capacity is, crucially, also augmented by the PMCs. The effectiveness of PMCs of course depends on the thickness of the local consulting market: consultancies capable of performing these role must be present, and the more of them there are, the higher the bar for their selection can be set while mitigating capture risks. The all-encompassing role of PMCs reduces coordination issues and increases the ability to deal with uncertainties as the project is being implemented. It is noteworthy that IL&FS has had both an advisory role for the government (“MoT-PMC”) and an execution role for the firms (“SPV-PMC”) in 80% of the parks it is PMC for, and the ratio was said to be similar for other PMCs. Although there are risks to such integration (examined in the last part of this section), the advantages are increased flexibility in adapting design as implementation proceeds, and the removal of one agent in the execution chain. The PMC can then have few excuses for delays executing plans it has devised. The PMCs also seem to effectively intermediate between the government and entrepreneurs. Their incentives are partially aligned with both: to remain empanelled, they must demonstrate their performance to the Ministry by getting parks built and functional in a timely manner, while they are ultimately chosen for each park by the entrepreneurs themselves. They do not though share profits with the entrepreneurs and do not have a formal interest in which parks or states perform better (though the potential for informal collusion is dealt with below). Their intermediation between central Ministry and local entrepreneur has benefits both practical and political: it enables the Ministry to oversee a large number of parks by monitoring a small number of PMCs; it overcomes potential distrust between private sector and government; and creates distance, so the government does not negotiate directly with individual entrepreneurs receiving grants. Overall this means PMCs are key to the potential for the scheme to learn about itself, and to foster learning among the firms in parks. On the former, the PMCs observe the scheme’s performance in a wide variety of contexts, and, by competing among each other, are incentivized to innovate within the existing scheme design. This has resulted, for example, in a number of the financing innovations described above. When the scheme design is refreshed, as is occurring at the time of writing, the knowledge gained by the PMCs can be rapidly consolidated. This does of course raise risks of capture and distortion by the PMCs. The Ministry’s ability to balance these risks and benefits will be critical to the ability of the scheme to improve its results, or conversely, to drop in performance. The PMCs also enhance the ability of firms themselves to learn, acting as transmission mechanisms for knowledge between widely separate firms across India. It is not clear to what extent this potential is being actively exploited at present; the potential to do so more actively would clearly form part of any more general initiative to exploit the social infrastructure built by the scheme, in this case doing so across parks as well as within them. That brings attention back to the entrepreneurs themselves. The scheme distinguishes itself by leaving the initiative and ultimate decision-making to them. The first advantage is that entrepreneurs represent the “demand” for the park and therefore are naturally best placed (in terms of incentives) to decide on an appropriate location: convenient for entrepreneurs in the SPV, for backward and forward linkages 30 and so that the price of land does not make it prohibitive. Similarly, dimensioning of the parks is naturally best discussed with those who will be occupying them, i.e. SPV members, with technical advice from the PMC. Demand and dimensioning are thus subject to verification by the ultimate users in a way costly to themselves, i.e., in a more reliable means than by the signing of an ‘expression of interest’ or ‘MoU’, or by the completion of a “demand study” (whose point estimates of the future will almost always be wrong).56 Infrastructure design and procurement are then left under the flexible discretion of the entrepreneurs themselves, subject to monitoring and other means to control fraud (discussed further below). In contrast, for example, previous policies such as APES fragmented the grants into many portions to be used for specific purposes: Rs. 10 crores for infrastructure facilities, Rs. 5 crores for effluent treatment plants, Rs. 2 crores for training facilities and so forth. Grant size, structure and collusion risk Turning from the division of roles to the grants themselves, we first note their size. The cap on the central grant of Rs. 40 crore and 40% of total cost in effect limits park to an investment cost of around Rs. 100 crore ($20m), corresponding in India to a size of around 100 acres.57 The policy originally had a target of at least 50 entrepreneurs per park, but has subsequently introduced flexibility conditional on a floor on investment commitments: the aggregate minimum investment in land, factory buildings and plant & machinery in the park is Rs. 100 crores if there are more 50 entrepreneurs, Rs. 150 crores if there are between 25 and 49 entrepreneurs, and Rs. 300 crores between 5 and 24 entrepreneurs. In practice, park size has averaged 106.5 acres, and 56.5 entrepreneurs per park. Field visits indicate that such a size is large enough to create some agglomeration economies, but small enough to allow land acquisition in most cases, to not unduly disturb the local, state and national political economy (by creating large incentives for interference, as discussed above), and to allow for a functional organization of the entrepreneurs in the SPV. Achieving this “right-sizing” of parks is probably one of the most important effects of the specific SITP grant design, reinforcing the capacity and incentives implications of the role allocation design. By contrast, many of India’s SEZs were considered too small, with only one or two firms, to result in agglomeration economies. On the other end of the scale, huge parks known as PCPIRs involve enormous land acquisition and infrastructure provision for areas of several thousand hectares.58 SITP parks seem to strike a sweet spot in this respect: modest enough to get things done and fly under the radar politically, but large enough for their SPV to be an effective local actor, and for them to generate some agglomeration economies.59 The method of grant disbursement in SITP is also important for reinforcing both capacity and incentives. The first disbursal (of 10%) occurs after final approval of the project, which itself happens only after monies have been disbursed for land procurement and the SPV can prove it has received 56 Neufville, R., & Scholtes, S. (2011). Flexibility in Engineering Design. The MIT Press. 