_ Wanyu Chung, University of Birmingham; Carlo Perroni, University of Warwick. August, 2020. This is a brief of a research paper first published by CESifo.
Free Trade Areas (FTAs) typically impose origin requirements as a condition for goods originating in a member country to be exportable to another member country without incurring a tariff. These rules of origin (ROOs) are meant to prevent importers of goods originating outside the FTA from using trans-shipments within the FTA as a way of minimizing payments of customs duties, but in practice they can translate into excessive protection of domestic producers of intermediate goods. This has been well documented both theoretically (Krishna and Krueger, 1995; Falvey and Reed, 2002; Krishna, 2005; Bombarda and Gamberoni, 2013) and empirically (Conconi et al., 2018).1
This paper studies how the domestic content requirements imposed by FTAs affect competition and prices. Our contribution is twofold. First, in a model of oligopolistic competition between producers of differentiated intermediate goods, we show that binding and stricter content requirements under oligopoly are associated with higher markups and more firm entry. These effects could dampen the well-documented procompetitive effects of preferential trade liberalization, and should thus be factored in when assessing the pros and cons of alternative preferential trade arrangements.2 Second, we verify our theoretical predictions empirically by focusing on the 1989 CanadaUnited States Free Trade Agreement (CUSFTA), using both Canadian trade data and US Census data. To the best of our knowledge, our paper is the first to document such effects for FTAs both in theoretical and empirical terms.
The broader debate on how different types of trade barriers may produce different effects on market outcomes is an old one, but has traditionally been restricted to the comparison between tariffs and quotas. The main conclusion is that price-based instruments (such as tariffs) and quantity-based instruments (such as quotas) produce the same effects under conditions of perfect competition but not so if competition is imperfect (Bhagwati, 1965).3 Under monopoly, for example, a quota removes a portion of domestic demand faced by a domestic monopolist, but its monopoly pricing power remains unchanged for the residual portion of demand. In contrast, trade in the presence of a tariff fully removes the monopoly power of the sole trader by comparison with autarky.
As a trade barrier, ROOs cannot be readily slotted into either category—they are neither price-based nor quantity-based. We argue here that they indeed amount to a hybrid between the two types of instruments. Specifically, we show that a regional value content (RVC) requirement can result in partial market segmentation, lowering the elasticity of demand for domestic intermediates produced within the FTA. When the market for intermediate inputs is oligopolistic (rather than monopolistically competitive), the lower demand elasticity can be shown to give rise to higher markups. In turn, absent barriers to entry, higher markups encourage inefficient entry. These effects on prices and market structure go beyond those that would be implied by an equivalent tariff barrier—a tariff that has the same effect on the volume of intermediate goods imports—because a tariff does not fundamentally change the elasticity of demand faced by suppliers of intermediate inputs.
Our theoretical analysis generates three main testable predictions. The first is that markups should be higher under a binding domestic content requirement than they are in its absence. The second is that a given content requirement is more likely to be binding the larger is the required level of domestic content, the smaller are Most Favoured Nation (MFN) input tariffs, and the larger are MFN output tariffs. To see why, assume there is absolutely free trade between FTA partners while there are non-zero MFN tariffs applied to non-FTA trading partners. High MFN input tariffs would induce final good producers to opt for inputs of FTA origin irrespective of input requirements, making the input requirement less likely to be binding. On the other hand, if MFN tariffs on final goods are non-negligible, obtaining origin status for final goods and being able to export them at zero tariff to other FTA regions has positive value for FTA producers. Intuitively, a more binding content requirement makes FTA producers of final goods more willing to pay a premium for intermediate inputs originating within the FTA, translating into greater pricing power for oligopolistic producers of intermediate inputs operating within the FTA.
The role origin requirements play in support of oligopolistic markups in turn limits the pro-competitive effects of preferential trade liberalization: absent barriers to entry/exit, the number of intermediate goods producers remains inefficiently higher, and their size remains inefficiently smaller, than would be the case under a preferential trade agreement that does not incorporate ROOs—our third prediction.
To test these predictions empirically, we focus on the 1989 CUSFTA and construct a novel product-level index that measures ROOs restrictiveness based on the input-output linkages in CUSFTA’s rules of origin constructed by Conconi et al. (2018).4 We first use monthly province-level trade statistics from Statistics Canada for the period of 1989-1993, and provide post-CUSFTA evidence that stricter and binding RVC requirements are associated with higher export unit values for Canadian exports to the US in comparison with exports of comparable products to other destinations (by roughly ten percent on average). Such gap can be interpreted as reflecting a differential markup applied by Canadian intermediate goods exporters on sales to FTA producers (who face origin requirements).5 To relate price changes to ROOs in comparison with the pre-CUSFTA levels, we turn to annual US PPI data for manufacturing industries from the US Bureau of Labor Statistics, matched with concentration measures (Herfindahl-Hirschman Index, HHI) from the Economic Census for identifying market structures for the years of 1987 and 1992. In difference-in-difference specifications, we find strong support for our theoretical predictions, and we show that our findings apply to only oligopoly industries, consistent with our theoretical setup.
