when institutional reform is embedded in trade ... · when institutional reform is embedded in...

When Institutional Reform is Embedded in Trade

Liberalization: Evidence from Chinese Exporters∗

Amit K. Khandelwal†

Columbia Business School & BREAD & NBER

Peter K. Schott‡

Yale School of Management & NBER

Shang-Jin Wei§

Columbia Business School & NBER & CIER

First Draft: October 2010

This Draft: March 2011

Abstract

When the removal of trade barriers coincides with the elimination of inecient institutions

created to manage them, the gains from trade increase. We investigate the change in produc-

tivity associated with the removal of quotas on Chinese textile and clothing exporters and the

concomitant elimination of the institution which allocated quotas. When quotas were abolished in

2005, Chinese export value and quantity surged and export prices declined. We show that these

responses are due predominantly to the extensive margin: entrants gained market share at the

expense of incumbent state-owned enterprises, and they entered with relatively low prices. These

reactions are inconsistent with an ex ante assignment of quotas on the basis of rm productivity.

We estimate that roughly half of the overall productivity gain among China's textile exporters

following quota removal is due to the elimination of the quota licensing institution.

∗We thank JaeBin Ahn, David Atkin, A.V. Chari, Thomas Moore, Siddharth Sharma, Nina Pavcnik and JonathonVogel, and seminar participants at the University of Toronto for helpful comments and suggestions. Di Fu providedexcellent research assistance.†Uris Hall, 3022 Broadway, New York, NY 10027, tel : (212) 854-7506, fax : (212) 316-9219, email :

[email protected]‡135 Prospect Street, New Haven, CT 06520, tel : (203) 436-4260, fax : (203) 432-6974, email : [email protected]§Uris Hall, 3022 Broadway, New York, NY 10027, tel : (212) 854-9139, fax : (212) 316-9219, email :

[email protected]

1

Embedded Institutional Reform

1 Introduction

Standard models of international trade generally predict small welfare gains from trade liberalization(Arkolakis et al. 2010). Empirical research, on the other hand, often nds substantial increases inproductivity or income coinciding with the removal of trade barriers (e.g., Pavcnik 2002 and Feyrer2010). One explanation for this discrepancy is that removing a particular tari or non-tari barrier totrade also eliminates other (un-modeled) policy distortions that evolved to manage the trade barrier.The welfare losses associated with taris, for example, can be amplied by corrupt customs agents orbureaucratic red tape that substantially increases the time goods spend in transit.1 If such policiesinuence how resources are allocated among existing rms, or favor incumbents at the expense ofentrants, they can have a sizable eect on aggregate outcomes. Trade liberalization that removes boththe trade barrier and the accompanying distortions can yield gains that are larger than the predictedbenet of removing just the trade barrier.

This paper estimates the productivity gain to China from the removal of a particular trade barrier,export quotas, and decomposes that gain into two parts: that which is due to the removal of thetrade barrier itself versus that which is accounted for by the removal the (inecient) export licensinginstitution that assigned the quotas. We analyze China's textile and clothing industry before and afterthe January 2005 expiration of the global Agreement on Textiles and Clothing, previously known (andreferred to in this paper) as the Multiber Arrangement (MFA). Under the MFA, exports of textileand clothing products by China and other developing economies to the United States, the EuropeanUnion and Canada were subject to quotas. In China's case, the licenses permitting rms to export aportion of the country's overall quota were distributed by the government according to a complex setof rules that may not have emphasized eciency. We use rm-level Chinese trade data to examinehow the distribution of textile and clothing exports changed within and across rms as quotas wereremoved, and to gauge whether these changes are consistent with an allocation of quotas to the mostproductive rms prior to their removal.

Our assessment of the eciency with which China assigned export licenses is guided by a modelof ecient allocation. This model is adapted from Irarrazabal et al. (2010), who introduce specic(i.e., per-unit) taris into the heterogeneous-rm framework of Melitz (2003) and Chaney (2008).Here,we interpret the specic tari as a quota license fee which rms must pay in order to access restrictedforeign markets. We assume that the government does not know the productivity of rms, but that itachieves ecient allocation by assigning export licenses via this (common) license fee. Firms self selectinto the quota-constrained export market based on their productivity, as only the most productiveexporters remain protable net of the fee. We also consider a model of inecient allocation, wherethe government assigns export licenses to politically favored rms. In that model, the most productiverms do not necessarily receive quotas, and the contribution of the extensive margin to export growthfollowing the end of the MFA is larger than under ecient allocation.

