WPS7874 Policy Research Working Paper 7874 Does the Elimination of Export Requirements in Special Economic Zones Affect Export Performance? Evidence from the Dominican Republic Fabrice Defever José-Daniel Reyes Alejandro Riaño Miguel Eduardo Sánchez-Martín Trade and Competitiveness Global Practice Group October 2016 Policy Research Working Paper 7874 Abstract Special economic zones, one of the most important instru- requirements. The findings show that entry increased ments of industrial policy in developing countries, often among firms in special economic zones, while the aver- feature export share requirements. That is, firms located in age value of export transactions fell for existing exporters these zones are obliged to export more than a certain stated following the reforms. At the same time, continuous export- share of their output to enjoy the wide array of incentives ers were unaffected by the policy change, possibly because available there, a practice prohibited by the World Trade these firms were not constrained by the export requirement. Organization. This paper exploits the staggered removal Overall, special economic zones became more important of export requirements across products and over time in with respect to the number of exporters based there but not the special economic zones of the Dominican Republic in terms of the value of exports. The findings suggest that to evaluate whether the importance of exports originating the elimination of performance requirements made it more from the zones was affected by the elimination of export attractive for firms to be based in special economic zones. This paper is a product of the Trade and Competitiveness Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at danielreyes@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Does the Elimination of Export Requirements in Special Economic Zones Affect Export Performance? Evidence from the Dominican Republic¦ X Jos´ Fabrice Defever, Y , e-Daniel Reyes § Alejandro Ria˜ no, Miguel Eduardo S´ ın¶ anchez-Mart´ ¦ We thank Thomas Farole, Michael Ferrantino, Justo Mirabal and participants at DEGIT XXI-Nottingham and the UNIDO-Kiel Vienna Investment Conference for their thoughtful comments and suggestions. All remaining errors are our own. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not represent the view of the World Bank, its Executive Directors, or the countries they represent. X City, University of London, CEP, GEP and CESifo. f.defever@lse.ac.uk Y The World Bank.danielreyes@worldbank.org § University of Nottingham, GEP, CFCM and CESifo. alejandro.riano@nottingham.ac.uk ¶ The World Bank.msanchezmartin@worldbank.org 1 Introduction On July 31, 2007, the General Council of the World Trade Organization (WTO) set December 31st 2015 as the final deadline for the elimination of export subsidies in the Dominican Republic and 18 other developing countries that had previously been exempted from complying with the Agreement on Subsidies and Countervailing Measures (ASCM).1 The majority of subsidy programs to be scrapped were fiscal incentives available to firms operating in special economic zones (SEZ) — geographically-bounded areas in which customs, tax and investment regulations are more liberal than in the rest of the country (Farole and Akinci, 2011). Although SEZ are not explicitly forbidden by the WTO, firms operating in them are often subject to export share requirements (ESR) (Defever no, 2016); that is, they are required to export at least a certain stated share of their output and Ria˜ in order to be eligible to receive the incentives available in SEZ.2 The imposition of ESR makes an SEZ contingent upon export performance, thereby falling under the category of prohibited subsidies under the ASCM.3 SEZ are pervasive across the world, and are one of the most important tools of industrial policy in developing countries (Rodrik, 2004); Boyenge (2007) reports the existence of 3,500 (more recently, The Economist (2015) puts this figure at 4,300) SEZ in 130 countries, accounting for approximately US$ 200 billion worth of exports. These zones are crucially important for a small open economy like the Dominican Republic, where exports originating from SEZ comprised 70% of aggregate exports and 12% of GDP over the last decade, a period which was also characterized by a chronic trade ublica Dominicana, 2014). balance deficit, averaging 8.4% of GDP (Banco Central de la Rep´ Our objective in this paper is to investigate whether the removal of export share requirements affected export performance in the special economic zones of the Dominican Republic. To do so, we use transaction-level data covering the period 2006-2014, and exploit the fact that the elimination of ESR took place in a staggered fashion over time and across different industries, in order to identify the effect of this reform on the importance of exports originating from SEZ. To 1 See: General Council decision of July 31, 2007 WT/L/691. The other beneficiaries of this extension were Antigua and Barbuda, Barbados, Belize, Costa Rica, Dominica, El Salvador, Fiji, Grenada, Guatemala, Jamaica, Jordan, Mauritius, Panama, Papua New Guinea, St. Kitts and Nevis, St. Lucia, Saint Vincent and the Grenadines, and Uruguay. The notification also lists the subsidy programs that needed to be reformed. 2 In fact, as Creskoff and Walkenhorst (2009) point out, they are not mentioned explicitly in any WTO agreement. 3 Only least developed WTO members and countries whose per capita gross national product is below US$1,000 in 1990 dollars are currently exempt from the disciplines of the ASCM (Creskoff and Walkenhorst, 2009). 1 be more precise, prior to 2007 all firms based in Dominican SEZ were subject to an 80% ESR — i.e. they could sell at most 20% of their output in the Dominican customs territory. In May 2007, law 56-07 first amended the regulatory framework for SEZ by declaring the textiles, clothing and accessories, hides and skins, footwear, and leather articles sectors to be of ‘national priority’, based on their importance in terms of exports and employment. The 2007 reform eliminated ESR for SEZ firms in national priority sectors only, and at the same time extended some of the tariff and tax incentives available to firms in SEZ to local producers outside the zones. In June 2011, law 139-11 fully eliminated ESR for all firms in SEZ regardless of their sector of operation. However, and in contrast to the 2007 reform, the latter law did not extend any incentives to non-SEZ firms. Our empirical analysis is underpinned by a model of subsidies with ESR in a heterogeneous- no (2016). Based on this theoretical framework we firm environment proposed by Defever and Ria˜ develop a set of predictions regarding the consequences of eliminating ESR on firms’ extensive and intensive margins of exports, as well as on the share of exports originating from SEZ (both in terms of values and number of exporters) within narrowly-defined HS-6 products. The model shows that ESR reduce the profitability of firms in SEZ which would have preferred to sell a higher share of its sales in the domestic market than what the ESR allows — which we call ‘constrained exporters’. The model predicts that these firms would increase the prices they charge abroad and thereby reduce export sales after the ESR is lifted. Conversely, firms that would have chosen to export 80% or more of their output even in the absence of ESR — i.e. unconstrained exporters — are unaffected by the reform. If, as it was the case in the Dominican Republic, the reform of SEZ maintains the incentives available to firms based there, then the elimination of ESR will make them more attractive locations for firms to be based. Since the latter goes in the opposite direction as the effect of the ESR removal on the intensive margin of exports for existing constrained exporters, the question of the importance of the response of SEZ exports to the policy change is, ultimately, an empirical one. Our results can be summarized as follows. In terms of the extensive margin, we find that in line with the predictions delivered by our model, export entry among firms in SEZ increased after the removal of ESR — particularly so after the second wave of reform, which targeted firms producing non-priority goods. Moreover, we also find that the exit rate out of export markets increased among non-SEZ firms in priority sectors, possibly due to more intense competition from 2 firms operating in the zones. Moving on to the intensive margin of exports, we find that the value of export transactions among SEZ firms fell following both waves of reform, as predicted by our model. However, when restricting our analysis to the subset of firms that export in every year throughout our period of study, we find that the change in the value of exports after the policy change becomes statistically insignificant. This result suggests that continuing exporters do not find the ESR constraint binding, whereas more sporadic exporters seem to have reallocated their sales towards the domestic market once this alternative became feasible. Examining the impact of the removal of ESR on the share of exports — both in terms of value and number of exporting firms — at the product level paints a similar picture to that conveyed by our analysis of export margins. We find that the removal of ESR increased the share of SEZ exporters