57 Since the grant is 40% of project costs, the cap of Rs. 40 crore grant is reached for a Rs. 100 crore. Since larger projects will typically generate larger scale economies, it is optimal to design a Rs. 100 crore park. 58 Petrochemical Investment Region. See the analysis of the SEZ scheme referenced previously. 59 For example in all four SITP parks visited in Tamil Nadu, it was clear that the grouping in an SPV structure is what allowed entrepreneurs to negotiate a specific electricity feeder for the park with state authorities. Similarly in the case of the Surat Super Yarn Park in Gujarat, last-mile road connectivity was negotiated with and provided by the state infrastructure development corporation (GIDC) – thanks to the existing high quality highway nearby, this meant the park was accessible from the city even though land had been bought in an under-developed area to reduce costs. 31 ownership of the land. Given land markets in India, the initial disbursal is then a selection mechanism on the capacity of entrepreneurs to navigate the local political economy. Subsequent disbursals are made in advance after the completion of specific milestones and the full utilization of the previous disbursement. The schedule usually has five milestones, with 10%, 30%, 60%, 90%, and finally 100% disbursal on completion of the park and start of production by at least 25% of the units. By disbursing in advance, the scheme reduces liquidity constraints faced by SMEs, in contrast to schemes that disburse on a reimbursement basis; disbursing in tranches linked to milestones (including a final one that checks proper demand for, and functioning of the park), as verified by the PMC for the MoT, incentivizes and disciplines both SPV and PMC. The structure also makes it difficult to use the scheme for subsidized land speculation: grants will not be disbursed if parks are not built. The structure of the grant may though raise a concern of subversion of the scheme for personal gain, e.g., through diversion of the grants by the entrepreneurs. To date, to the authors’ knowledge, there has not been an accusation of misuse of funds or misappropriation against either the scheme or any of the parks. It is too early to rule this out in every case, and one cannot rule it out in theory. The scheme’s design does though seem to provide relatively strong controls against it, and while this does not mean informal bargains might not be made, those do not so far appear to be impacting scheme performance. To understand why this might be the case, we return to the roles and incentives in the scheme, and how those interact with grant disbursal. The recourse to PMCs as a new capacity provider and in fine as intermediaries between entrepreneurs and the government is vital to the scheme, but creates risks on both sides. First there could be collusion with the government to be selected to be on the panel of PMCs. On the other side, PMCs on the panel could bargain with entrepreneurs to be selected for projects. Both would potentially impact scheme performance. At the park development stage, in theory there could also be collusion of PMCs with entrepreneurs on progress reports and utilization certificates to share grant money instead of investing it in the parks. Conversely when PMCs combine design and “execution” responsibilities (“MoT-PMC” and “SPV-PMC” roles) they could budget more for themselves in the DPR. Finally PMCs could reduce their efforts (and save money) in park development and hide underperformance to both the government and entrepreneurs, since, as MoT-PMC, they are also responsible for monitoring. Collusion on PMC selection on the panels is limited, in theory, by the use of a competitive two-stage bidding process. Early signs are that this process is quite rigorous, with a PMC having been dropped, and only seven selected in total.60 Being on the panel in one 5-year Plan Period does not guarantee presence in the subsequent period, creating a strong incentive for performance over the period. Further performance discipline on the PMCs is provided by their competition to be selected by firms to develop a project (and then have this project be approved),61 such selection and approval being a precondition for the PMC to be paid the consulting fee by the government. Collusion with the entrepreneurs on selection is itself limited by the firms’ own profit motive and the phased disbursal process. In short, a low-performing PMC would have difficulty enticing firms to select it, as those firms would then be unlikely to receive the grants. The PMC would also simply not be paid, if its work was too low quality to result in an approved DPR. 60 More details would be needed on the exact assessment criteria to assess their appropriateness, but even a perfect list would not prevent collusion and subversion. Rather, the risk would be most affected by the incentives environment, which is what we focus on here. 61 As mentioned earlier, the DPR approval procedure includes passing first a Project Scrutiny Committee, an inter- ministerial committee headed by the Joint Secretary for SITP in the MoT, and then a Project Approval Committee headed by the Minister of Textiles. 