In the ecient allocation model, the exports of the most productive incumbents jump dispropor-tionately once quotas are removed. This asymmetric reaction by the intensive margin is driven byremoval of the per-unit license fees, which impose a greater distortion on high-productivity rm's low-priced exports than low-productivity rm's high-priced exports. The removal of export quotas mayalso cause less-productive rms to enter the export market: because obtaining a costly export licenseis no longer necessary, relatively unproductive rms may nd it protable to export the previouslyconstrained goods. This potential contribution by the extensive margin depends upon the density ofhigh-productivity rms. If they are suciently numerous, the aggregate price decline associated withtheir post-quota growth can shut low-productivity entrants out of the export market or even inducethe lowest-productivity incumbents to exit the export market.

1Djankov, Freund and Pham (2010) document substantial variation across countries in the time required to export astandard container. Sequeria and Djankov (2010) show that this variation is driven in part by institutions arising fromtrade barriers: rms choose to export through more distant ports if customs agents at nearby ports are more corrupt.Kaufmann and Wei (2000) show that, contrary to the hypothesis that bribery greases the wheels of commerce, red tapeand bribery tend to be positively correlated.

2


Empirically, we employ a dierence-in-dierences strategy to examine how Chinese exporters reactto the removal of quotas in 2005. In particular, we compare the pre- and post-January 2005 behaviorof rms exporting quota-constrained textile and clothing products to exporters of very similar textileand clothing products that are exported quota free. This comparison isolates the eects of potentialinecient quota allocation from other factors that aect Chinese textile and clothing exporters morebroadly. Exports of cotton slips (HS 62081920) to the United States, for example, were subjectto quotas in 2004, while exports of silk slips (HS 62081910), were not. Contrasting their growthin the years before and after quotas are removed allows us to control for shocks to supply, such asprivatization, or shocks to demand, such as changes in the preferences of consumers, that are commonto both goods.

As documented by Harrigan and Barrows (2009) and Brambilla et al. (2010), China's exports ofpreviously constrained textile and clothing products jump following quota removal, while their exportprices decline. In contrast with the implications of the ecient allocation model, however, we ndhere that both trends are driven largely by the extensive margin. Furthermore, several trends suggestthat entrants are more productive than incumbents. First, entrants' prices are on average 27 percentlower than incumbents, and account for two-thirds of the overall 22 percentage point decline in relativeprices. Second, incumbents with the highest market share under quotas experience the largest declinein market share when quotas are removed. Under ecient allocation, these incumbents possess thehighest productivity and therefore should have beneted disproportionately from the removal of licensefees. Finally, entrants emerge primarily from the private sector, and gain market share at the expense ofstate-owned enterprises (SOEs). As SOEs are well-known for their relatively low productivity (Dollarand Wei 2007, Brandt, Tombe and Zhu 2010), this trend suggests China's quota-licensing institutionfavored relatively unproductive rms.

Back-of-envelope calculations suggest that the replacement of SOEs with privately owned rmsfollowing quota removal resulted in an 18 percent increase in aggregate TFP. This increase is substan-tial relative to the 9 percent growth implied by the decline of SOEs among exporters of non-quota-constrained textile and clothing goods in the same year. It is also large relative to the average 4 percentTFP growth of all non-agricultural rms in China over the 1998 to 2007 period found by Brandt andZhu (2010). We quantify the relative contribution of the removal of the licensing institution using nu-merical solutions of the model that estimate productivity growth after quota liberalization assumingboth ecient and inecient ex ante allocation. We nd that slightly less than half 44 percent of the overall productivity gain can be attributed to the elimination of licensing, with the remaining56 percent due to the removal of the quotas themselves. These results suggest that gains from tradeliberalization may be amplied substantially by the removal of inecient institutions that grow uparound them.

The results of this paper contribute to literatures in both economic development and internationaltrade. Our analysis of how trade aects institutions complements research by Acemoglu, Johnson andRobinson (2005), who argue that the rise of international trade after 1500 increased the power of themerchant class and led to changes in institutions that increased their private property rights. Here,we provide an example of embedded institutional reform: how the removal of trade barriers can leadto the automatic elimination of inecient institutions that grew up around them. By quantifyingthis distortion, we contribute to the theoretical literature on the welfare consequences of poor policyimplementation under the MFA (Krishna and Tan 1998),and Anderson's (1985) analysis of similarinecient allocation of U.S. cheese quotas in the early 1980s.