within HS-6 products by 5.4-7 percentage points relative to the situation in which per- formance requirements were in place. At the same time, the share of export value accounted for by SEZ does not change significantly after the reforms. As the requirements to operate in SEZ were eased, more exporters chose to locate there; however, the bulk of export value originating from SEZ seems to be accounted for by firms that were based there with the clear objective to export the majority of their output, and were therefore not affected by the removal of ESR. As noted above, the first wave of SEZ reforms implemented in 2007 sought to level the playing field between SEZ and non-SEZ firms by lowering the tariffs imposed on key inputs utilized by producers in priority sectors (firms in SEZ do not pay tariffs on intermediate inputs). Thus, with a view to assess the effectiveness of this element of the Dominican reform strategy, we investigate if the share of these liberalized products purchased by SEZ declined after 2007. Our conclusions in this dimension are univocally negative; we do not find any substantial change in the share of imports of these goods accounted for by SEZ after the trade liberalization. Thus, although a large number of goods were made duty-free (126 HS-6 products), the ones that were primarily imported by non-SEZ firms were, for the most part, general purpose rather than inputs specific to priority sectors. This rendered the tariff cuts ineffective as an instrument to foster the expansion of firms outside the zones in these sectors. Taken altogether, our results suggest that the removal of ESR without eliminating the fiscal incentives available to firms in SEZ has increased the attractiveness of operating in special economic zones. Achieving a greater degree of compliance with the WTO disciplines on subsidies has not 3 substantively reduced the Dominican Republic’s reliance on SEZ – if anything, the removal of ESR has helped in cementing their importance in the country’s export basket, at least in the short-run. Achieving compliance with WTO disciplines on subsidies poses a complex challenge for policy makers in countries that rely intensively on exports originating from SEZ in the face of large and chronic trade deficits. Opening up the domestic market to firms based in SEZ can result in unduly intense competition for local firms given the tax advantages available in the zones. Curbing these concessions, however, entails the risk that multinational firms would decide to relocate their oper- ations to countries offering more generous incentive packages. Alternatively, equalizing incentives inside and outside SEZ might not be feasible from a fiscal policy standpoint (FIAS, 2008; World Bank, 2014). Thus, taking stock of the Dominican Republic’s experience with eliminating export requirements can help to shed light on the future role of SEZ as an export promotion policy oper- ating within a regulatory framework compatible with the WTO agreements. Related Work. Export performance requirements have been studied from a theoretical perspec- tive by Davidson et al. (1985), Rodrik (1987) and Chao and Yu (2014), with the aim of determining under which circumstances these measures can improve welfare in countries enacting them. The main conclusion arising from this body of work is that ESR can operate as second-best policies, improving domestic welfare in the presence of pre-existing distortions.4 For instance, ESR can shift rents towards the enacting country that would otherwise accrue to foreign-owned firms, or can also lessen the overproduction distortion caused by tariffs imposed in import-competing sectors. More no (2015, 2016), have studied how the use of subsidies with ESR affects recently, Defever and Ria˜ the intensity of competition and welfare using a quantitative model calibrated to evaluate the Chi- nese experience with this class of incentives. We contribute to this literature by focusing on the microeconomic effects of export share requirements on export performance. Moreover, evaluating the consequences of ESR in the context of the reforms undertaken by the Dominican Republic no (2015, 2016) — has the advantage — relative to the case of China studied by Defever and Ria˜ that the set of firms facing ESR are readily identifiable with our data and are only subject to one 4 The theoretical literature on SEZ has also reached similar conclusions. See e.g. Hamada (1974), Young and Miyagiwa (1987), Young (1991), Devereux and Chen (1995). Heid et al. (2013), on the other hand, show that an expansion of export-oriented maquiladoras (i.e. firms exporting all their output) increases labor informality and lowers aggregate welfare. 4 such requirement.5 Additionally — and quite importantly — the fact that ESR were eliminated in a staggered way across different sectors provides valuable variation across time and sectors that allows us to identify the consequences of this policy change. The Dominican Republic being one of the pioneers, and arguably, most successful exponents of the use of SEZ in the Western Hemisphere, stands at the heart of a long-standing debate over their desirability as an industrial policy to foster economic development (Volpe Martincus, 2010). On the one hand, Kaplinsky (1993) argues that SEZ induced the Dominican Republic to specialize in unskilled labor-intensive production, which due to declining terms-of-trade has resulted in im- miserizing growth. Willmore (1995), on the contrary, argues that judged by its ability to promote exports without threatening local producers, the Dominican SEZ program has been an unqualified success. Sanchez-Ancochea (2006) and Schrank (2008) qualify this assertion and suggest that the success of SEZ beyond export promotion hinges on the extent to which they have integrated with ıguez-Clare (1996). Our paper contributes the local economy — an argument formalized by Rodr´ to this discussion by assessing how recent policy efforts aimed at promoting a new wave of SEZ that goes beyond the low-skill-intensive enclaves of the past, have affected export outcomes — arguably, one of the key performance dimensions for SEZ. Crucially, it does so by utilizing highly detailed transaction-level micro data which allow us to elucidate the effect of the policy changes while controlling for confounding factors that would contaminate the results obtained using only aggregate data. Our paper is related to work by Bagwell and Staiger (2006) studying the implications of GATT- WTO rules on subsidies on economic outcomes, as well as to broader efforts to quantify the con- sequences of eliminating “murky” forms of protectionism such as export subsidies, local content requirements, and public procurement measures (Baldwin and Evenett, 2009; Evenett and Wer- melinger, 2010). To the best of our knowledge, this paper is the first in carrying out an empirical assessment of whether policy reforms aimed at achieving compliance with WTO rules on subsi- dies have affected trade outcomes. This is an issue of tremendous importance, since the inherent difficulty in defining and measuring export subsidies, combined with a lack of comparable cross- country data has resulted in fewer empirical studies investigating them than any other instrument 5 no (2016) document a wide range of often overlapping policy measures subject to In contrast, Defever and Ria˜ different ESR thresholds in China. 5 of commercial policy under the aegis of the World Trade Organization (World Trade Organization, 2006). Our paper also contributes to a small, but flourishing empirical literature that investigates the effect of SEZ and export processing regimes on exports (Ianchovichina, 2007; Fernandes and Tang, 2012; Wang and Yu, 2012; Alder et al., 2013; Wang, 2013; Davies and Mazhikeyev, 2015; ucer and Siro¨ Y¨ en, 2016). The rest of the paper is organized as follows: Section 2 summarizes the regulations governing SEZ in the Dominican Republic and the changes introduced by laws 56-07 and 139-11, which mandated the elimination of ESR. Section 3 sketches a simple model of trade with heterogeneous firms featuring subsidies subject to export share requirements and outlines predictions regarding how the elimination of ESR would affect export performance at the firm and product level. Section 4 describes our data and provides a set of stylized facts characterizing export patterns in the Dominican Republic, comparing firms in SEZ and those exporting through the national customs regime as well as firms in national priority sectors and other industries. Section 5 discusses our empirical strategy and presents our results. Section 6 concludes. 