This paper builds on and contributes to the literature that examines how different trade policies may impact on market power and firm entry. A large body of literature has documented a pro-competitive effect under imperfectly competitive market structures and showed that trade liberalization reduces markups, both theoretically (Melitz and Ottaviano, 2008) and empirically (Levinsohn, 1993; Harrison, 1994; Feenstra and Weinstsein, 2017).6 It has also been well-documented that trade liberalization can lead to exit by the least productive firms exit the market: see Pavcnik (2002) for the case of Chile, and Trefler (2004) for the case of CUSFTA.7 Our contribution here is in showing that binding rules of origin under oligopolistic competition generate the opposite effects in terms of markups and firm entry, and therefore should not be neglected when assessing the impact of FTAs.
Among the papers that examine the effects of rules of origin, most are theoretical papers that focus on the protection of domestic producers of intermediate goods. Ju and Krishna (2002; 2005) are probably the first studies to formalize how ROOs could affect the prices of intermediate goods in the FTA region. In a framework with inelastic supply and perfect substitution between FTA and non-FTA inputs, they document a non-monotonic effects due to demand shifts for FTA inputs, depending on whether het erogeneous firms choose to comply with ROOs. Our paper differs from theirs by studying how ROOs affect trade prices via changes in market power and markups rather than through decreasing returns in production.
Empirical attempts to measure the restrictiveness of ROOs in different industries and evidence of their trade effects have been scarce. One notable example is Conconi et al. (2018) who provide a mapping of input-output product linkages of ROOs in NAFTA and CUSFTA, upon which our ROO measures are based.8 Their focus, however, is on how NAFTA ROOs lead to significant reductions in imports of intermediate goods from third countries relative to NAFTA trading partners, rather than on the effects of ROOs on prices. A study that examines effects on both trade volumes and prices is Romalis (2007), which documents a substantial boost in trade between NAFTA partners but only a modest increase in relative output prices of traded goods between NAFTA members versus the rest of the world for very protected sectors with high MFN tariffs. His structural estimations, however, do not account for ROOs. Our paper complements these studies by providing evidence on how ROOs affect trade prices for intermediates, employing a novel measure of ROO tightness that accounts for variations in the levels of protection across traded goods.9
Rules of origin in FTAs generate a degree of market segmentation that boosts the market power of oligopolistic producers of intermediate goods, translating into higher markups and higher prices even in the absence of decreasing returns to scale in production. We should then expect to observe higher markups under a binding domestic content requirement than they in its absence. In turn, domestic content requirements should be more likely to be binding the tighter is the requirement, the smaller are MFN input tariffs, and the larger are MFN output tariffs. These predictions are borne out by evidence in Canadian export data and US PPI data.
The effects of ROOs on market power and markups imply that origin requirements have the potential of generating efficiency costs that go beyond those associated with the substitution of domestic intermediates for imported intermediates by domestic producers (as measured by the trade-barrier equivalent effect of ROOs). These additional efficiency costs stem from inefficient firm entry (and potentially inefficient selection of heterogeneously productive intermediate producers) due to ROOs sheltering domestic oligopolists from foreign competition.
1 For welfare implications, see Krueger’s (1997) survey on FTAs with ROOs versus Customs Union; and Brenton and Manchin (2003) for a discussion of implications for small developing countries such as the Balkans’ FTAs with the EU. Lloyd (1993) also argues that using a tariff on value added produced outside the FTA would be more efficient than using ROOs.
2 See Krueger (1999) for a review of the relevant literature.
3 Krishna (1989) has also shown how quantity-based trade barriers can facilitate collusion.
4CUSFTA came into force in January 1989 and was superseded by the North American Free Trade Agreement (NAFTA) in 1994 with the addition of Mexico. Our choice of CUSFTA over NAFTA is motivated by the fact that over 95% of NAFTA’s rules of origin were already in place in CUSFTA (Conconi et al., 2018). Another advantage is that it enables us to draw a clear distinction between preferential and external MFN tariff rates, as during our sample period of 1989-1993 both Canada and the US had no other FTA partners (with the only exception being the 1985 US-Israel agreement).
5 It should be noted that although Canada is a much smaller economy than the US, it is the largest export market for US producers—US exports to Canada account for 21.52% of total US exports in 1989 and the share remains steady to date at around 20%—and Canada is also one of the largest suppliers to the US —US imports from Canada account for roughly 18% of annual total US imports from 1989 to 2002 before slowly dropping to about 12% in 2018. This means that origin requirements do matter to US producers of final goods.
6 Other studies that have contributed to this debate are Cox and Harris (1985), Head and Ries (1999), and Caliendo and Parro (2015).
7 For a survey of both theory and empirics on heterogeneous firms and trade, see Melitz and Redding (2014).