Our ndings also relate to the growing set of papers that use micro-data to estimate the eectsof market distortions on rms.2 This research generally focuses on existing rms (i.e., the intensivemargin).3 Here, we focus on the extensive margin, and also identify misallocation using weaker as-sumptions. Hsieh and Klenow (2009), for example, assume identical production functions across time

2See, for example, Hsieh and Klenow (2009), Dollar and Wei (2007), Restuccia and Rogerson (2010), Alfaro, Charlton,and Kanczuk (2008), Midrigan and Xu (2010), and Petrin and Sivadasan (2010).

3A recent exception is Chari (forthcoming) who analyzes the aggregate productivity eects of rm entry and sizerestrictions under India's industrial licensing policy.

3


and countries in their comparison of rm productivity distributions in the United States, India andChina. Cross-country comparisons in Alfaro et al. (2008) and Restuccia and Rogerson (2010), on theother hand, assume both that the U.S. allocation of factors is distortion free and that entrepreneurialability is drawn from the same distribution across countries. However, if entrepreneurial ability isshaped by the economic environment (such as the quality of educational institutions), the distributionneed not be identical across countries.4 In the dierence-in-dierences strategy used here, by contrast,we only require the assumption of identical technology and entrepreneurial ability across similar typesof textile and clothing products within China, e.g., silk versus cotton slips.

The eect of distortions on the extensive margin is studied most widely in the context of creditconstraints in developing countries (Banerjee and Duo, 2005). Banerjee and Duo (2004), for example,use an exogenous change in the supply of credit to specic rms to identify constraints on obtainingcredit among Indian rms. Their results suggest the existence of talented entrepreneurs who areunable to borrow from the formal banking sector. Recent theoretical contributions to this literaturehave shown that the potential eect of this extensive-margin misallocation on aggregate productivitycould be quite large.5 We nd empirical evidence for these large eects in the context of a preciselydened government institution.

The rest of the paper proceeds as follows. Section 2 presents a model of ecient quota allocationthat is used to guide the empirical analysis. Section 3 presents a brief summary of the MultiberArrangement. Section 4 performs the empirical analysis. Section 5 explores alternative explanationsfor our ndings. Section 6 decomposes that gain into two parts: that which is due to the removal ofthe trade barrier itself versus the part accounted for by export licensing institution that managed theallocation of quotas. Section 7 concludes.

2 Theory

In this section we outline a simple, ecient-allocation model of exporting under quotas to guide ourempirical analysis. We use this model to derive rm-level implications for how export quantities andprices respond to the removal of quotas assuming quota licenses are allocated to the most productiveexporters.

The model delivers two key results. First, the removal of quotas induces less productive rms toenter the export market. Second, even with this entry, the preponderance of export quantity increasesand export price declines is accounted for by incumbents. The intuition for these results is straightfor-ward. With the elimination of quotas, potential exporters whose costs inclusive of the license fees werepreviously too high to attract enough foreign consumers to overcome the xed costs of exporting cannow enter the export market. However, the removal of license fees exerts a disproportionately largeeect on low-price (high-eciency) rms than high-price (low-eciency) rms because they representa larger fraction of high-eciency rm's low prices. In demonstrating these implications, we employnumerical solutions where analytic results cannot be obtained.

2.1 Exporting Under Quotas

Our model is a re-interpretation of Irarrazabal et al. (2010), which analyzes exporting by heterogeneousrms in a trading system where importing countries make use of both specic (i.e., per unit) and ad

valorem taris. As in Demidova, Kee and Krishna (2009), we interpret quota license fees as equivalentto per-unit increases in the cost of exporting.

Irarrazabal et al. (2010) is an N -country version of Melitz (2003) that collapses to Chaney (2008)when specic taris are set to zero. We assume that in order to export a quota-bound good from origin

4The fact that entrepreneurs in developing countries often do not adopt best practices (Bloom and Van Reenan 2007)may indicate that these technologies are slow to diuse across countries.

5Banerjee and Moll (2010), for example, model misallocation due to nancial frictions that prevent entrepreneursfrom entering markets, while Buera and Shin (2008) and Buera, Kaboski and Shin (2010) quantify the role of nancialconstraints on productivity and growth in a related calibration exercise.

