مشوق ها برای ایجاد مشاغل: یارانه منطقه ای، سیاست تجاری ملی و سرمایه گذاری مستقیم خارجی

A national authority wishes to attract foreign direct investment (FDI) to create local jobs. We analyse the optimal national trade policy when local authorities might offer subsidies to convince a multi-national enterprise (MNE) to invest in their jurisdiction. With centralised decision-making or with allocation of investment to particular localities, the central authority’s optimal policy is to use a high tariff to avoid payment of any subsidy to the MNE. Despite this, some socially undesirable (but locally desirable) FDI cannot be avoided. If local authorities compete to offer subsidies to attract local investment, then the central government’s optimal policy is to try to discourage FDI by choosing a low tariff. Despite this, some socially undesirable – and even locally undesirable -- FDI prevails. We conduct our analysis both assuming an upper bound on tariffs, as would be consistent with trade liberalisation, and allowing tariffs to vary freely. The effect of increasing trade liberalisation depends heavily on the system of granting local subsidies: if the system is centralised, trade liberalisation decreases the range of parameters for which FDI occurs; if the system is decentralised and competitive, it increases this range.

Regional policy to attract foreign direct investment (FDI) and generate new jobs has been prominent in the recent discussion of how to stimulate local economies and relieve the effects of the global recession.1 Jones and Wren (2008) note that, under European Union state aid rules, regional grants are one of the few means by which states can attract FDI. Where serious underemployment exists, for example, economic incentives are permitted to attract foreign firms as a way of resolving underemployment problems. Indeed, these authors comment that the UK and France devote half their regional policy budgets to financial incentives to attract FDI. Many countries have similar local stimulus policies aimed at attracting foreign investment. A UNCTAD (2000) global survey notes that nearly all countries offered incentives targeted at specific sectors, while seventy per cent of countries offered regional incentives. In many cases regional and sectoral incentives were integrated, so that only certain sectors received incentives in certain regions. More generally, these incentives take a variety of forms and may be offered over time or as a lump sum to assist with entry. Davies (2003) and OECD (2008) indicate that such incentives can affect FDI location decisions significantly.2 Offering more detail on this for the case of the UK, Ernst and Young (2011) finds that tax/subsidy benefits, supporting infrastructure investments, and low administrative requirements are all important factors in the decision of firms to locate in a region or not. A major reason, also singled out in the report, for a state to offer these policies is employment gains, with 21,000 jobs created by FDI in the UK in 2010. As noted by the UNCTAD (2000) survey, in a federal system the package of incentives offered to the investor may include central as well as region-based incentives, while the process of agreeing a package may involve differing degrees of competition among regions. Such competition can create windfall benefits for investors: the report cites the case of Mercedes-Benz, which wished to establish a new car plant in the United States and contacted six states before deciding to accept a (generous) location package from Alabama. Similar competition among states to attract a Ford Motors assembly plant occurred in Brazil. The UNCTAD (2000) report goes on to enumerate an exhaustive list of regional policies towards FDI, illustrating that different countries have chosen different degrees of centralisation. Roughly speaking, the US and Europe seem to take a relatively decentralised approach (although this varies by country), many developing countries seem to take the approach of designating a limited number of regions (sometimes only one) that are allowed to offer the incentives without internal competition among regions, and some smaller countries (such as Singapore) take a purely centralised approach, where FDI packages can only be obtained from the national government. When a region is the designated destination, the actual negotiations for the incentive package can be delegated to the local authority.3 Jones and Wren (2008) note that centralisation and the degree to which competition is permitted among regions within a country can also vary over time, documenting the vacillations in the UK system.4 FDI location decisions are affected by more than just regional incentive policies, however. A recent OECD report stated that “Trade policy is one of the main determinants of foreign firms in their investment decisions…High barriers to imports can include tariff-jumping FDI – FDI as an alternative to trade.”5 Even in federalised countries, trade policy typically is in the hands of the central government. For federal governments concerned that competition among regions can dissipate the rents that would otherwise accrue to the country from FDI, trade policy as a tool to avoid this destructive competition is one way forward. Intuitively, trade policy can deal with the problem of excessive local bidding in two ways: first, in setting high tariffs the central government can decrease the “bargaining power “of the multinational enterprise (MNE). This policy does not discourage FDI, but it can decrease the rents captured by the firm in the bidding. Another approach is to lower the tariff so as to make the (local) incentives required to attract the FDI prohibitively high for the region(s). This policy potentially eliminates FDI entirely in favour of imports, but also eliminates costly subsidy competition in the process. Where local subsidies would mount to levels that outweigh the country’s gains, this can be a better choice for the nation as a whole. This paper explores this intuition. We study whether and how trade policy can be used effectively with incentives when incentives can be decentralised by region and where regions may or may not be allowed to compete for the FDI. Following Brander and Spencer (1987) and the sense of the literature we have quoted as motivation, we postulate that FDI can increase local levels of employment. A single MNE considers investment into a country (or group of countries). Local authorities try to attract the foreign firm by offering subsidies, which can be thought of broadly in our model as any