2 Regulatory Changes in the Special Economic Zones of the Dominican Republic In this section we provide a brief summary of the incentives and performance requirements faced by firms operating in SEZ in the Dominican Republic; we then discuss the reforms introduced by laws 56-07 and 139-11 in 2007 and 2011 respectively. Law 8-90 of January 1990 established the regulatory framework governing SEZ (Zonas Francas ) in the Dominican Republic. According to the law, its key objectives were to attract local and foreign investment, provide training, and foment the transfer of technology and know-how in order to create employment, particularly in economically deprived areas, such as the border with Haiti. Law 8-90 sought to encourage the expansion of SEZ by providing a generous array of fiscal incentives to firms located there: these included duty-free access to imported inputs and capital goods, and a 15-year (20 years for firms located in border zones) 100% exemption of registration, construction, gross sales and transfer of industrial goods taxes. In order to be eligible for these concessions, firms operating in SEZ faced an 80% export share requirement (ESR), while firms located outside SEZ 6 and exporting through the national customs regime were not subject to any performance obligations regarding their export behavior. SEZ firms were also required to pay the corresponding full import duty on the goods sold in the Dominican Republic. Until 2007 the ESR applied equally to all SEZ firms regardless of their sector of operation as can be seen in the first row of Table 1. Law 56-07 signed in May 2007 amended law 8-90 and declared the textile, footwear and leather industries ‘national priority’ sectors. The law’s motivation noted that worldwide structural change in the textile, footwear and apparel industries driven by China’s increasing preponderance in the world economy and changes in WTO agreements such as end of the MFA agreement have eroded the competitiveness of these sectors in the Dominican Republic. Due to their importance in terms of job creation, the law sought to provide support to producers in these industries in the face of adverse external circumstances. Besides this objective, law 56-07 also aimed at fostering the convergence between SEZ and non-SEZ firms in view of the country’s agreement to eliminate performance requirements under the CAFTA-DR trade agreement and the ASCM. This constituted a gradual first step forward towards the SEZ regime achieving compliance with the ASCM — following on the footsteps of China, Vietnam, Mauritius and Costa Rica (Waters, 2013; World Bank, 2014). Table 1: Changes in SEZ Regulations in the Dominican Republic, 2006-2014 Period National Priority Sectors Non-priority Sectors 80% ESR; duty-free imports of intermediate inputs and capital goods; full exemption of gross 2006-07 sales, registration, construction and transfer of industrial goods (ITBIS) taxes for 15 years (20 years for firms in border SEZ) SEZ firms ESR fully removed; duty-free access to 80% ESR remains; duty-free access on domestic 2008-11 sales if product is not produced in DR or has at domestic market least 25% of local input content Domestic sales remain free of import duties but ESR fully removed; Domestic sales are subject to 2012-14 are subject to a 3.5% tax on gross sales and import duty, 3.5% tax on gross sales and 18% 18% VAT VAT 2006-07 No ESR; subject to national customs regime Non-SEZ firms Duty-free access to 126 'priority' intermediate 2008-11 inputs; exemption of tax on transfer of No change industrial goods (ITBIS) 2012-14 No change Source: Law 56-07 (available in Spanish at https://www.dgii.gov.do/legislacion/leyesTributarias/Documents/ 56-07.pdf) and law 139-11 (available in Spanish at https://www.dgii.gov.do/legislacion/leyesTributarias/ Documents/139-11.pdf). 7 The second and fifth rows of Table 1 outline the main policy changes brought about by law 56-07 for SEZ and non-SEZ firms in priority and non-priority sectors. SEZ firms in priority sectors saw the full removal of ESR, which meant that they could now sell all their output in the Dominican Republic if they wished to do so; moreover, these domestic sales were not subject to import duties. Priority-sector firms operating outside SEZ received tax concessions similar to those available to their SEZ counterparts and enjoyed duty-free access to 126 HS-6 key imported inputs. The 2007 reform maintained the 80% ESR for SEZ firms producing non-priority goods but offered them duty-free access to the domestic market provided that either the good in question was not produced in the Dominican Republic or, that it incorporated at least 25% of locally-sourced intermediate inputs in value terms. Tax concessions available to SEZ firms in both priority and non-priority sectors did not change with this reform. Firms producing non-priority goods and located outside SEZ were not directly affected by law 56-07. Law 139-11 (June 2011) completely eliminated export share requirements for all SEZ firms regardless of their sector of operation in accordance to the compromises signed under the CAFTA- DR free trade agreement. SEZ firms in priority sectors retain their duty-free access to the Dominican market whereas their non-priority counterparts need to pay the customary import tariffs mandated by the national customs regime. All SEZ firms are required to pay a 3.5% gross sales tax and 18% VAT on their domestic sales (see row 3 of Table 1), while non-SEZ firms were not directly affected by this law. 3 How Does Eliminating Export Share Requirements Affect Ex- port Performance? In order to understand how the elimination of export share requirements for firms in special eco- nomic zones affects export performance, we use a theoretical framework developed by Defever and no (2016), which is in turn based on the workhorse Melitz (2003) model of international trade Ria˜ with heterogeneous firms, to guide our empirical analysis. However, since our focus lies on indi- vidual firm and product-level effects of the policy change, we abstract from general equilibrium considerations. To fix ideas, consider an environment with two countries, Home (H ), which we take to be the 8 Dominican Republic, and Foreign (F ), which represents the rest of the world. There is a continuum of firms operating at Home, each of which produces a unique differentiated good ω using labor as the sole input. We assume, for simplicity, that all firms are homogenous in terms of their productivity, which implies that all producers have a constant marginal cost, c.6 The demand function faced by a firm selling product ω in market i € tH, F u, is: qi pω q  Ai pω qpi pω q¡σ , (1) where σ ¡ 1 denotes the elasticity of demand and pipωq is the price charged by firm ω in market i.7 The term Ai pω q encompasses both aggregate variables such as a country’s GDP and price level — both of which are common across all producers — as well as product appeal factors that are firm-destination-specific. Including the latter is necessary to produce a non-degenerate distribution of export intensity. As discussed below, this implies that some firms will export the majority of their output even without export share requirements. We assume that firms can sell domestically without incurring any additional cost, but face a fixed cost fx and an iceberg transport cost τ ¥ 1 if they export. Crucially, firms can choose to sell their output abroad either through the national customs regime or via a special economic zone. Non-SEZ firms are free to sell however much they want domestically but do not receive any incentives. On the other hand, firms located in SEZ stand to receive a subsidy sr on their sales or marginal costs (e.g. duty-free access to imported inputs, corporate income tax exemptions or lower utility rates) and/or a subsidy sf on their fixed costs (e.g. lower registration costs or access to better port infrastructure) provided they export more than a fraction η of their sales, which prior to 2007 would have been set at 0.8 for all firms in Dominican SEZ. We explain the profit maximization problem of firms operating in these two regimes below. All firms sell some of their output domestically. As it is well known, given the demand function (1), firms set a price at Home, pH pω q  σc ¡ which is a constant mark up above their marginal cost, σ 1,   σ¡1 ¨σ¡1 thus generating domestic sales rH pω q  σc AH pω q, which are increasing in the domestic de- 6 Assuming instead that firms differ in terms of productivity, as is standard in the literature, does not alter the conclusions of the analysis that follows. This is the case because firm-level export intensity is independent of productivity when demand functions are iso-elastic. 7 Such a demand function would obtain if individuals have love-for-variety Dixit-Stiglitz utility functions as in Krugman (1980) or Melitz (2003). 