4


country o to destination country d, rms must pay aod > 0 per unit exported as well as an ad valorem

tari τod > 1 of the value of the product exported. Productivities are drawn from the distributionG(ϕ) with density g(ϕ), and the price of variety ϕ in export market d is given by

pod(ϕ, aod) =σ

σ − 1ωo

(τodϕ

+ aod

), (1)

where σ > 1 is the constant elasticity of substitution across varieties and ωd is the wage in the homecountry.6 The corresponding export quantities are given by

qod(ϕ, aod) =

(σ

σ − 1ωo

)−σ (τodϕ

+ aod

)−σ(Pd)

σ−1Yd (2)

where Pd and Yd are the price index and expenditure in the destination market, respectively.In this ecient allocation model, a Walrasian auctioneer nds the license price by equating the

aggregate demand for exports,´ϕ∗qod(ϕ, aod)g(ϕ)dϕ, and the size of the quota, Qod, where the latter

is determined exogenously through bilateral negotiations between the origin and destination countriesThis setup is similar to Anderson (1985), who demonstrates that the most ecient allocation of quotasimplies a common license price. Lower license prices connote less restrictive quotas, and vice versa.

A productivity cuto,

ϕ∗od =

[λ

(fodYd

) 11−σ Pd

ωoτod− aodτod

]−1

, (3)

determines the marginal exporter who is indierent between paying the xed costs of exporting and

remaining a purely domestic rm. In equation (3), λ =(σ−1σ

)σ

11−σ is a constant and fod is the xed

costs of exporting from country o to country d.As in Irarrazabal et al. (2010), there is no closed-form solution for the price index Pd = Pd(ϕ

∗od)

when aod > 0. With Pd xed, it is easy to verify that a lower license price implies a lower cuto for

exporting,dϕ∗oddaod

> 0. If the exporting country is large, the foreign price index will fall with quotaliberalization and, depending on the magnitude of the decline, the productivity cuto ϕ∗od could rise,implying that the least-productive exporters exit. To x ideas, we assume for the remainder of thissub-section that the exporting country is small relative to the importer and therefore that the priceindex is insensitive to changes in the license fee.7 In the numerical analysis below, we relax thisrestriction.

When license fees are zero, the ratio of output quantities between two rms with productivitiesϕ > ϕ′ is independent of ad valorem trade costs (Melitz, 2003). The existence of such fees, however,breaks this independence because per-unit costs disproportionately raise the price of low-price (high-productivity) rms compared to high-price (low-productivity) rms. As a result, with Pd xed, reduc-tions in the license fee induce relatively greater growth in export quantities among higher-productivityincumbents,

∂[qod(ϕ,aod)qod(ϕ′,aod)

]∂aod

= −σ

[τodϕ + aodτodϕ′ + aod

]−σ−1 τod

(1ϕ′ −

1ϕ

)(τodϕ′ + aod

)2 < 0. (4)

Thus, while the entry of low-productivity rms causes the overall share of incumbents to fall with aod,among incumbent rms the market shares of the largest and most productive rms rise. Removingthe license fee contributes to a gain in weighted-average productivity because these high-productivityrms increase their market shares after liberalization.

6Since rms pay the additive fee and pass this fee on to customers, prices are a constant markup above marginal cost.This contrasts with Berman, Martin and Mayer (2009) who also introduce additive transport costs but have variablemarkups since the consumer pays the fee. Wages are pinned down by a perfectly competitive outside sector.

7Berman, Martin and Mayer (2009) adopt a similar, small-country assumption in their model with per-unit costs.

5


The simple average productivity of exporters, ϕ(ϕ∗), is given by

ϕ(ϕ∗) =1

1−G(ϕ∗)

ˆϕ∗ϕg(ϕ)dϕ. (5)

With Pd xed, the average productivity of exporters falls in response to quota liberalization,

∂ϕ

∂a=

g(ϕ∗)

1−G(ϕ∗)

∂ϕ∗

∂a[ϕ(ϕ∗)− ϕ∗] > 0. (6)

Intuitively, as the license price falls and ϕ∗ declines, less-productive rms enter the export market,driving down the average productivity of all exporters. Given that an individual rm's productivity isxed by assumption, there is no change in the average productivity of incumbents.