32 There is also a tradeoff for PMCs once they are appointed for a project between the gains from colluding, overcharging or lowering efforts as described above and the risk of losing further appointments and corresponding fees if they are caught, or discarded because of under-performance. If fees are high enough and performance is rewarded in practice, PMCs are likely to choose not to engage in such activities. Of course, this depends on PMCs facing a credible risk of being dropped from the panel in case of underperformance. But this has, it seems, already occurred, as MoT removed a PMC, SREI Capital Markets Ltd,62 from the SITP PMC panel after it had suddenly “abandoned” the SPV it was supposed to support (though the cause of abandonment is not clear). Similarly, the fact that failing parks were abandoned relatively early might be indicative of the constraints on accurate reporting by PMCs.63 Finally, misappropriation is rendered practically difficult by the process through which grant money is channeled for project development. Except for the limited share of grant money that can be allocated to administrative expenses, the SPV has no direct access to grant money, which goes directly from an escrow account at a designated state-owned bank (usually the same that would provide the term loan) to the private contractor executing construction plans. This mechanism also allows verification that money is spent in the 60/40 proportion by the entrepreneurs (and other contributors) and the MoT, respectively for completion of the milestones. Maintaining the proportions (with some flexibility) along the payment schedule reduces further the opportunity for abuse of grant money. The fee payment structure for PMCs follows those same milestones, which further incentivizes PMCs to support completion of the parks in a timely manner.64 Vendor selection must be approved by the SPV board, which includes entrepreneurs, state officials and others. An underlying support to many of these controls may, again, be the modest size of the individual parks, interacting with the structure of the industry. The average industrial plot size per entrepreneur in the parks is around 6.1 acres, with a median of only 1.9 acres. Participants are small, likely without the financial resources to incentivize national agents and firms (officials or PMCs) to an extent that it would justify the risks that such national agents would face on a national scale (even if they were hypothetically susceptible). Even joining forces within a park, with 50 entrepreneurs per park on average, and the limit imposed of a 20% equity stake in the SPV, entrepreneurs would face a coordination issue – the maintenance of the secret – if attempting to defraud the scheme. Collusion would require a dozens of firms and officials to take significant risks in concert over a period of years, to misappropriate a relatively small grant. So it may be that the mechanisms that limit subversion described earlier might not have worked in a less fragmented industry, or if larger firms had been involved, if parks were much larger, or if the grants available per park were more substantial. These might prove to be preconditions to replicating the model successfully in environments more conducive to informal bargains and subversion, or at least conditions that are strongly supportive of success in other cases. The high fragmentation of the textile sector in India means individual firms may have less influence on the government, and 62 SREI Capital Markets Ltd is a large, reputable firm with a 50/50 JV with BNP Paribas Leasing Solutions, the largest leasing group in Europe. 63 The two IL&FS cancelled parks, already evoked earlier, were the Hyderabad Hi-Tech Weaving Park in Andhra Pradesh and the Shri Dhairyashil Mane park in Maharashtra. The MoT took the decision on observing the lack of progress by the SPV in moving forward after disbursement of 30% of the committed grant amount. The other two were the Vaigai Hi-Tech Weaving Park in Tamil Nadu (managed by CS Architects) and the Bharat Fabtex & Corporate Park in Rajasthan (managed by SREI Capital Markets), both cancelled after 10% disbursement, i.e. the first milestone. 64 The conclusion in not changed in the case where PMC roles are divided for advisory and execution, as the MoT-PMC will still be paid according to the grant disbursal milestones, and the SPV-PMC will be paid according to an SPV-determined schedule that would align with the SPV interest, which is also to complete the project. 33 therefore less leverage for subversion of the scheme. This might not be true even in the same value chain, in input sectors such as Man-Made Fibers (MMF), which is almost duopolistic. It might also be of concern if the scheme is used for very large parks, whose SPVs might become locally dominant, particularly if the remainder of the local industry is undeveloped and unorganized and so lacks countervailing sources of voice. VI. REMAINING PITFALLS AND THE CHALLENGE OF SCALING UP Reinforcing and sustaining cluster effects Despite the observations made in Section IV, there were many indications that cluster effects could be more actively strengthened and sustained. In some parks, entrepreneurs stated that they would like to pursue agglomeration effects further, but that liquidity pressures on the SPV had prevented them doing so. In other parks, some of the firms merely expressed interest, when prompted, in activities such as common purchasing. Various schemes at both State and Central level are available to fund such activities in many cases, indicating that either a funding scheme is not an appropriate remedy, or those other schemes could be more effectively designed, perhaps learning from the SITP itself. It could also be argued that, with firms already co-located, and receiving a grant, they should engage in these activities without subsidies, so that it is not clear what market failure remains in place. On the other hand, as mentioned earlier, it seems that the scheme design has been successful at creating strong cohesion between the entrepreneurs, by involving them from the (difficult) starting point of park development, the identification and purchase of land. This is in contrast with other park programs where firms are called to fill in a park only after its completion, and therefore would be expected to have less sense of common belonging. In other words, the means by which the scheme creates “hard” infrastructure also creates “soft”, in t he form of a platform for the development of cluster effects. In that case, it may be that some form of limited support for demonstration effects may be an appropriate addition to SPV costs in SITP, but this would require further investigation.65 It