9 mand shifter. Domestic profits are in turn given by πH pω q  κAH pω q, where κ  pσ ¡1qσ¡1 σ ¡σ c1¡σ . Let us now consider the problem of an exporter selling its output through the national customs regime. Since the marginal cost of production is constant, this firm will choose to export if the profit from selling abroad exceeds the fixed cost fx ; otherwise it would prefer to only sell domestically. A non-SEZ exporter sets price pF pω q  τ pH pωq in the foreign market; its export sales are given 1¡σ σ ¡1 σ ¡1   ¨ by rF pω q τ σc AF pω q, and export profits are πF pω q  κτ 1¡σ AF pω q ¡ fx . It follows that only firms with sufficiently high foreign demand appeal choose to export; that is, those for which AF pω q ¥ A¦ F  pfx {κqτ σ ¡1 . Total profits for non-SEZ exporters are therefore given by   ¨ π pω q  κ AH pω q   τ 1¡σ AF pω q ¡ fx . Define the ‘natural’ export intensity of an exporter, η pω q, as the share of its total sales accounted for by exports in the absence of subsidies or ESR: rF pω q τ 1¡σ AF pω q η pω q   A pω rH pω q   rF pω q q   τ 1¡σ AF pωq . (2) H It is straightforward to show that any deviations from (2) would reduce firms’ profits by distorting the necessary condition that marginal revenue has to be equalized across markets. As (2) makes clear, firms can operate at very high export intensities in the absence of ESR, for instance, if they conduct assembly operations as a link in a global value chain, if they produce highly sophisticated products with little domestic demand (medical instruments would be a relevant example for the Dominican Republic) or have built an extensive network of foreign customers.8 The profit maximization problem of a firm located in the SEZ facing an ESR η € p0, 1s and receiving subsidies psr , sf q is: 5  %  $ 1¡σ 1¡σ ¡σ   τ A pωq ppz pωqq¡σ pz max p q p q z p1 sr q AH p ω q pp z H pω qq   AF pω q pp z F pω qq ¡c AH p ω q pp z H pω qq F F H ω ,pF ω C 1¡σ F pω qq AF pω q ppz ¡ p1 ¡ sf qfx subject to:   ¨1¡σ   ¨1¡σ ¥ η, (3) H pω q AH pω q pz   AF pω q F pω q pz no (2016) show where the superscript z indexes variables for firms based in SEZ. Defever and Ria˜ that the solution to problem (3) involves two types of firms operating in SEZ, which we denote 8 If fixed costs associated with selling domestically were included in the model, then ‘pure exporters’, i.e. firms selling all their output abroad, would arise when domestic profits are not sufficiently high to cover the domestic fixed cost. 10 unconstrained and constrained exporters. Unconstrained exporters are those that do not find the ESR constraint binding; i.e. firms with natural export intensity of at least η . These firms do not need to distort their allocation of sales across domestic and foreign markets to receive the subsidies available in SEZ. Thus, H pω q  ¡ and pF pωq  H pω q and realize total profits π pω q  1 σc they charge prices pz   1 sr σ 1 z τ pz z   ¨ p1   sr qσ κ AH pω q   τ 1¡σ AF pω q ¡ p1 ¡ sf qfx. For unconstrained exporters in SEZ any incen- tive conditioned on ESR operates in the same way as a standard unconditional production subsidy. This means that they charge lower prices, both domestically and abroad, and earn higher profits than in the absence of subsidies, while still operating at their natural export intensity. Constrained exporters, on the other hand, are firms with natural export intensity strictly below the ESR threshold, but that nevertheless choose to alter their export intensity in order to be eligible to operate in the SEZ and receive the subsidies available there. These firms curtail domestic sales, by increasing prices at Home while at the same time increasing export sales until they reach an export no, 2016, Proposition 1). Therefore, constrained intensity exactly equal to η (see Defever and Ria˜ exporters in SEZ set prices,  ' p1 ¡ η q   η ¡ τ A H pω q ¡ AF pω q σ¡1 σ 1 σ ¡ 1 pω q  1 σc σ 1 σ 1 σ 1 pz 1   sr σ ¡ 1 , (4) H p1 ¡ η q ¡ AF p ω q ¡1 σ 1 1 σ 1  ' p1 ¡ ηq ¡ AF pωq ¡   η ¡ τ AH pωq σ 1 σ ¡ 1 F pω q  σ 1 σ 1 σ 1 σ 1 1 σc pz 1   sr σ ¡ 1 , (5) η ¡ AH pω q ¡ 1 1 σ 1 σ 1 which result in profits:   ¨ π z pω q  κp1   sr qσ Θ AH pω q, AF pω q, η, τ ¡ p1 ¡ sf qfx, (6) where Θ, is defined as: AH pω qAF pω q Θ  % σ ¡1 . (7) p1 ¡ η q ¡ AF pω q σ¡1 σ σ 1 1  η ¡ τ AH pω q σ¡1 σ σ 1 1 The profit shifter term Θ is strictly concave in η reaching a global maximum at a firm’s natural export intensity, η pω q. Thus, firms with relatively high natural export intensity — yet lower than 11 the ESR threshold — are more likely to operate as constrained exporters in SEZ. Based on the characterization of optimal firm behavior above, we can establish a set of pre- dictions delivered by the model in response to the reforms introduced by laws 56-07 and 139-11. These predictions are based on a comparative static exercise which entails a marginal reduction in the ESR threshold η .9 We first note that unconstrained SEZ exporters would not modify their pricing or sales choices if the ESR constraint was relaxed, since they are already operating at their natural export intensity. Constrained SEZ exporters would, however, seize the opportunity to in- crease their domestic sales, and would do so by redirecting some of their output to be sold at Home. Thus, constrained exporters would simultaneously lower domestic prices and increase export prices, thereby either achieving their natural export intensity, or, at the very least, operating at the newer lower ESR threshold. Prediction 1 Relaxing the ESR constraint (i.e. a reduction in η ), everything else equal, would induce constrained SEZ exporters to lower domestic prices and increase export prices. This in turn implies an increase in domestic sales and a reduction in export sales for this group of firms. Unconstrained exporters do not change their prices nor their sales in either market in response to the policy change. Lowering the ESR threshold increases the profitability of being based in SEZ for constrained exporters if the incentives available to SEZ firms are maintained after the reform — as laws 56-07 and 139-11 did — thereby making SEZ more attractive locations for firms to operate from. This yields our second prediction, Prediction 2 Relaxing the ESR constraint, everything else equal, would induce some firms export- ing through the national customs regime to operate in SEZ. Taking Predictions 1 and 2 together implies an ambiguous response of exports originating from SEZ to a relaxation of the ESR constraint. On the one hand, export sales of constrained exporters 9 A full elimination of the ESR while maintaining the incentives to SEZ in the model would imply that all firms would prefer to operate in SEZ. This would not happen in reality because there is a limited supply of locations in industrial parks in which SEZ are established and also because firms that wish to operate in SEZ need gain the approval of the Consejo Nacional de Zonas Francas de Exportaci´ on, a public-private body that regulates SEZ. In the context of the model, a fixed cost associated with changing a firm’s status from the national customs regime to SEZ would prevent an equilibrium in which all firms choose to operate in SEZ; following this route would leave our conclusions unchanged. 12 fall in response to a reduction in η . On the other hand, new entry into SEZ would increase total export sales from SEZ. The combination of these two effects, therefore, yields our third prediction: Prediction 3 A relaxation of the ESR, everything else equal, has an ambiguous effect on the value of exports originating from SEZ. The magnitude of the overall effect depends on the relative importance of constrained exporters a-vis unconstrained exporters, as well as on the share of non-SEZ exporters operating in SEZ vis-` that would find profitable to switch to SEZ after the ESR is relaxed, a feature that depends crucially on the distribution of demand shifters Ai pω q across firms. In the remainder of the paper we set out to investigate whether the predictions delivered by our model bear out in the data. 4 Data and Stylized Facts In this section we first describe the data used in the paper and provide summary statistics illus- trating the broad export and import patterns observed over our period of analysis. We next shift focus towards documenting the importance of SEZ and national priority sectors in the Dominican Republic’s export basket and examine whether there are any discernable changes in export perfor- mance, at least at the aggregate level, following the reforms implemented in 2007 and 2011. We conclude by exploring the extent to which the tariff concessions mandated by law 56-07 affected priority-sector firms outside the SEZ. Our study exploits detailed firm-level customs data provided by the Dominican Republic Cus- on General de Aduanas, DGA). The data contains all export and import toms Agency (Direcci´ transaction values by product at the HS 6-digit level and by origin/destination for the period 2006- 2014. Throughout this period, the universe of firms consists of 29,682 firms reporting at least one positive export transaction in at least one of 4,466 HS-6 digit products to at least one of 230 customs territories. Crucially, our data identifies trade flows that originate or reach firms located in SEZ. The Dominican Republic is one of the world’s pioneers in the use of SEZ with a program ongoing for more than 40 years (Burgaud and Farole, 2011). Figure 1 shows that SEZ exports account for more than half of aggregate export value throughout our period of interest — although 13 their importance has dwindled both in terms of value and number of exporters.10 Conversely, SEZ firms comprise a substantially smaller share of the country’s imports and constitute less than 2% of all importing firms. This striking difference underscores the significance of the SEZ regime at the macroeconomic level for the Dominican Republic — the foreign exchange earnings generated by SEZ exports plays a key role in enabling the import flows required by the rest of the economy. Figure 1: Importance of SEZ on Export and Import Transactions, 2006-2014 Export Transactions Import Transactions 80 80 Value Value Number of firms Number of firms 70 70 60 60 50 50 % % 40 40 30 30 20 20 10 10 0 0 2006 2008 2010 2012 2014 2006 2008 2010 2012 2014 Year Year Source: Authors’ calculations based on customs transaction data from the DGA. Zooming in into firm-level export and import performance, Table 2 shows that SEZ firms differ substantially from firms trading through the national customs regime. SEZ firms are larger in terms of the value of their export and import transactions, export and import more products, sell to more destination markets and acquire imported inputs from a wider range of countries than non-SEZ firms. These differences remain when we compare median performance outcomes, which help to ensure that the results are not driven by outliers. These figures are in line with those reported by Fernandes et al. (2016) for countries at a similar stage of development. 