The response of quantity-weighted average productivity to quota reduction is more complex becauseit depends upon the redistribution of activity among incumbents,

∂ϕ

∂a=

1

1−G(ϕ∗)

ˆϕ∗

∂ [q(ϕ)/Q]

∂aϕg(ϕ)dϕ︸︷︷︸

Intensive

(7)

+g(ϕ∗)

1−G(ϕ∗)

∂ϕ∗

∂a

[ϕ(ϕ∗)− ϕ∗ q(ϕ

∗)

Q

]︸︷︷︸

Extensive

The rst term in equation (7) is the change in weighted-average productivity due to the intensivemargin. The sign of this term is negative as reductions in the quota license fee increase the relativemarket share of high-productivity incumbents at the expense of low-productivity incumbents. Thesign of the extensive-margin contribution, on the other hand, is positive: a reduction in the licenseprice enables less ecient rms to commence exporting, which drives down the weighted average. Theoverall eect of a change in the license price on weighted average productivity is ambiguous. It isnegative if the right tail of the distribution of rm productivity is relatively thin as low-productivityentrants will account for a larger fraction of growth. It is positive if incumbents account for a largerfraction of growth.

The model's one-to-one correspondence between productivity and price yields similar relationshipswith respect to export prices.8 The average price of exports is given by

p(ϕ∗) =1

1−G(ϕ∗)

ˆϕ∗p(ϕ)g(ϕ)dϕ. (8)

Here, the removal of quotas implies an increase in the average price of exports, net of the impact ofremoving the license fee

∂p

∂a=

σ

σ − 1︸︷︷︸Intensive

+g(ϕ∗)

1−G(ϕ∗)

∂ϕ∗

∂a[p− p(ϕ∗)]︸︷︷︸

Extensive

. (9)

The sign of the rst term is positive and represents the change in average price among incumbents dueto the reduction of the license fee (see also equation (1)). The second term represents the change inthe average price due to the extensive margin. This term is negative: as license prices fall, less ecientrms enter the market pushing up the average price. The key insight here is that only incumbentscontribute to lower prices following quota reductions.

The response of quantity-weighted average export prices to reductions in the quota is given by

8In a more general setting in which rms choose the quality as well as level of their output, this one-to-one mappingmight break down. We discuss this issue in greater detail below.

6


∂p

∂a=

1

1−G(ϕ∗)

ˆϕ∗

∂ [p(ϕ)q(ϕ)/Q]

∂aϕg(ϕ)dϕ︸︷︷︸

Intensive

(10)

+g(ϕ∗)

1−G(ϕ∗)

∂ϕ∗

∂a

[p(ϕ∗)− q(ϕ∗)p(ϕ∗)

Q

]︸︷︷︸

Extensive

The rst term represents the intensive margin and its sign is positive: when license prices fall, theprices of all incumbent rms will fall. The extensive-margin term is negative, as less-productive entrantsenter the market with relatively high prices. The overall change in the weighted-average export priceis ambiguous: if the most productive incumbents market share rises enough, it falls, else it rises.

Our model of ecient allocation assumes constant product quality before and after quotas areremoved. It is of course possible that rms decide to alter product quality as a result of the tradeliberalization, i.e., that some of the price declines observed after quotas are removed might be driven byquality downgrading. We discuss how our empirical results might be interpreted in this more generalsetting in Section 5.3.

2.2 Numerical Solutions

We make use of numerical solutions of the ecient allocation model described above to derive im-plications for the price index Pd = Pd(ϕ

∗od) when aod > 0.9 For reasonable parameters values, these

solutions yield predictions similar to those derived analytically above.We consider two countries and one industry. We assume symmetric country sizes and set Lo = 100

and Ld = 100. Iceberg trade costs are chosen so that the share of Chinese textile and clothing exports inU.S. imports and vice versa match the observed shares in 2005 (23 percent and 5 percent, respectively).This parametrization of the iceberg trade costs captures the fact that wages in the United States arehigher than in China. The iceberg cost within each country is set to one, and we assume σ = 4.10

We follow the literature in assuming rm productivity is Pareto distributed, G(ϕ) = 1 − ϕ−γ withshape parameter γ. This shape parameter and the xed cost of exporting are chosen to match thedistribution of Chinese exports to the United States, Canada and E.U (the regions that imposed quotason China's textile and apparel exports). and the fraction of rms in China that are exporters.11 Thecalibration yields a shape parameter of 3.8 and a ratio of export xed cost to domestic xed cost equalto 2.12

Using these parameters, we solve for productivity cutos and price indexes in a quota-free equi-librium. We then re-solve the model under increasing levels of quota restrictiveness, which we deneas 1 minus the ratio of exports under quotas to exports in the quota-free equilibrium. Quota restric-tiveness varies from 0 to 1 as quotas move from non-binding to prohibitive. The relative growth ofChinese textile and clothing exports between 2004 and 2005 observed in Table 1, discussed in greaterdetail below, implies a quota restrictiveness of 58.