should also be investigated how certain design features, for example the format of monitoring reports, might divert efforts from developing such effects, though they have proven relatively effective at getting the common infrastructure built. An open question is to what extent the scheme will have the capacity to learn about, and improve on such design features and their effects. The prospects of expansion The relative success of the scheme has motivated its expansion, both in terms of adding new parks under SITP itself, as well as the model spreading to other sectors. Several other government entities and ministries, as well as some states, have now devised schemes that are highly similar to SITP. As mentioned above, the Department of Industrial Policy and Promotion (DIPP) already had a similar scheme, the Industrial Infrastructure Upgradation Scheme (IIUS) launched in 2003. By definition it supports the upgrading of existing clusters, and did not initially require the involvement of a PMC or equivalent. IIUS was though modified in 2009 to add a “Project Management Agency” (PMA), engaged by DIPP to carry out the financial and technical appraisal of the projects, together with a requirement for entrepreneurs to “ensure that the project proposal is of highest quality and a consultancy organization of national repute is engaged for its preparation”. 65 Bloom, N., McKenzie, D. & al. (2012). Does management matter? Evidence from India. NBER Working Paper. 34 However it has had to rely on much more generous grant amounts and conditions to generate enough demand (see Annex 3). Other sectors also now have similar schemes in various stages under their respective Ministries: plastics under the Ministry of fertilizers and Petroleum products; the Mega Food Parks Schemes of the MoFP (replaces the more traditional FPS with an SITP-like scheme); leather under the Ministry of Commerce; engineering under the Ministry of Heavy Industries; and IT-Electronics with the recently launched Electronics Manufacturing Clusters (EMCs) of the Department of Electronics and Information Technology (DEITy). In such cases, it will be important that the relative roles, incentives and grant structures that work in the SITP are maintained and that the scheme is not pushed into sectors or contexts where its key features will no longer be effective. As noted above, this particularly relates to the industry structure, the capabilities of firms, and the size of the parks. We would, for example, be cautious about the prospects for “mega food parks”. Given their importance, the capacity and incentives of the PMCs must also be kept in mind. For example, despite the competitive selection mechanism, PMCs might have weaker incentives to perform on each additional park they are awarded, once they each have “enough” parks to manage, i.e., such that losing access to new parks would lose some of its effectiveness in discipline. Moreover, if there are too few PMCs, their market might become oligopolistic. Neither risk appears likely at the moment, and in its inception the scheme only had one PMC, IL&FS, with few to no signs of non- performance by it while a monopoly. As such, the competition between PMCs, while important, may be a lesser factor than other of the schemes’ design features. The quality and robustness of PMC empanelment will be crucial to mitigate such risks. In terms of their own capacity expansion to handle a larger volume of parks, IL&FS reported that they did not expect to face a shortage of adequate recruits, with the notable exception of “soft skills” requirements. At the Ministry level, a fast and effective expansion of the scheme might also mean stretched capacities to monitor, despite the consolidating work of PMCs, and therefore a dilution of attention and ability to react to individual project difficulties. This may be mitigated by careful attention to the project pipeline, i.e., so that new waves of approvals are given only when earlier waves are nearing the completion of their construction. This appears, for example, to have been followed by the SITP. The land issue In scaling up, both for SITP itself and for other parks, an issue outside the ambit of the scheme might be the availability and price of land, for the expansion of current parks and for building new parks. In existing parks, the price of land varied significantly across parks and across states. Considering the 32 IL&FS parks, prices range from an average of Rs. 1.7 lakhs per acre in Tamil Nadu to Rs. 1.3 crores (or Rs 130 lakhs) per acre for the park being built in West Bengal. Variations between parks within each state can also be quite significant, for example in Gujarat where prices ranged from Rs. 3.3 lakhs to Rs. 59.4 lakhs among the six parks built there. Much of this variance might be attributed to the dynamics of industrial land markets in India in the last seven years, since the first parks were launched. Land in general has been rising rapidly in price, particularly land around any industrial area. In all the parks visited, surrounding land prices had increased by factors of more than 20 or 30 in the 5 to 6 years since the start of the projects, according to the entrepreneurs. This was a much larger increase than that of land prices in general in each state, and was a serious impediment to the expected future growth of the park. Many SPVs had to some extent anticipated the problem by buying typically twice the amount of land initially needed. This should be kept in mind when evaluating any industrial park program in India (SITP or not): land 35 utilization may not be a good measure of whether a park has suffered a type (ii) failure. Exhibit 7 summarizes land prices and other land-related data, first according to the state, then according to the purchase process: private (with or without state assistance) or state. On the other hand, despite increasing prices, entrepreneurs were confident that they could find land for new parks. This may again reflect the modest size of the parks. They were keen on roles (1) and (3) given to them in the scheme, i.e., having the flexibility to choose and negotiate park location and size according to their own price and viability criteria.66 Available data on existing parks is not conclusive on this issue, especially as