10 Schrank (2008) reports that the share of aggregate exports accounted for by SEZ exceeded 80% in the early 2000s 14 Table 2: Importance of SEZ on Export and Import Transactions, 2006-2014 Special Economic Zones National Customs Regime Exports Imports Exports Imports Transaction value per firm Mean 44.98 54.05 5.62 3.32 Median 0.27 4.99 0.05 0.04 Products per firm Mean 6.38 46.32 4.03 9.91 Median 1.00 20.00 1.00 2.00 Destinations/origins per firm Mean 2.58 4.82 1.61 1.47 Median 1.00 2.00 1.00 1.00 Source: Authors’ calculations based on customs transaction data from the DGA. Export and import transaction values are denominated in hundreds of thousands US Dollars. Figures are averaged across the period 2006-2014. Export Performance across SEZ and National Priority Status. We now provide a first look at aggregate export performance according to firms’ SEZ status and whether they operate or not in national priority sectors, which experienced an earlier removal of ESR. Following the discussion in Section 2 regarding the reforms of the SEZ regime in the Dominican Republic, we consider three periods of analysis: pre-reform (2006-07), introduction of national priority sectors (2008-12) and full ESR elimination reform (2012-14). Figure 2: Export Performance across SEZ and National Priority Status 7000 SEZ firms Non-SEZ firms 6000 Total Exports (Millions of US Dollars) 5000 4000 3000 2000 1000 0 Priority Non-Priority Priority Non-Priority Priority Non-Priority 2006-07 2008-11 2012-14 Source: Authors’ calculations based on customs transaction data from the DGA. 15 Figure 2 shows very minor change in total export sales between 2006-07 and 2008-12, both for priority and non-priority sectors and SEZ and non-SEZ firms. This stability of course masks a significant fall in exports during the global financial crisis in 2008-09, which was subsequently followed by a rapid recovery in 2010. The textiles and apparel, leather and footwear industries accounted for approximately one-fifth of the Dominican Republic’s aggregate exports, almost all of which originated from SEZ throughout our period of study. The importance of national priority products in the Dominican export basket has declined secularly since 2000, both because of the erosion of trade preferences and more intense competition by low-wage producers at the regional (e.g. Haiti, Honduras and Nicaragua) and global level (e.g. Bangladesh, China and Vietnam). These two channels have resulted in Dominican exporters experiencing substantial market share losses, particularly in the US market. Non-SEZ firms account for a larger share of export value in non- priority sectors, with their importance increasing over time — a pattern consistent with a long-term trend of greater export dynamism for firms operating outside SEZ identified by Burgaud and Farole (2011) and World Bank (2014). At first pass, the elimination of the ESR for SEZ firms and the extension of fiscal incentives to non-SEZ firms in priority sectors mandated by law 56-07 does not appear to have had a strong impact on aggregate export flows in national priority sectors. SEZ firms still account for the overwhelming majority of priority sector exports in terms of value, with the share of non-SEZ exports increasing by only 1 percentage point, from 1.5 to 2.5%, after the 2007 reform. Extending the fiscal incentives available on SEZ to firms exporting through the national customs regime does not appear to have reoriented exports away from SEZ in priority sectors. The fall in the share of aggregate exports originating in SEZ presented in Figure 1, however, understates their continuing preponderance in the Dominican Republic’s international trade. Table 3 presents the top 20 HS-2 export products during our sample period, as well as the number of exporting firms and the share of exports originating in SEZ. For only one out of the top-10 exported products is the average share of SEZ exports below 40%, and in fact, for 7 out of 10 broadly defined sectors, this share exceeds 90%. Thus, even though the export portfolio of the Dominican Republic has gradually diversified over the last decade (World Bank, 2014), its main comparative advantage sectors are still highly reliant on SEZ. 16 Table 3: Top-20 Export Products, 2006-2014 HS-2 Sector Export # % Exported Value Firms from SEZ Optical and medical instruments 690.33 368.00 96 Precious metals and jewelry 615.27 390.22 42 Electrical machinery and equipment 539.83 492.11 94 Iron and steel 526.37 150.22 5 Tobacco 450.96 233.56 95 Apparel and clothing 396.13 336.67 99 Plastics 263.41 591.33 61 Cotton 263.32 88.56 99 Knitted goods 251.70 266.22 99 Footwear 225.77 114.67 98 Tin articles 178.14 567.67 8 Pharmaceutical products 154.50 177.89 89 Cocoa 153.88 80.00 39 Sugar confectionaries 132.10 100.67 5 Beverages 125.33 161.22 9 Paper 113.84 300.11 78 Mineral fuels 108.34 68.00 2 Misc. edible preparations 87.15 134.00 43 Salt, earth and cement 85.30 115.22 1 Nuclear reactors and related machinery 83.50 617.78 63 Source: Authors’ calculations based on customs transaction data from the DGA. Export and import transaction values are denominated in millions of US Dollars. Figures are averaged across the period 2006-2014. Shaded rows indicate national priority sectors. Although both Figure 2 and Table 3 clearly illustrate the fact that SEZ account for the lion’s share of exports in national priority sectors, Table 4 shows that there is still substantial hetero- geneity — both in terms of export performance and the importance of SEZ — across individual products. In terms of export values, SEZ account for the majority of exports across priority prod- ucts, with the exception of special yarns and carpets (HS 56 and 57 respectively) — both of which have quite low export sales. The number of non-SEZ firms is also more important in low export value products, such as leather goods, yarns and crocheted fibres (HS 42, 56 and 60 respectively). 17 Table 4: Exports of National Priority Sectors Special Economic Zones National Customs Regime Sector HS 2 Export # % Export # % code Value Firms Exports Value Firms Exports Leather 41 11.49 14.44 66 5.83 29.33 34 goods 42 10.03 47.78 93 0.79 74.89 7 43 0.02 1.57 60 0.01 1.80 40 50 0.47 2.89 88 0.06 5.56 12 51 0.41 7.89 80 0.10 3.89 20 52 261.07 60.56 99 2.25 28.00 1 53 0.12 6.00 63 0.07 5.11 37 54 3.09 17.63 94 0.19 8.78 6 Textiles 55 4.17 42.67 96 0.16 15.89 4 and 56 0.53 12.78 32 1.12 36.56 68 textile 57 0.02 2.67 35 0.04 10.11 65 articles 58 6.67 51.44 92 0.57 36.89 8 59 0.52 8.11 96 0.02 9.89 4 60 7.29 21.11 65 3.87 139.89 35 61 249.14 151.33 99 2.56 114.89 1 62 392.28 188.67 99 3.86 148 1 63 42.54 73.33 90 4.54 163.33 10 Footwear 64 221.12 37.33 98 4.65 77.33 2 Source: Authors’ calculations based on customs transaction data from the DGA. Export and import transaction values are denominated in millions of US Dollars. Figures are averaged across the period 2006-2014. Law 56-07 also provided duty-free access for 126 imported inputs to firms in national priority sectors operating outside SEZ. Table 5 presents the top 20 HS-6 liberalized inputs in terms of import value as well as the share of value imported destined to SEZ and the percentage of imports purchased by national priority exporters outside SEZ. This table suggests that the reduction in tariffs mandated by law 56-07 was likely to have a minor impact on non-SEZ firms in priority sectors, given that the most important inputs liberalized were almost exclusively imported by SEZ firms — which could already purchase these products duty-free. The last column of Table 5 also shows that the majority of import value of products that are primarily purchased by non-SEZ firms (with the exception of ammonium sulphate) is accounted for by exporters selling products other than textiles, leather and apparel. Thus, it seems that the liberalized products were either mostly imported by SEZ firms, or were general-purpose rather than inputs specific to priority sector producers. 18 Table 5: Top 20 HS-6 ‘Priority Input’ Imports HS-6 Product HS6 code Import % SEZ % of non-SEZ Value Imports Imports used in Priority Sectors White spirit 271011 1437.24 1 4 Plastic articles nes 392690 278.8 90 2 Soles and heels for footwear of rubber or plastic 640620 22.71 96 2 Bovine leather, vegetable pre-tanned 410411 18.1 99 0 Cartons, boxes & cases, of corrugated paper 481910 16.12 78 2 Other bovine leather, vegetable pre-tanned 410449 14.26 98 1 Footwear uppers and parts thereof, except stiffeners 640610 12.64 99 0 Ammonium sulphate 310221 12.4 0 52 Pigments and preparations based on titanium dioxide 320611 11.63 2 15 Cotton sewing thread for retail 520420 11.56 99 1 Pesticides, rodenticides 380892 11.4 1 23 Composition leather, in slabs, sheets or strip 411510 11.27 99 1 Parts of footwear nes, gaiters and leggings 640699 10.05 99 0 Cartons, boxes & cases, folding, non-corrugated paper 481920 7.55 65 6 Sheet etc, cellular of polyurethane 392113 7.52 85 7 Finishing agents, dye carriers, dressing, mordants 380991 7.18 81 1 Knitted non-pile fabrics 600330 7.03 94 1 Chemical products and preparations 382490 6.93 10 17 Goat or kid skin leather 410622 6.9 95 2 Buckles and clasps 830890 6.71 83 3 Chemical colouring matter 320416 5.79 98 0 Adhesives based on rubber 350691 5.61 38 12 Cotton sewing thread (¤ 85%) 520419 5.44 86 11 Tulles and other net fabric 580410 5.24 98 0 Made up articles of textile material 630790 5.2 57 9 Source: Authors’ calculations based on customs transaction data from the DGA. Import values are denominated in millions of US Dollars. Figures are averaged across the period 2006-2014. Shaded rows indicate HS-6 products that are primarily imported by non-SEZ firms. 