9We are grateful to Andreas Moxnes for providing the Matlab code used to derive the numerical solutions in Irarrazabalet al. (2010). We modify their code by adding the quota constraint and solving for an equilibrium license fee given thisconstraint.

10According to estimates from Broda, Greeneld and Weinstein (2006), the median elasticity of substitution in appareland textile products subject to quotas is 4.14

11The parameters are calibrated using the post-quota distribution of Chinese textile and clothing export values. Theshare of exports accounted for by the 50th, 75th, 90th, 95th and 99th percentiles in 2005 are 2, 8, 16, 14, 29 and 31percent, respectively. China's Annual Survey of Industries reports that 45 percent of rms in the textile and clothingsectors (Chinese Industrial Classication 17 and 18) exported in 2005. We explore the implications of using alternativeshape parameters and a lognormal productivity distribution in Section 6.

12The shape parameters used in numerical simulations such as ours vary. Bernard et al. (2007), for example, assumeγ = 3.4 in their analysis of comparative advantage and heterogenous rms. We examine the sensitivity of our results tothis parameter choice in Section 6.

7


The solid curve in Figure 1 plots the home country's change in average (exporter) productivitybetween an ecient allocation and the quota-free equilibrium as quota restrictiveness changes from0 to 1. Consistent with the comparative static in equation (6), which relied on a xed aggregate priceindex, this curve is downward sloping: when more-restrictive quotas are removed, average productivitydeclines more because there is greater the entry of low-productivity rms.

The dashed curve in Figure 1 plots the change in weighted-average productivity against quota re-strictiveness. As noted in the previous section, this relationship depends upon the extent to which thehighest-productivity incumbents gain market share following quota removal. Here, our numerical solu-tion indicates a positive slope: as more-restrictive quotas are removed, weighted-average productivityrises more because export growth is relatively higher among previously constrained, high-productivityincumbents (the intensive margin) than among low-productivity entrants (the extensive margin).

The dierences between the average and weighted-average productivity curves highlights the im-portance of the intensive margin for post-quota export growth under ecient allocation. Given ourparameters, relaxation of quotas with quota restrictiveness 58 percent implies that virtually all 99percent of the growth in export quantity following quota removal is accounted for by the intensivemargin.

3 A Brief Summary of the MFA

China's textile and clothing industry accounts for a substantial share of its overall economy. In 2004, itemployed 12.9 million workers, or 13 percent of total manufacturing employment (2005 China EconomicCensus). China's textile and clothing exports account for 15 percent of the country's overall exports,and 23 percent of world-wide textile and clothing exports (which equaled $487 billion dollars in 2005).

The Multiber Arrangement (MFA) and its successor, the Agreement on Textile and Clothing,grew out of restraints imposed by the United States on Japanese imports during the 1950s. Overtime, it grew into a broader institution that regulated the exports of clothing and textile productsfrom developing countries to the United States, EU, Canada (the UEC), and Turkey. (Though weare unable to locate the list of products under quotas in Turkey, they accounted for less than 0.5%of China's textile and clothing exports in 2004.) Bargaining over these restrictions was kept separatefrom multilateral trade negotiations until the conclusion of the Uruguay Round in 1995, when the UECagreed to eliminate the quotas over four phases. At the beginning of 1995, 1998, 2002 and 2005, theUEC were required to remove textile and clothing quotas representing 16, 17, 18 and the remaining49 percent of their 1990 import volumes, respectively. The order in which goods were placed intoa particular phase varied across importers, with each country generally choosing to place their mostsensitive textile and clothing products into the nal phase (Phase IV) to defer politically painfulimport competition as long as possible (Brambilla et al. 2010). This aspect of the liberalizationsuggests that the reaction of Phase IV exports relative to a control group is likely stronger than asimilar comparison in earlier phases. However, the fact that Phase IV goods were determined in 1995implies that their choice was not inuenced by demand or supply conditions in 2005.