land prices have sharply increased since the inception of the scheme. Price variability was lower in existing parks for land acquired through, or with support of, the local state authority, but the average price was not significantly lower (excluding one outlier), and average completion time was slightly higher. The highest expected completion time (9.1 years) is for the one park in Tamil Nadu where the state acquired the land for the park, in a state where otherwise completion time has been shorter than average. Similarly one of the two parks cancelled in the IL&FS sample also received state support, though the ratio of parks receiving some form of support for land is more around one third. However the sample is too small to draw any firm conclusions: only 10 parks received state support, and those are spread over 5 states, with parks of variable size and structure (e.g. area per entrepreneur). Some observations can still help think about how to treat the land issue upon expanding the scheme. First, entrepreneurs interviewed systematically dismissed any interest in placing parks in SEZs, even when pressed on the potentially large fiscal gains and regulatory advantages. Some mentioned the requirement for exclusive export orientation as a constraint, but most suggested interference from the government (state or center) as the strongest impediment. In other words, given the controversy it causes, the more capable entrepreneurs seemed to prefer the government not be involved in land acquisition. Such acquisition would also raise the prospect of the scheme being accused of being a “land grab”, as occurred for SEZs, causing a public backlash. The MoT recently removed land purchase costs from project costs to calculate grants, to avoid financing soaring land prices and to put more pressure on entrepreneurs to negotiate these prices. This move does not appear to have prevented the excess demand for more parks mentioned above. Second, early anecdotal evidence from the Ministry of MSMEs’ Cluster Development Program, targeted at Micro and Small Enterprises (hence its name MSE-CDP), tends to confirm the view that systematic state involvement in land acquisition causes multiple delays and cost increases. MSE-CDP is almost identical to SITP both in its grant design and in its role allocation, except that it makes the state responsible for the critical step of land location selection and involves it in acquisition. The result so far has been political exploitation of the land issue, with countless trials in process and rocketing costs, and a general slow-down because projects have to be processed through the whole state bureaucracy. When the state intervened for land acquisition in SITP, it was on an ad-hoc basis, and location selection (role 1) was still in the hand of the SPV, which likely means the SPV had already considered the local political economy of land acquisition. We would note in closing that the position of the entrepreneurs within the SITP seems rather unique. In public discourse in India, and in individual and group discussions, there is a general clamor from the private sector for public support in acquiring land. Prior negative experiences seem slow to dampen this. It may be that the long history of public attempts to “help” textile SMEs, including the 66 Typically depending on the type of manufacturing activity considered: close to ports if export-oriented; close to raw materials if backward integration is most critical; close to the parent manufacturer for its ancillary/auxiliary activities etc. 36 chronicle of earlier industrial park schemes, has led to firms in that sector being among the first to acknowledge that the interference that comes along with public acquisition is rarely worth the cost that is saved. This may be a mirror effect to the MoT’s own innovation as it developed the SITP. CONCLUSIONS: LESSONS AND FURTHER RESEARCH Why do some industrial park programs result in operating parks, and many not? This paper has attempted to answer this by decomposing industrial park implementation into elementary steps and analyzing how role allocation among them interacts with the political economy. This was motivated by the examination of a Government of India scheme that introduced two key innovations: leadership left to entrepreneurs on all aspects of scheme implementation; and a shift by both government and its consultants from building parks and then inviting firms, to organizing and monitoring prospective firms, which then lead the building of the park themselves. We found that not only these innovations, but how the design deployed them in the political economy, resulted in a relatively successful scheme, particularly in comparison to other industrial park programs in the same context. However, we found mixed evidence for avoiding type (iii) failures, i.e., for realizing “soft” cluster effects. Though our findings have been mostly anecdotal so far, there has been widespread agreement among SPV and IL&FS representatives alike that the key driver, or “root cause”, of current achievements has been the cohesion between the entrepreneurs themselves. Such cohesion was frequently reported to stem from the “leadership” and “vision” of one of the entrepreneurs, but was likely reinforced by the capacity building efforts of the PMC 67 . Means to systematize these effects should be considered next, for example potentially through the addition of funding to already built parks for initial demonstration effects in “soft” clustering. 