5 Empirical Analysis and Results Our main objective in this paper is to determine whether the removal of export share requirements in SEZ, first in national priority sectors and subsequently across all other industries, affected the prevalence of SEZ in the exports of the Dominican Republic. We will also evaluate whether the input tariff reduction experienced by non-SEZ firms in priority sectors had an impact on the import behavior of these firms. 19 Our identication strategy relies on the comparison between priority and non-priority sectors, which experienced the removal of ESR at different points in time. We first describe the construc- tion of our key explanatory variables related to the removal of the ESR for firms in SEZ, which occurred in July 2007 (law 56-07) and June 2011 (law 139-11). As we noted in Section 4 above, we can consider three distinctive periods in our analysis: 2006-2007, when the ESR is in place for SEZ firms in both priority and non-priority sectors, 2008-2011, when the ESR was lifted only for SEZ firms operating in priority sectors and, 2012-2014, when the ESR had been removed for all firms based in the zones regardless of their sector of operation. Since is likely that firms required some time to adapt their behavior to changes stipulated in the two laws, we aggregate the data at a yearly frequency and consider the impact of the first and second regulatory changes starting from the beginning of 2008 and 2012 respectively. Thus, we define the variable POST08t as taking the value 1 from 2008 onwards and 0 otherwise, and likewise, POST12t is a dummy variable that turns on from 2012 onwards. Did the removal of ESR affect the Extensive and Intensive Margin of Exports at the Firm-level? We first investigate if the elimination of export performance requirements affected the export entry and exit behavior of firms within HS-6 products over time, i.e. the extensive margin of exports. Our identication strategy relies on the comparison between priority and non-priority sectors and the different time in which they were subject to the elimination of ESR. We aggregate our data at the firm-product-location-year level, with products being defined at the HS-6 digit level and location indicates whether a firm is located or not in an SEZ. We then estimate the impact of the two waves of reform separately for firms located in SEZ and outside them. Our OLS estimating equation for the extensive margin of exports is therefore given by: Entry/Exitij  β1pPost08t ¦ Priorityj ¦ SEZ q   β2pPost08t ¦ Priorityj ¦ non-SEZ q  t jt   εij t , β3 pPost12t ¦ Non-Priorityj ¦ SEZ q  β4 pPost12t ¦ Non-Priorityj ¦ non-SEZ q  fj   f t   T˜ (8) where i indexes firms, j , HS-6 products, , locations, and t, years. Entryij t is a dummy variable that takes the value 1 if a firm-product-location combination reports a positive export value at time 20 t but not in the previous year. Similarly, Exitij t takes the value 1 if a firm-product-location exports in year t but does not report any export sales in year t   1. As noted above, SEZ is a dummy that takes value 1 if a firm is located in a special economic zone and 0 otherwise, while non-SEZ jt is an HS-2-product-specific linear trend which intends to control for is the opposite image. T˜ broadly-defined time-varying sectoral supply and demand shifts. We estimate the effect of policy changes within and outside the zone independently. To this end, the set of HS-6 digit product fixed effects fj and year fixed effects f t , are set to be specific to location . Including this battery of location-specific fixed effects allows for differential average effects on export performance according to firms’ SEZ status as well as controlling for time-varying differences within and outside the zones. The standard errors in regressions (8) and (9) below are clustered at the HS-6 product level. Table 6 presents our findings regarding the effect of laws 56-07 and 139-11 on the extensive margin of exports for firms according to their SEZ and national priority status. Prediction 2 suggests that inasmuch as the removal of ESR induces more firms to locate in SEZ, we should expect a positive effect on export entry among firms located in SEZ. At the same time, firms’ relocation towards SEZ might also increase the exit rate among firms exporting through the national customs regime. The results reported in columns (1) and (2) of Table 6 provide support for this hypothesis. Export entry increased for SEZ firms in both priority and non-priority sectors after the removal of the ESR constraint — although the effect is statistically significant only for SEZ producers in non-priority sectors. The entry rate into exporting for these firms increased by 10 percentage points following the reform implemented in 2011. We also observe an increase in the exit rate of firms exporting through the national customs regime in priority sectors after 2008 (column 2). This is expected as eliminating the export requirement would intensify the competition faced by non-SEZ firms. Nevertheless, it is somewhat puzzling to observe this response following the 2007 reform, since the government extended some of the tax incentives available in SEZ to priority sector producers outside the zones; in contrast, law 139-11 did not provide any incentives to non-SEZ firms. As we discuss in more detail below when we investigate the input tariff reduction mandated by law 56-07, this dimension of the SEZ reform was largely ineffective in providing incentives to firms outside SEZ. We next investigate the response of the intensive margin of exports. More specifically, we explore how the export value of firms operating in a given location, selling a given HS-6 product, 21 was affected by the elimination of ESR for each of the four types of firms considered in our analysis of the extensive margin above. Equation (9) presents our OLS estimating equation for the intensive margin, where the dependent variable is the log of export value by firm i exporting HS-6 product j from location in year t, and all other variables have been defined above: ln Yij t  β1pPost08t¦Priorityj ¦SEZ q β2pPost08t¦Priorityj ¦non-SEZ q β3pPost12t¦Non-Priorityj ¦SEZ q jt   εij t . (9)   β4pPost12t ¦ Non-Priorityj ¦ non-SEZ q   fij   f t   T˜ The results reported in column (3) of table 6 show that firms operating in SEZ experienced a reduction in the value of individual export transactions at the HS-6 product level after ESR were eliminated, with the magnitude of the effect being quite similar for SEZ firms independently of their sector of operation. Firms exporting through the national customs regime were, on the other hand, largely unaffected along this margin. This result is consistent with Prediction 1 from our model. After the limit on domestic sales was lifted, constrained exporters increase export prices lowering exports sales. However, this average effect encompasses both the change in behavior of constrained exporters, which were previously forced to export 80% of their output, as well as unconstrained exporters, which would have been unaffected by the elimination of the export requirement. In order to disentangle these two different responses, we re-estimate regression (9) on the set of continuing exporters, i.e. those firms that have conducted export transactions in every year of our sample.11 ek´ We do so with the view that perennial exporters are more likely to be unconstrained (B´ es and ozy, 2012).12 The results reported in column (4) show that although the negative effect of the Murak¨ removal of ESR on the intensive margin of exports for SEZ firms remains, it becomes statistically insignificant as we expected. Lastly, we do not find any evidence that the removal of ESR had any discernable effect on the size of export transactions sold by non-SEZ exporters. 11 Notice that since we do not observe the value of firms’ domestic sales, we cannot directly distinguish constrained and unconstrained exporters in SEZ. 12 Interviews with managers at several firms in SEZ carried out in December 2015, reveal that large foreign-owned firms that export permanently sell almost all their output abroad, perhaps with the exception of defective orders that are sometimes sold locally. Interestingly, some of these large multinationals even choose to serve the Dominican market via affiliates located in other CAFTA-DR country members, since by doing so they avoid paying import tariffs that they would otherwise face if they were to sell directly from their SEZ plant. 