China did not become eligible for quota removal until it joined the WTO at the end of 2001. In early2002, its quotas on Phase I, II and III goods were relaxed immediately. Removal of quotas on PhaseIV goods the focus of our empirical work occurred according to schedule in 2005, and coincidedwith China's agreement under the WTO to eliminate export licensing in all products by 2005. As of2004, China still required export licenses for 12 percent of other (i.e., non-quota-requiring) textile andclothing products, accounting for 5 percent of their 2004 export value.13 To the extent that removalof export licensing promoted entry into these products in 2005, the relative contribution of entry intoMFA product markets discussed in the next section will be biased downwards.

13Note that possession of a quota was a more binding constraint on exporting to UEC than possession of an ex-port license. The products that were subject to export restrictions are listed in China's WTO accession document(WT/ACC/CHN/49).

8


Like other countries under the MFA, China ocially allocated quotas on the basis of past perfor-mance, i.e., rm's ability to export their quota successfully in the previous year (Krishna and Tam1998). China's actual allocation of quotas, however, deviated from this principle, at times substan-tially (Moore 2002). In the 1980s, rent-seeking and political favoritism were rampant. The People'sLiberation Army (PLA), for example, received quotas in return for their support of the government,and these allocations were increased in 1989 following their backing of the state during the Tainanmencrisis. Likewise, the central Ministry of Commerce provided quota allocations to provincial authoritiesin an eort to promote the spread of textile and clothing manufacturing geographically (citation).

Although trading quotas in China was illegal throughout the MFA, anecdotal evidence suggeststhat an active black market emerged during the 1980s. One consequence of this illegal trading wasunused quota, which occurred when rms were unable to sell their allocations on the black market.14

To prevent quota from going unused, the government stepped up enforcement of allocations based onpast performance, and tried to prevent non-producing rms from receiving quotas (Moore 2002). Thesereforms are generally believed to have reduced black-market activity, though verication of this claimis dicult given rms' (understandable) reluctance to discuss illegal trading (Moore 2002; interviewsconducted by the authors). The sensitivity of our results to legal or illegal subcontracting, as well asempirical exercises designed to measure it, are discussed further in Section 5.1.

Starting in 2000, the government experimented with auctioning up to 30 percent of the total quotaallocation of a subset of MFA goods. To bid in these auctions, however, rms were required to winapproval from the government. Unfortunately, we have been unable to determine the precise criteriagoverning appoval, though we do know that bidding was open only to existing textile exporters.

4 Reallocation of Chinese Exports Following Quota Removal

From the perspective of the ecient allocation model described in Section 2, we expect quota liber-alization to coincide with three outcomes: a reallocation of market share towards the largest, mostproductive incumbent exporters under the quota institution; a reduction in incumbents export pricesdue to the removal of license fees; and the entry of less-productive exporters with relatively high exportprices. As discussed further in the counterfactual section below, an alternate hypothesis of inecientallocation implies a stronger role for the extensive margin.

4.1 Data

Our empirical analysis relies on data from several sources. The rst is Chinese export customs data byrm, eight-digit Harmonized System (HS) category and destination country. For each rm-product-country observation, we observe the total nominal value and quantity traded as well as whether rmsfall into one of three ownership categories: state-owned enterprises (SOEs), domestically ownedprivate rms (domestic) and foreign-owned private rms (foreign).15 Quantity units are availablefor 99 percent of observations representing 99 percent of export value, and vary across products, e.g.,square meters of fabric. We combine the value and quantity data to construct nominal unit values,also referred to as prices. As documented in Schott (2004), unit values can be noisy and we thereforefollow the literature in trimming outliers for some of our results as noted below.

We partition China's exports into six mutually exclusive and time-invariant groups based ondestination market and product type. Destination markets fall into two blocs: the rst encompassesthe United States, the members of the European Union and Canada and is referred to as "UEC"; the

14Although 2004 was the nal year of the MFA, there is anecdotal evidence that rms believed that quotas would bereimposed in subsequent years which would mitigate their incentives to subcontract in this year.

15The customs data separate rms into seven groups. We classify state-owned rms as SOEs; collective-owned,other and private domestic rms as domestic, and foreign-exclusive owned and two joint venture classications asforeign.