package of incentives to attract the firm (involving tax breaks, infrastructure investments and so on). Trade policy takes the form of a per unit import tariff set by the central government. In choosing the tariff, the central authorities take into account its effect on the bidding behaviour of the local authorities and the investment decision of the MNE. Our first result is that, if both the trade and FDI attraction policies are centralised, FDI only occurs when it raises the country’s welfare. FDI is induced optimally through a high tariff so that no subsidies are paid. This is consistent with the first of the two mechanisms outlined above: the tariff has the advantage of affecting the decision to locate but also the “bargaining position” of the firm, since it affects the attractiveness of the alternative of exporting. We next consider the fully decentralised case where different regions compete for FDI. The crucial effect of this competition among regions is that the central government can no longer induce subsidy-free FDI by setting a high tariff. To the contrary, by fully committing the firm to the FDI route, a high tariff can increase the level of subsidy offered by the states in their attempt to compete for the jobs that the foreign firm surely will create in one of the local jurisdictions. In this case, then, the central government may find it optimal to avoid socially undesirable subsidized FDI by setting a low tariff, and so provide an incentive for the firm to switch to exports. By improving the outside option of the firm, the central government makes FDI a more expensive proposition for the localities, potentially making FDI prohibitively expensive. The central government curtails excessively expensive subsidy competition by lowering the tariff, which generates substitution into imports but reduces subsidy expense. Finally, we consider the case where FDI is assigned to a unique zone within the country, which is allowed to provide subsidies for FDI but which does not compete with other regions. We find that in this case, the only difference with the fully centralised case may occur in the upper regions of costs, where the tariff may optimally be lowered to prevent subsidies’ being offered. Overall, when we compare regimes, we find that the range of levels of production efficiency for which FDI occurs under the optimal trade policies is larger with full centralisation or non-competitive bidding than with competitive bidding. Competitive bidding makes FDI socially less desirable because FDI is associated with the payment of positive subsidies. Since, for high cost levels, the government can effectively avoid undesirable FDI by setting a low tariff (so that the firm exports, by preference), the equilibrium range of FDI is curtailed. Unsurprisingly, competitive bidding leads to higher levels of equilibrium subsidies. On the other hand, tariffs with centralisation are at least as large as with competitive bidding: a centralised system relies on high tariffs to reduce subsidy payments, whereas a decentralised system relies on low tariffs to achieve the same end. The recent debate about the benefits of a US-European free trade zone has brought into relief the continuing significance of even modest tariffs. While some markets involve much higher tariffs than one finds between the US and Europe6 , tariffs cannot be set in an unconstrained way under WTO rules. It is important, then, in order to broaden the scope of the paper beyond our object of exploring the interaction of unconstrained trade policy and FDI policy, to derive the effect that tariff caps place on the optimal policies we have just outlined. We therefore proceed to impose a maximum tariff level and examine the effect of progressively reducing these tariffs on FDI and incentive levels. We find that the effect of trade liberalisation depends crucially on the institutional regime we consider. In the fully centralised case, a tightening of tariff caps leads to higher subsidy levels for the FDI that occurs, as it reduces the effectiveness of tariffs as a tool to lower subsidy payments, but to less FDI overall since FDI becomes less attractive for the host country as it becomes more expensive in subsidies. Hence, both the amount of FDI and the instrument balance to induce it change. With competition among regions, however, moderate caps may affect neither subsidy levels nor equilibrium FDI patterns since the optimal policy involves lower tariffs in any case. On the other hand, more drastic trade liberalisation increases the range for which FDI subsidies are observed and can increase the range over which FDI occurs. This suggests that local governance is a factor to consider when determining whether trade liberalisation is likely to impact on FDI decisions. While the extension is useful, it is also important to realise that when we refer to “high” tariffs, we simply mean tariffs that are high enough to make a firm choose FDI over exports. Such “high” tariffs can therefore be quite small in absolute terms, in which case our main analysis without tariffs remains the most pertinent. We also briefly consider how our results might change when we allow for the use of per unit subsidies rather than lump sum subsidies. Output related subsidies create two main additional effects. Firstly, higher subsidies lead to higher output and hence higher employment. This additional employment benefit is perfectly internalised by the authority that offers the subsidy, be it federal or local. As such, it does not affect the nature of our results. The second effect is that output related subsidies also lead to lower output prices, which benefits all consumers. Since local authorities only care about local consumers, this creates a tendency for subsidies to be lower with decentralised decision making than with centralised decision making. This mitigates the tendency for competing local authority to offer subsidies that are higher than would be optimal for the country as a whole. Finally, we argue that the main conclusions of our analysis would still hold in a world where there are several MNEs that can serve the home market through exports and FDI. An interesting collateral result of this discussion is that, in the intermediate case where subsidy setting is decentralised but there is no competitive bidding between states, it is optimal to “steer” all firms that wish to invest in the same industry toward the same region. This offers a possible explanation for the often observed regional concentration of industries that does not depend on the existence of