22 Table 6: Response of Firm-level Intensive and Extensive Margins of Exports to the Removal of ESR Extensive margin Intensive margin Entry Exit Value Value [1] [2] [3] [4] Post08t ¦ Priorityj ¦ SEZ 0.035 -0.024 -0.234** -0.146 (0.025) (0.020) (0.102) (0.361) Post08t ¦ Priorityj ¦ non-SEZ 0.010 0.040** 0.106 0.579 (0.021) (0.026) (0.108) (0.824) Post12t ¦ Non-Priorityj ¦ SEZ 0.097*** -0.039 -0.218** -0.148 (0.021) (0.026) (0.098) (0.286) Post12t ¦ Non-Priorityj ¦ non-SEZ 0.002 -0.016 0.183 -0.331 (0.019) (0.022) (0.151) (0.440) Observations 188,623 170,991 203,137 3,672 R2 0.047 0.019 0.010 0.090 ***, significant at the 1% level; **, significant at the 5% level; *, significant at the 10% level. All regressions are estimated by OLS. Robust standard errors clustered at the HS-6 product level. Specifications [1] and [2] include HS-6 product ¢ location and year ¢ location fixed effects, as well as HS-2 product-specific linear trends. Specifications [3] and [4] include firm ¢ HS-6 product ¢ location, year ¢ location fixed effects and HS-2 product-specific linear trends. Column [3] includes all firms, whereas the regression reported in column [4] is restricted to only include firms that have conducted export transactions in a given HS-6 product in every year of our sample. Did the elimination of ESR affect the importance of exports originating from SEZ? We now present our main conclusion concerning the change in the importance of exports originating from SEZ after the elimination of ESR. As we discussed in Section 3, if the constraint on domestic sales is lifted for SEZ firms and the level of fiscal incentives available in the zones remains unchanged, we would expect that more firms find profitable to be based in SEZ, and this in turn should increase the share of SEZ exports for a given HS-6 product (Prediction 2). On the other hand, if a substantial number of firms in SEZ were constrained by the 80% ESR, we would expect that the elimination of ESR would induce these producers to reallocate some of their sales towards the domestic market, thereby reducing the importance of exports originating from the SEZ (Prediction 1). Therefore, Prediction 3 tells us that, overall, the SEZ reform is expected to have an ambiguous effect on the share of SEZ exports. We now aggregate our data at the HS-6 product-year level and use as our dependent variable the shares of export value originating from SEZ and the number of exporters located in SEZ in a given HS-6 product. We are interested in gauging the effect that the 2007 and 2011 reforms had in 23 products belonging to priority and non-priority sectors. Thus, we focus in the coefficients β1 and β2 in the following OLS regression equation: ShrSEZjt jt   εjt ,  β1pPost08t ¦ Priorityj q   β2pPost12t ¦ Non-Priorityj q   fj   ft   T˜ (10) where j indexes HS-6 products and t indexes years as before; fj and ft denote product and year fixed effects respectively. This set of fixed effects and product trends control both for time-invariant characteristics that affect the attractiveness of exporting a particular product from SEZ, as well as for time-varying secular changes at the sectoral level, which could confound the interpretation of the difference-in-differences coefficients. As in our previous regressions, robust standard errors in regression (10) are clustered at the HS-6 product level. Table 7 presents our estimates of regression (10). Our results confirm the initial impression provided by Table 6 and are also consistent with Predictions 2 and 3 of our model. The share of exports accounted for by SEZ did not change significantly after the reforms. Conversely, we find that scrapping the ESR had a positive and significant effect on the share of exporting firms operating in SEZ. More precisely, the share of firms in priority sectors exporting through SEZ increased by 7 percentage points on average relative to the situation with ESR, while law 139-11 led to an increase of 5.4 percentage points in the share of SEZ exporters in non-priority sectors. The fact that the increase in the share of SEZ exporters is slightly higher in priority sectors suggests that the efforts to extend the incentives available in SEZ to firms based outside them were largely ineffective. Overall, the results reported in Table 7 suggest that the effect of the elimination of ESR on the attractiveness of locating in SEZ dominated the negative effect on the intensive margin of constrained exporters. Thus, granting access to the domestic market to firms in the zones has helped in consolidating their importance in the export basket of the Dominican Republic — at least in the short term. 24 Table 7: Share of SEZ in Export Value and Number of Firms at the HS-6 Product-level Export Value Number of Firms [1] [2] Post08t ¦ Priorityj 0.033 0.069*** (0.024) (0.019) Post12t ¦ Non-Priorityj 0.039 0.054** (0.031) (0.025) Observations 19,141 19,145 R2 0.055 0.109 All regressions are estimated by OLS and include HS-6 product and year fixed effects as well as HS-2 product-specific linear trends. Robust standard errors clustered at the HS-6 product level. ***, significant at the 1% level; **, significant at the 5% level; *, significant at the 10% level. Input Tariff Reductions for Non-SEZ Firms in Priority Sectors. Lastly, we investigate whether the tariff cuts mandated by law 56-07 had any significant impact on the share of imports of these products accounted for by SEZ across HS-6 products, both in terms of value and number of importing firms. Thus, our estimating equation is given by: ShrSEZjt jt   εjt ,  β1pPost08t ¦ Priority Inputj q   fj   ft   T˜ (11) where Priority Inputj takes the value 1 if HS-6 product j import tariff was set to zero under law 56-07, and all other control variables have been defined previously. Notice that since law 139-11 did not extend any import tariff reductions, equation (11) does not include an interaction term for liberalized inputs after 2012. As we discussed in Section 4, Table 5 shows that the most important goods (in terms of total import value) that were made duty-free for non-SEZ firms operating in national priority sectors were either almost exclusively imported by SEZ firms, or, general-purpose inputs imported by a large number of firms in priority and non-priority sectors alike. Thus, we examine whether there is a difference in the share of imports destined to SEZ depending on the importance of these goods in the production of national priority sectors. Since we do not have an input-output table that would allow us to categorize inputs as specific to priority sectors or general-purpose, we follow an alternative strategy and classify each of the 126 HS-6 products liberalized as specific, if the first two digits of their product nomenclature correspond to those of a national priority product. Thus, 25 for instance, the HS-6 product ‘bovine leather’ is classified as a priority input because its first two digits correspond to the leather goods sector, which is listed as a priority sector. Conversely, ‘white spirit’, which is also a priority input, is instead considered to be a non-priority product. Columns (3) and (4) of Table 8 present estimates of the following OLS regression that modifies equation (11): ShrSEZjt  β1pPost08t ¦ Priority Inputj ¦ Priority Productj q  jt   εjt , β2 pPost08t ¦ Priority Inputj ¦ Non-Priority Productj q   fj   ft   T˜ (11’) Column (1) of Table 8 shows an insignificant reduction in the share of liberalized inputs imported by SEZ firms after 2008, and similarly, column (2) reveals that there is no change in the share of SEZ firms importing these goods either. Column (3), however, shows that the reduction in the share of imports of priority inputs destined to SEZ occurred primarily among products that were not used specifically by firms in the apparel, textiles and leather industries. If — as the results in Table 8 imply — non-SEZ firms did not receive a sufficient boost in incentives after the removal of ESR, tougher competition from SEZ firms in the domestic market could contribute to the rise in the share of SEZ exports after the elimination of ESR. Table 8: Share of SEZ in Import Value and Number of Firms at HS-6 Product-level Import Value # Firms Import Value # Firms [1] [2] [3] [4] Post08t ¦ Priority inputj -0.033 -0.004 (0.021) (0.015) Post08t ¦ Priority inputj ¦ Priorityj -0.019 0.034 (0.043) (0.036) Post08t ¦ Priority inputj ¦ Non-Priorityj -0.040* -0.022 (0.022) (0.013) Observations 40,770 40,784 40,770 40,784 R2 0.018 0.029 0.018 0.029 All regressions are estimated by OLS and include HS-6 product and year fixed effects as well as HS-2 product-specific linear trends. Robust standard errors clustered at the HS-6 product level. ***, significant at the 1% level; **, significant at the 5% level; *, significant at the 10% level. 