68 However, programs to fund these types of “soft” capabilities are at least as prone to failure as industrial park programs, so the same innovation and care would need to go into any such addition. It is important to note as well that the MoT already engaged early in an on-going fine-tuning process of SITP, facilitated by the intermediation of PMCs, following feedback from the ground and analysis of early results. “Soft issues” have been part of the discussion, with for example prospects of including the salary of a professional manager of the SPV in project costs and of refining and intensifying capacity building exercises. In other words, the scheme shows early signs of building in the capacity to learn about itself. It remains to be seen if this will become a core feature of the scheme; whether the learning will increase, decrease or leave neutral the quality of scheme implementation and the generation of outcomes; and whether this learning will repeat regularly or be sporadic and largely ad-hoc. However, as the scheme is scaled up, changes in the policy should be careful to preserve what has contributed to success so far. This includes avoiding the temptation of much larger parks, given the importance of the modest size of SITP parks. In this context, many mid-sized parks seem preferable to a few very large ones (or a multitude of too small ones). Similarly, we would exercise caution on having the government more involved in the acquisition of land for the parks. Moreover, as the scheme spreads to other sectors, the design details will be crucial. For example, care must be taken that the industry structure and target firms (through the exact design parameters of the grants) are of the characteristics described in sections above. As importantly, the roles of location 67 As in most of our analysis, we extrapolate from our observations in IL&FS parks and our discussions with IL&FS representatives. We currently lack details on exactly how this capacity building effort is being conducted. 68 Such capacity building exercises would likely be even more critical in the case of smaller entrepreneurs, such as those targeted by the MSE-CDP scheme mentioned earlier, as was confirmed by a state official deeply involved in the scheme 37 selection and park dimensioning should rest entirely or principally on the initiative and decision of the entrepreneurs, though at the outset catalyzed by the government or the PMC or both. If grants and parks are to be much larger than those in the SITP, the reasons for this should be carefully understood. In time, though, as these schemes progress, comparisons of them to the SITP will yield a rich set of lessons. Perhaps the principal lesson so far is that role allocation is the most sensitive parameter of industrial policy design for effective implementation. These roles are more diverse than the labels “PPP” or similar would have us believe, and can be taken on by a wide variety of actors, whose incentives must be carefully understood in designing grant disbursal mechanisms. In other words, the manner in which the park program fits the political economy likely matters more to avoiding type (i) and (ii) failures than does sector selection, demand estimation by technical consultants, the payment structure for such consultants, or other areas that typically receive more attention. This must be borne in mind when considering the possible replication of such a scheme, not only in other sectors in India, but in other countries. Such replication would be subject to an adaptation to the particular capacity and political economy conditions of each country. Variations such as industry structure might be relatively easy to measure, whether between sectors in the same country or between countries in the same sector. But the general political settlement, formal rules and institutions, and even more critically local informal rules would require a more in-depth analysis to assess their impact on the local political economy of scheme implementation. Political drive at the center might be different; institutions that can play the role of PMCs might be missing; entrepreneurship and the relationship of entrepreneurs with the local political environment might not be as developed; the entrepreneurs able to organize and build the parks might not yet exist; the scheme, if hastily combined with regulatory exemptions, might remove both the pressure for broader reform and the most effective advocates of it. In all, the most important lesson from the SITP for policymakers elsewhere may then arise not from the specific features of the scheme itself, but from its demonstration that thinking in new and creative ways about who does what, and how, within a given political economy, can lead to policy innovations and relative successes in areas previously characterized by a long record of desultory performance. 38 EXHIBITS Exhibit 1: List of 40 first parks and their basic characteristics Target Target Target Target Project GOI's GOI Grant No of Land Private Direct Indirect Annual No Project Name PMC State Location Cost Grant Released firms Area Investment Jobs jobs Production Rs. Rs. Units Crores Crores Rs. Crores Acres Rs. Crores Rs. Crores 1 Hyderabad Hi-tech IL&FS AP Mahboob Nagar 58.0 23.2 12.0 29 60 208.0 2 500 2 500 370 Weaving 2 Hindupur Vyapar Apparel IL&FS AP Ananthpur 102.3 40.0 24.0 2 73 265.5 10 500 22 000 340 3 Pochampally Handloom IL&FS AP Pochampally 34.0 13.6 13.6 6 22 50.0 5 000 3 000 40 4 Brandix India Apparel IL&FS AP Vishakhapatna 134.4 40.0 40.0 17 1 000 4 878.0 60 000 90 000 7 000 City 5 MAS Fabric IL&FS AP Nellore 254.7 40.0 12.0 16 582 1 982.0 31 000 15 000 2 500 6 Gujarat Eco Textile IL&FS GJ Surat 128.8 40.0 40.0 33 104 705.0 8 000 17 000 850 7 Mundra SEZ Textile & IL&FS GJ Kutch 103.5 40.0 40.0 11 116 775.0 3 077 4 500 800 Apparel 8 Fairdeal IL&FS GJ Surat 105.6 40.0 24.0 53 54 312.7 2 900 4 300 1 200 9 Vraj ITP IL&FS GJ Kheda 114.8 40.0 36.0 21 70 550.0 6 250 12 500 617 10 Sayana ITP IL&FS GJ Surat 116.8 40.0 36.0 50 57 298.6 3 155 4 733 1 312 11 Surat Super Yarn IL&FS GJ Surat 104.8 40.0 36.0 27 43 230.6 1 000 2 000 1 250 12 RJD ITP Magus GJ Surat 106.5 40.0 36.0 579 56 352.7 4 270 6 405 598 13 Metro Hi-Tech IL&FS MH Ichalkaranji 106.5 40.0 36.0 86 99 335.0 5 000 5 000 212 Cooperative 14 Pride india cooperative IL&FS MH Ichalkaranji 58.2 23.3 21.0 85 27 203.0 1 500 2 500 300 15 Baramati Hi Tech IL&FS MH Baramati 108.5 40.0 34.8 22 60 250.0 5 500 6 000 380 16 Shri Dhairyashil Mane IL&FS MH Ichalkaranji 72.3 28.9 8.7 167 65 376.6 3 300 5 000 370 Co-op Society 17 Deesan Infrastructure Magus MH Dhul 103.1 40.0 12.0 50 106 721.7 4 410 6 615 2 135 18 Asmeeta Infrastructure Magus MH Bhawandi 200.8 40.0 4.0 65 73 673.2 7 634 11 451 1 046 19 Islampur Integrated Technopak MH Sangl 102.1 40.0 40.0 12 35 334.3 10 000 0 904 20 Latur Integrated Technopak MH Latur 102.6 40.0 40.0 20 50 257.4 10 000 0 617 39 21 Purna Global IL&FS MH Hingoli 91.8 36.7 22.0 41 55 205.0 1 100 550 200 22 The Great Indian Linen & IL&FS TN Perundurai 149.5 40.0 12.0 20 121 