any favourable local conditions (such as vertical linkages or factor costs) or any network effect. Trade policy, FDI, and tax competition have been treated extensively in the literature. Trade policy and tax competition have been examined jointly by Horst (1971) and Janeba (1996) but the tax competition occurs between two different countries that also can set their own trade policies. In contrast, we focus on regional tax/subsidy competition within a single tariff-setting country. Decentralisation in the presence of tax competition among countries has been studied by Wilson and Janeba (2005), with the result that decentralisation can improve welfare by serving as a commitment that changes the strategic behaviour of the competing countries, but this paper does not interact tax with trade policy.7 While Brander and Spencer (1987) consider trade and tax policies in a setting where FDI generates local employment, they do not introduce competition among local authorities. Also, the timing of their policies differs from ours: while they assume that the decision to enter a market is made before tariffs and taxes are set, we assume the contrary in order to focus on the incentive effects on firm location of these policies.8 Trade and FDI policies have been interacted in other papers.9 Blanchard, in a series of papers (2007, 2010 and Blanchard and Matschke (2012)), explores this interaction both theoretically and empirically, although she does not investigate the role of federalism or employment objectives as we do.10 Blanchard and Matschke (2012) provide some empirical support for a relation between offshoring and preferential trade agreements. Vézina (2010) argues empirically that unilateral tariff cutting in Asia, 1988-2006, was driven by FDI competition for intermediate goods, although she does not observe the same relation for consumer goods. Pflüger and Südekum (2009) also provide empirical results on the relation between entry subsidies and trade openness, where entry barriers are measured as non-tariff barriers and openness is measured as export plus import share relative to GDP. They find a U-shaped relation between trade openness and effective entry cost. Our setting generates both a potentially positive and negative relation between tariff level (“openness”) and entry cost (“subsidy”), but the sign of the relation depends on local governance suggesting that internal governance should be controlled for in this kind of empirical study. The closest paper to ours is Raff (2004), who examines the interaction between a local tax instrument (profit taxes) and trade policy. Raff’s paper considers the interaction between tax and trade policy, deriving conditions on when FDI occurs and when it is or is not welfare-improving. While the two papers are similar at this general level, they depart substantially on their focus, institutional framework, and instrument choice. Raff examines the effect of customs unions and free trade areas on FDI, where his focus is on which system should be chosen in a first stage of his game among three countries. This difference yields a different modelling strategy where the baseline case is that of three completely independent countries that may set profit taxes and trade policy independently. Our fully centralised case is the polar opposite, then, of his point of departure, and at no point can the regions in our paper set an independent trade policy. While Raff’s question concerns what sorts of “constraining” agreements the countries should enter into and their relative advantages, in our framework the institutional structure is fixed and we ask how local and national policies interact. Hence, a main point of his paper is that a customs union can serve to coordinate local trade policy. This is not a concern of ours. In terms of instrument, states control a profit tax in Raff’s model, with the associated benefits from locally generated profits rising monotonically with lower costs of production; in our framework the local benefit is not profits but employment, which is related non-linearly to production costs. This generates a contrasting benefits profile, where lower cost production is not necessarily more beneficial since it could mean lower local employment. This tension between efficiency and desirability is not present in his framework. Further, he assumes that an initial asymmetry between the production costs is possible in the two states whereas we consider – in the first instance of the text – either very intense competition between symmetric states or very light competition where FDI is assigned to a single state. On top of this, our framework is that of bargaining rather than Nash equilibrium, so that the function of the tariff to affect the “bargaining position” of the foreign firm, which is an emphasis of our paper is not a concern of his. Despite these differences, the papers can be linked: the fully decentralised case in our paper generates effects that could be present in a symmetric version of Raff’s customs union case, even though his actual emphasis is on asymmetric structures. Our decentralised case without bidding can be seen as generating effects that could be present in a highly asymmetric version of Raff’s customs union case, where one region is so high cost that it is irrelevant to the location of investment. Hence, while the structures of the two papers are different enough that full nesting would not be possible, the sense of this paper is to study cases that could be viewed as “limits” of the asymmetric structure adopted by Raff. Raff’s and our paper should be viewed as complementary, then, in the sense that they investigate the interaction of trade and tax policy under different combinations of the governmental level at which policy instruments are set, and different sources of local benefits from FDI. The rest of the paper is organised as follows. The basic model is presented in section 2. The baseline case where trade and FDI policies are centralised is solved in section 3. Section 4 analyses the cases of full-fledged bidding between local authorities, and section 5 compares the centralised and decentralised regimes. Section 6 analyses the case of decentralisation without bidding competition. Section 7 revisits these three cases in the presence of tariff constraints. Section 8 examines the case of structured subsidies, which can affect both the efficiency of the firm under FDI and can even amount to a negative lump sum paid upon investment. Finally, section 9 discusses the robustness of our results and proposes directions for future research.