26 6 Conclusions For more than four decades the Dominican Republic has relied extensively on providing fiscal incentives to firms located in special economic zones to promote exports. An instrumental element of this developmental strategy was the requirement that firms located in SEZ had to export at least 80% of their output, thus limiting the extent of competition faced by local producers outside the zones. The impending deadline to make the SEZ program compliant with the disciplines stipulated by the WTO Agreement on Subsidies and Countervailing Measures by December 2015, led to the elimination of export share requirements in the SEZ. A distinctive feature of the reform strategy pursued by the Dominican Republic — which we take advantage of in order to identify the impact of the removal of ESR on the export performance of SEZ — was that the requirements were eliminated in a staggered fashion over time and across different industries. Namely, the export share requirement was first removed for producers of leather, textiles and apparel and leather, the so called ‘national priority’ sectors in 2007, and afterwards for all SEZ producers in 2011. From a theoretical standpoint, ESR distort the allocation of sales across domestic and export markets for exporters taking advantage of the incentives available in SEZ, but that would prefer to sell a higher share of their output locally than what the export requirement allows them to. On the one hand, eliminating the constraint on domestic sales would induce these exporters to increase the prices they charge abroad in order to reach their undistorted export intensity, thus reducing export sales. On the other hand, removing the ESR while retaining the incentives available in SEZ, would encourage firms to locate there, thereby increasing the share of exports originating from SEZ for a given product. Since these two mechanisms work in opposite directions, the question of how did the ESR reform affect the importance of SEZ exports — which is the focus of this paper — is ultimately, an empirical one. Our results show that removing the ESR in SEZ has made them a more attractive location for firms to operate from in the Dominican Republic. Our empirical assessment of the response of export margins to the elimination of ESR shows, in line with our theoretical framework, that the reforms promoted entry into export markets within SEZ, while at the same time producing a reduction in the value of export shipments for existing exporters in SEZ (although the latter effect stops being significant when we focus on continuing exporters). When we look at how the 27 importance of SEZ exports in narrowly-defined HS-6 products was affected by the two waves of elimination of ESR, we find that the share of exporters from SEZ increased between 5.5 to 7 percentage points relative to the situation with ESR, whereas the increase in the share of export value after the reforms was largely insignificant. These findings suggest that the lion’s share of exports (in terms of value) originated in SEZ is accounted for by unconstrained exporters that did not find the ESR constraint binding. Nevertheless, the entry of new exporters after the removal of export requirements more than compensated the reduction in the value of individual export shipments by established constrained exporters, and therefore, we see that the importance of SEZ in terms of the number of exporting firms for a given product, increased following the reforms. In regard to the objective of harmonizing the incentives between firms inside and outside SEZ, we find that the tariff reductions introduced by law 56-07 did not provide a sufficiently strong boost, in terms of the import behavior of firms relying on the national customs regime, to elicit a significant change in the composition of exports in the Dominican Republic. Overall, we conclude that at least in the short-run, making the special economic zones regime compatible with WTO rules on export subsidies has not significantly diminished the Dominican Republic’s reliance on the zones. References Alder, S., L. Shao, and F. Zilibotti (2013): “Economic Reforms and Industrial Policy in a Panel of Chinese Cities,” UBS Center Working Paper Series, Working Paper No. 5. Bagwell, K. and R. W. Staiger (2006): “Will International Rules on Subsidies Disrupt the World Trading System?” American Economic Review, 96, 877–895. Baldwin, R. and S. Evenett (2009): “The Collapse of Global Trade, Murky Protectionism, and the Crisis: Recommendations for the G20,” VoxEU.org Publication. Banco Central de la Repu ´ blica Dominicana (2014): “El Reto de las Zonas Francas con el alisis de Coyuntura Internacional No. 29. Cambio en las Reglas de Juego en 2015,” An´ ´ke Be ´s, G. and B. Murako ¨ zy (2012): “Temporary trade and heterogeneous firms,” Journal of International Economics, 87, 232–246. Boyenge, J. P. S. (2007): “ILO Database on Export Processing Zones, Revised,” International Labour Organization. Burgaud, J.-M. and T. Farole (2011): “When Trade Preferences and Tax Breaks are no Longer Enough: The Challenges of Adjustment in the Dominican Republics Free Zones,” in Special Economic Zones: Progress, Emerging Challenges and Future Directions, ed. by T. Farole and G. Akinci, World Bank. 28 Chao, C.-C. and E. S. H. Yu (2014): Trade-Related Investment Measures: Theory and Appli- cations, London: Imperial College Press. Creskoff, S. and P. Walkenhorst (2009): “Implications of WTO Disciplines for Special Economic Zones in Developing Countries,” World Bank Policy Research Working Paper 4892. Davidson, C., S. J. Matusz, and M. E. Kreinin (1985): “Analysis of Performance Standards for Direct Foreign Investments,” Canadian Journal of Economics, 18, 876–890. Davies, R. B. and A. Mazhikeyev (2015): “The Impact of Special Economic Zones on Exporting Behavior,” University College Dublin Working Paper 15/28. Defever, F. and A. Rian ˜ o (2015): “Gone for Good? Subsidies with Export Share Requirements in China: 2002-2013,” World Bank Economic Review Papers & Proceedings, 29, S125–S144. ——— (2016): “Protectionism through Exporting: Subsidies with Export Share Requirements in China,” CEP Working Paper 1431. Devereux, J. and L. L. Chen (1995): “Export Zones and Welfare: Another Look,” Oxford Economic Papers, 47, 704–713. Evenett, S. J. and M. Wermelinger (2010): “A snapshot of contemporary protectionism: how important are the murkier forms of trade discrimination?” in Rising Non-Tariff Protectionism and Crisis Recovery, United Nations Economic and Social Commission for Asia and the Pacific (ESCAP). Farole, T. and G. Akinci (2011): Special Economic Zones: Progress, Emerging Challenges and Future Directions, Washington DC: The World Bank. Fernandes, A. M., C. Freund, and M. D. Pierola (2016): “Exporter behavior, country size and stage of development: Evidence from the exporter dynamics database,” Journal of Development Economics, 119, 121–137. Fernandes, A. P. and H. Tang (2012): “Determinants of Vertical Integration in Export Pro- cessing: Theory and Evidence from China,” Journal of Development Economics, 99, 396–414. FIAS (2008): Special Economic Zones: Performance, Lessons Learned, and Implications for Zone Development, Washington, DC: World Bank. Hamada, K. (1974): “An Economic Analysis of the Duty Free Zone,” Journal of International Economics, 4, 225–241. ˜ o (2013): “The Rise of the Maquiladoras: A Mixed Blessing,” Heid, B., M. Larch, and A. Rian Review of Development Economics, 17, 252–267. Ianchovichina, E. (2007): “Are Duty Drawbacks on Exports Worth the Hassle?” Canadian Journal of Economics, 40, 881–913. Kaplinsky, R. (1993): “Export Processing Zones in the Dominican Republic: Transforming man- ufactures into commodities,” World Development, 21, 1851–1865. Krugman, P. (1980): “Scale Economies, Product Differentiation, and the Pattern of Trade,” American Economic Review, 70, 950–959. 29 Melitz, M. J. (2003): “The Impact of Trade on Intra-Industry Reallocations and Aggregate Productivity,” Econometrica, 71, 1695–1725. ıguez-Clare, A. (1996): “Multinationals, Linkages, and Economic Development,” Ameri- Rodr´ can Economic Review, 86, 852–873. Rodrik, D. (1987): “The Economics of Export-Performance Requirements,” The Quarterly Jour- nal of Economics, 102, 633–650. ——— (2004): “Industrial Policy for the Twenty-First Century,” CEPR Discussion Paper 4767. Sanchez-Ancochea, D. (2006): “Development trajectories and new comparative advantages: Costa Rica and the Dominican Republic under Globalization,” World Development, 34, 996– 1015. Schrank, A. (2008): “Export Processing Zones in the Dominican Republic: Schools or Stopgaps?” World Development, 36, 1381–1397. The Economist (2015): “Special Economic Zones: Not so Special,” April 3, 2015. Volpe Martincus, C. (2010): Odyssey in International Markets: An Assessment of the Effective- ness of Export Promotion in Latin America and the Caribbean, Washington, DC: Inter-American Development Bank. Wang, J. (2013): “The Economic Impact of Special Economic Zones: Evidence from Chinese Municipalities,” Journal of Development Economics, 101, 133–147. Wang, Z. and Z. Yu (2012): “Trading Partners, Traded Products, and Firm Performance: Evi- dence from China’s Exporter-Importers,” The World Economy, 35, 1795–1824. Waters, J. J. (2013): “Achieving World Trade Organization Compliance for Export Process- ing Zones while Maintaining Economic Competitiveness for Developing Countries,” Duke Law Review, 63, 481–524. Willmore, L. (1995): “Export processing zones in the Dominican Republic: A comment on Kaplinsky,” World Development, 23, 529–535. World Bank (2014): How to Sustain Export Dynamism by Reducing Duality in the Dominican Republic, Washington, DC: World Bank. World Trade Organization (2006): World Trade Report 2006: Exploring the Links Between Subsidies, Trade and the WTO, Geneva: World Trade Organization. Young, L. (1991): “Unemployment and the optimal export-processing zone,” Journal of Devel- opment Economics, 37, 369–385. Young, L. and K. F. Miyagiwa (1987): “Unemployment and the formation of duty-free zones,” Journal of Development Economics, 26, 397–405. ¨n (2016): “Trade Performance of Export Processing Zones,” The ¨ cer, A. and J.-M. Siroe Yu World Economy. 30