418.1 5 000 7 500 520 Textile 23 SIMA Processing Centre IL&FS TN Cuddalore 111.6 40.0 12.0 10 248 475.0 5 000 15 000 500 24 Palladam Hi-Tech IL&FS TN Palladam 55.4 22.2 22.2 90 65 161.3 2 500 3 500 300 Weaving 25 Komarapalayam Hi-Tech IL&FS TN Komarapalayam 34.8 13.9 12.5 57 31 125.7 1 500 1 500 350 Weaving 26 Karur ITP IL&FS TN Karur 116.1 40.0 40.0 42 104 227.0 3 000 4 000 623 27 Madurai ITP IL&FS TN Madurai 87.3 34.9 31.4 15 110 409.8 3 000 4 000 900 28 Vaigai Hi-Tech Weaving C.S. TN Vaigai, Theni 65.1 24.0 2.4 90 41 145.2 6 080 0 361 Architects 29 Kanchipuram AACM C.S. TN Kanchipuram 87.0 33.5 0.0 115 75 119.9 18 000 0 118 Handloom Silk Architects 30 Jaipur Texweaving IL&FS RJ Kishangarh 96.8 38.7 23.2 51 95 250.0 3 000 9 000 550 31 Kishangarh Hi-Tech IL&FS RJ Kishangarh 110.6 40.0 36.0 37 40 416.7 4 000 8 000 800 Textile 32 Next Gen Textile IL&FS RJ Pali 101.4 40.0 24.0 53 100 416.2 9 450 10 000 1 050 33 Jaipur Integrated IL&FS RJ Jaipur 53.5 21.4 19.3 20 23 45.9 4 400 8 800 90 Texcraft 34 Bharat Fabtex & SREI RJ Pali 103.1 40.0 4.0 27 120 416.5 9 450 0 922 Corporate 35 Lotus Integrated Tex IL&FS PB Barnala 110.3 40.0 36.0 8 100 847.7 2 400 2 950 1 740 36 Rhythm Textile & Apparel IL&FS PB Nawanshehar 125.5 40.0 24.0 14 19 339.8 11 000 14 000 350 37 Ludhiana ITP IL&FS PB Ludhiana 116.2 40.0 24.0 55 57 217.0 10 000 10 000 500 38 EIGMEF Apparel IL&FS WB Kolkata 130.5 40.0 24.0 73 13 160.0 10 000 30 000 500 39 Doddabalapur ITP IL&FS KA Doddabalapur 80.3 32.1 32.1 72 48 132.7 2 000 2 000 350 40 CLC ITP IL&FS MP Chhindwara 95.7 38.3 11.5 20 47 301.7 2 000 1 000 400 40 Exhibit 2: Park completion time statistics and graphs (IL&FS parks only) Completion time statistics by state Average Average Average estimated No of Grant Average area/ project completion States Parks Released entrepreneur cost time Min Max Cancel Units % Acres Rs crore Years Yrs Yrs % AP 5 68% 27.5 117 5.9 4.0 8.0 20% GJ 6 93% 3.5 112 5.8 4.6 7.0 0% KA 1 100% 0.7 80 5.9 5.9 5.9 0% MH 5 71% 1.2 87 7.0 6.1 7.6 20% MP 1 30% 2.4 96 6.0 6.0 6.0 0% PB 3 70% 5.0 117 6.0 5.5 6.3 0% RJ 4 75% 1.5 91 7.0 5.1 8.6 0% TN 6 73% 7.0 92 6.7 3.9 9.1 0% WB 1 60% 0.2 131 8.0 8.0 8.0 0% Total 32 75% 7.2 102 6.4 3.9 9.1 6% Completion time statistics by park (30 total, 2 cancelled not included here) Estimated Grant Land Area/ Compl Park Name released Area entrepreneur Project Cost Time Units % Acres Acres Rs. crore Years Baramati Hi Tech Textile 87% 60 2.7 109 7.0 Brandix India Apparel City 100% 1 000 58.8 134 4.0 CLC Textiile 30% 47 2.4 96 6.0 Doddabalapur Integerated Textile 100% 48 0.7 80 5.9 EIGMEF Apparel 60% 13 0.2 131 8.0 Fairdeal Textile 90% 54 1.0 106 5.8 Gujarat Eco Textile 100% 104 3.2 129 4.6 Hindupur Vyapar Apparel 60% 73 36.6 102 8.0 Jaipur Integrated Texcraft 90% 23 1.2 54 5.1 Jaipur Texweaving 60% 95 1.9 97 8.6 Karur Integrated Textile 100% 104 2.5 116 3.9 Kishangarh Hi-Tech Textile 90% 40 1.1 111 7.0 Komarapalayam Hi-Tech Weaving. 90% 31 0.5 35 7.0 Lotus Integrated Tex 90% 100 12.5 110 6.3 Ludhiana Integrated Textile 60% 57 1.0 116 5.5 Madurai Integrated Textile 90% 110 7.3 87 6.3 41 MAS Fabric (India 30% 582 36.4 255 6.8 Metro Hi-Tech Cooperative 90% 99 1.2 107 7.6 Mundra SEZ Textile & Apparel 100% 116 10.6 104 5.1 Next Gen Textile 60% 100 1.9 101 7.3 Palladam Hi-Tech Weaving 100% 65 0.7 55 4.7 Pochampally Handloom 100% 22 3.7 34 4.7 Pride india cooperative Textile 90% 27 0.3 58 7.4 Purna Global Textiles 60% 55 1.3 92 6.1 Rhythm Textile & Apparel 60% 19 1.3 125 6.1 Sayana Textile 90% 57 1.1 117 5.3 SIMA Textile Processing Centre 30% 248 24.8 112 9.1 Surat Super Yarn 90% 43 1.6 105 7.0 The Great Indian Linen & Textile Infrastructure Company 30% 121 6.1 149 8.9 Vraj Integrated Textile 90% 70 3.3 115 7.0 Average 77% 119 7.6 105 6.4 Correlations with estimated completion times No correlation between park size and estimated completion time 10.0 Estimated Completion Time (y) 9.0 8.0 7.0 6.0 R² = 0.036 5.0 4.0 3.0 0 200 400 600 800 1,000 Park land area (acres) Figure Ia 42 Low correlation between park size and estimated completion time (excluding the 2 largest) 10.0 Estimated Completion Time (y) 9.0 8.0 7.0 R² = 0.0955 6.0 5.0 4.0 3.0 0 50 100 150 200 250 Park land area (acres) Figure Ib No correlation between park complexity (project costs) and estimated completion time 10.0 Estimated Completion Time (years) 9.0 8.0 7.0 R² = 0.0248 6.0 5.0 4.0 3.0 0 50 100 150 200 250 300 Project Cost (Rs. Crore) Figure IIa Low correlation between park complexity (project costs) and estimated completion time (excluding the 2 largest) 10.0 Estimated Completion Time (years) 9.0 8.0 7.0 6.0 R² = 0.0736 5.0 4.0 3.0 0 50 100 150 200 Project Cost (Rs. Crore) Figure IIb 43 No correlation between average park business size and expected completion time 10.0 Estimated Completion Time (years) 9.0 8.0 7.0 6.0 R² = 0.0008 5.0 4.0 3.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Area per entrepreneur (Acres) Figure IIIa Low correlation between average park business size and estimated completion time (excluding the 2 largest 10.0 Estimated Completion Time (years) 9.0 8.0 7.0 R² = 0.119 6.0 5.0 4.0 3.0 0.0 10.0 20.0 30.0 40.0 Area per entrepreneur (Acres) Figure IIIb 44 Exhibit 3: Comparisons between SITP and SEZ Act implementation Metric SEZ Act SITP Official Start Date February 2006 July 2005 Number of projects formally approved by end 2011 580 40 Number of projects notified by end 2011 380 36 Conversion ratio 1: notified/approved 66% 90% Operational projects by end 2011 124 24 Conversion ratio 2: operational/approved 21% 60% Of which non-IT 44 24 Notes Formally approved means, for SEZs, having put in an application - the same definition is used for SITP Notified means having secured necessary land and being ready to start building - the same definition is used for SITP Operational means being at least in the first stage of production - the same definition is used for SITP Sources http://www.sezindia.nic.in/writereaddata/pdf/List_of322notifiedSEZsunderSEZAct2005as19.6.09- masterlist.pdf World Bank (2012), “Learning from SEZs in India”, Unpublished 45 Exhibit 4: Investment statistics for IL&FS SITP parks Investment statistics by state (average numbers, including all 32 parks) Target Current No of Area Project Target Grant Investment investment investment States Parks /unit cost investment Released achieved /grant ratio /grant ratio Units Acres Rs crores Rs crores % % AP 5 27.5 117 1477 68% 25% 38.2 4.7 GJ 6 3.5 112 479 93% 43% 12.0 5.8 KA 1 0.7 80 133 100% 121% 4.1 5.0 MH 5 1.2 87 274 71% 36% 8.4 3.4 MP 1 2.4 96 302 30% 5% 7.9 1.4 PB 3 5.0 117 468 70% 24% 11.7 4.0 RJ 4 1.5 91 282 75% 40% 7.4 2.6 TN 6 7.0 92 303 73% 33% 9.3 3.0 WB 1 0.2 131 160 60% 25% 4.0 1.7 Total 32 7.2 102 518 75% 36% 13.8 3.9 Investment statistics by park (removing the 2 cancelled parks) Target Current Area Project Target Grant Investment investment investment Park Name /unit cost investment Released achieved /grant ratio /grant ratio Units Acres Rs crores Rs crores % % Baramati Hi Tech Textile 2.7 109 250 87% 60% 6.3 4.3 Brandix India Apparel City 58.8 134 4 878 100% 12% 122.0 14.2 CLC Textiile 2.4 96 302 30% 5% 7.9 1.4 Doddabalapur Integerated Textile 0.7 80 133 100% 121% 4.1 5.0 EIGMEF Apparel 0.2 131 160 60% 25% 4.0 1.7 Fairdeal Textile 1.0 106 313 90% 40% 7.8 5.2 Gujarat Eco Textile 3.2 129 705 100% 56% 17.6 10.0 Hindupur Vyapar Apparel 36.6 102 265 60% 19% 6.6 2.1 Jaipur Integrated Texcraft 1.2 54 46 90% 93% 2.1 2.2 Jaipur Texweaving 1.9 97 250 60% 24% 6.5 2.6 Karur Integrated Textile 2.5 116 227 100% 55% 5.7 3.1 Kishangarh Hi-Tech Textile 1.1 111 417 90% 32% 10.4 3.7 Komarapalayam Hi- Tech Weaving. 0.5 35 126 90% 42% 9.0 4.2 Lotus Integrated Tex 12.5 110 848 90% 29% 21.2 6.9 Ludhiana Integrated Textile 1.0 116 217 60% 24% 5.4 2.1 Madurai Integrated Textile 7.3 87 410 90% 23% 11.7 3.0 MAS Fabric India 36.4 255 1 982 30% 2% 49.6 2.6 46 Metro Hi-Tech Cooperative 1.2 107 335 90% 37% 8.4 3.5 Mundra SEZ Textile & Apparel 10.6 104 775 100% 57% 19.4 11.0 Next Gen Textile 1.9 101 416 60% 10% 10.4 1.8 Palladam Hi-Tech Weaving 0.7 55 161 100% 68% 7.3 5.0 Pochampally Handloom 3.7 34 50 100% 84% 3.7 3.1 Pride india cooperative Textile 0.3 58 203 90% 54% 8.7 5.3 Purna Global Textiles 1.3 92 205 60% 24% 5.6 2.3 Rhythm Textile & Apparel 1.3 125 340 60% 20% 8.5 2.9 Sayana Textile 1.1 117 299 90% 32% 7.5 2.7 SIMA Textile Processing Centre 24.8 112 475 30% 4% 11.9 1.5 Surat Super Yarn 1.6 105 231 90% 57% 5.8 3.6 The Great Indian Linen & Textile Infrastructure Company 6.1 149 418 30% 4% 10.5 1.5 Vraj Integrated Textile 3.3 115 550 90% 17% 13.8 2.6 Average 7.6 105 533 77% 38% 14.0 4.0 47 Exhibit 5: Scatter plots of investment target achievement Completed parks are close to their Completed parks are close to their investment targets investment targets (without Brandix & MAS) 125% 125% % Target achived % Target achived 100% 100% 75% 75% 50% R² = 0.4639 50% R² = 0.5141 25% 25% 0% 0% 20% 40% 60% 80% 100% 20% 40% 60% 80% 100% % GoI grant released % GoI grant released Slower progress in more complex Slower progress in more complex projects projects (withouth Brandix and MAS) 125% 125% % Target achived % Target achived 100% 100% 75% 75% 50% 50% R² = 0.2922 25% R² = 0.306 25% 0% 0% 0 100 200 300 0 50 100 150 200 Project Cost (Rs. crore) Project Cost (Rs. crore) Low negative correlation between Low negative correlation between investment target achivement and the investment achivement and size of size of businesses in the park businesses in the park (ex-Brandix, MAS) % Target achived 150% 150% % Target achived 100% 100% 50% 50% R² = 0.0885 R² = 0.1503 0% 0% 0.0 20.0 40.0 60.0 0.0 10.0 20.0 30.0 40.0 Average plot size (acres/entrepreneur) Average plot size (acres per entrepreneur) Apparent strong correlation between No correlation between achieved investment/grant achieved investment/grant ratios and ratios and targets, but many have have achieved targets targets (without Brandix and MAS) Current investment/grant Current investment/grant 15 12.5 R² = 0.4458 10.0 10 ratio ratio 7.5 5.0 R² = 0.0008 5 2.5 0 0.0 0 20 40 60 80 100 120 140 0 5 10 15 20 25 Target investment/grant ratio Target investment/grant ratio 48 Exhibit 6: Jobs, production and efficiency statistics (average numbers, IL&FS parks) Target Target Target Current Target Current No of Annual Direct Production direct jobs grant /job grant /job States Parks Production Jobs /Job ratio (Feb12) ratio ratio Units Rs. Crore Rs. Lakhs Rs. Lakhs Rs. Lakhs AP 5 10 250 109 000 8 23 700 0.2 5.0 GJ 6 6 029 24 382 42 2 840 1.5 11.5 KA 1 350 2 000 18 250 1.6 12.8 MH 5 1 462 16 400 12 7 280 1.6 3.4 MP 1 400 2 000 20 0 1.9 N/A PB 3 2 590 23 400 27 1 137 0.8 10.6 RJ 4 2 490 20 850 13 1 150 0.8 10.1 TN 6 3 193 20 000 18 5 420 1.0 2.1 WB 1 500 10 000 5 0 0.4 N/A Total 32 27 264 228 032 20 41 777 1.1 7.4 49 Exhibit 7: Land statistics by state and by purchase type (IL&FS parks) Land statistics by state (average numbers, 32 IL&FS parks) State/acquisition No of Area Project Land cost Min Max Land process Parks Land Area /unit Cost /acre /acre cost/acre cost/acre Units Acres Acres Rs. Crore Rs. Lakhs Rs. Lakhs Rs. Lakhs AP 5 347 27.5 0.9 2.8 0.0 5.4 Private 4 184 19.7 1.1 3.4 1.5 5.4 State 1 1000 58.8 0.1 0.0 0.0 0.0 GJ 6 74 3.5 1.7 21.0 3.3 59.4 Private 6 74 3.5 1.7 21.0 3.3 59.4 KA 1 48 0.7 1.7 32.7 32.7 32.7 State 1 48 0.7 1.7 32.7 32.7 32.7 MH 5 61 1.2 1.6 4.9 1.6 8.0 Private 1 55 1.3 1.7 1.6 1.6 1.6 State 4 63 1.1 1.5 5.8 4.1 8.0 MP 1 47 2.4 2.0 2.7 2.7 2.7 Private 1 47 2.4 2.0 2.7 2.7 2.7 PB 3 59 5.0 3.3 35.1 21.0 54.7 Private 3 59 5.0 3.3 35.1 21.0 54.7 RJ 4 64 1.5 1.8 11.1 0.7 36.3 Private 1 100 1.9 1.0 0.7 0.7 0.7 State 3 53 1.4 2.0 14.6 3.5 36.3 TN 6 113 7.0 0.9 1.7 0.5 3.5 Private 5 86 3.4 1.0 1.3 0.5 1.9 State 1 248 24.8 0.5 3.5 3.5 3.5 WB 1 13 0.2 10.1 125.8 125.8 125.8 Private 1 13 0.2 10.1 125.8 125.8 125.8 Grand Total 32 116 7.2 1.8 15.2 0.0 125.8 Land statistics by purchase type: private (with or without state support), or state acquired Project Estimated Acquisition No of Land Area Cost Land cost Min Max Land Completion process/IDC Parks Area /unit /acre /acre cost/acre cost/acre Time Units Acres Acres Rs. Crore Rs. Lakhs Rs. Lakhs Rs. Lakhs Years Private 22 91 6.2 2.0 17.4 0.5 125.8 6.2 APIIC, AP 1 582 36.4 0.4 1.5 1.5 1.5 6.8 (blank) 21 68 4.8 2.1 18.1 0.5 125.8 6.2 State 10 170 9.3 1.5 10.3 0.0 36.3 6.9 APIIC, AP 1 1000 58.8 0.1 0.0 0.0 0.0 4.0 KIADB, KA 1 48 0.7 1.7 32.7 32.7 32.7 5.9 MIDC, MH 4 63 1.1 1.5 5.8 4.1 8.0 7.3 RIICO, RJ 3 53 1.4 2.0 14.6 3.5 36.3 6.9 SIPCOT, TN 1 248 24.8 0.5 3.5 3.5 3.5 9.1 Grand Total 32 116 7.2 1.8 15.2 0.0 125.8 6.4 50