9


second bloc contains all other countries and is referred as "rest of the world" or "ROW".16 Within acountry, products are partitioned into three types: textile and clothing products subject to a quotaprior to 2005 (MFA), other textile and clothing products not subject to a quota (OTC), and non-textileand clothing products, e.g., electronics or steel (NTC).17

A given product-country is assigned to one of the six resulting ROW,UEC×MFA,OTC,NTCgroups. MFA-UEC, for example, refers to product-country exports that are subject to quotas, whileOTC-UEC encompasses to product-country exports of textile and clothing products not subject toquotas. Note that our classication is product-country specic, and so it is possible for a given HSproduct to be part of two dierent groups. An export of a textile and clothing product subject to aquota only in the United States to the United States, for example, is MFA-UEC, but an export of thatsame product to the EU is OTC-UEC.18

The mutual exclusivity of product-country assignments to groups is an important element of ouridentication strategy, as we use the OTC-UEC group to construct dierence-in-dierences estimatesfor the MFA-UEC groups' reactions to quota elimination. These comparisons assume that the textileand clothing products in the two groups are subject to similar demand and supply shocks. Amongthe 554 products that are subject to quotas by any of the three countries in 2004, 163 are subject toquotas by all three destinations, while 160, 50, and 6 are subject to quotas solely in the United States,solely in the EU and solely in Canada, respectively.

Some of our results also exploit variation in the extent to which quotas are binding. FollowingUSITC (2002), we dene a quota as binding if its ll rate exports divided by the respective quota exceeds some threshold. Using data on the level of U.S., EU and Canadian quotas available fromwebsites maintained by each country, we nd that 32 percent of the 1,017 product country observationsin the MFA-UEC group have ll rates exceeding 95 percent in 2004.19

4.2 Export Growth Following Quota Removal

Chinese export growth in 2005 is disproportionately large for textile and clothing goods released fromquotas, and generally occurs at the expense of state-owned enterprises.

As indicated in the top panel of the Table 1, the MFA-UEC group's 307 percent increase in exportvalue between 2000 and 2005 is the largest among all six groups over this period. By comparison,export growth is 205 and 113 percent for OTC-UEC and MFA-ROW, respectively, and 236 percentfor Chinese exports as a whole. The MFA-UEC group's dierentially large 2000 to 2005 growth isdue primarily to the 119 percent jump in export value that occurs in the nal year of the sample,when quotas are removed. MFA-UEC growth in prior years, by contrast, averages just 17 percent peryear.20 Likewise, the MFA-UEC group's 2005 growth is substantially larger than the growth exhibitedby OTC-UEC and MFA-ROW.

16We treat the EU as a single block of countries throughout our analysis given that quotas are set for the union as awhole.

17Phase I, II and III products, whose quotas were removed prior to 2005, are classied as OTC goods in our analysis.We note that changes to China's export classication scheme each year results in small changes to the number of productsin each type between 2000 and 2005. The set of textile and clothing products are: two-digit HS chapters 50-63; four-digitHS chapter 6406; ve-digit HS chapters 30059 and 65059; six-digit HS chapters 701919 and 94049. We identify the quotaproducts among these based on a concordance made available by the Embassy of China's Economic and CommercialAairs oce. This concordance identies the set of products subject to quotas in each destination market in 2004.

18A more concrete example: "cotton slips" to the United States are subject to quotas in 2004, while exports of "silkslips" are not. Our classication treats exports of cotton slips to the US as "UEC-MFA" and exports of silk slips to theUS as "UEC-OTC". As neither silk nor cotton womens' slips are subject to quotas in the EU in 2004, exports of theseproducts to the E.U. are classied as OTC-UEC. Note that groups do not vary within HS products for exports to ROWas the assignments depend only on quotas in UEC.

19Data on U.S., EU and Canadian ll rates are obtained from OTEXA, Système Intégré de Gestion de Licenses, andForeign Aairs and International Trade Canada, respectively.

20U.S., EU and Canadian quotas on China's MFA export quantities grew an average of 2 to 3 percent per year onceChina was admitted to the WTO in December 2001 (Brambilla et al. 2009). The relatively high value growth displayedbefore 2004 in Table 1 reects a combination of this growth in quantity as well as sizable increases in prices.

10


Data in the lower panel of Table 1 indicates that the surge in MFA-UEC export value is accompaniedby a similarly large increase in the number of MFA-UEC exporting rms. Between 2004 and 2005,the number of MFA-UEC exporters jumps 96 percent, from 9,523 to 18,628. Here too, this jumpis disproportionately large compared to prior years, and to both the 19 percent increase in Chineseexporters overall as well as the 39 and 16 percent increases exhibited by OTC-UEC and MFA-UEC,respectively. The relatively large increase in rms exporting MFA-UEC in 2005 is the rst indicationof the importance of the extensive margin in China's response to quota removal.21

MFA-UEC export growth is uneven across ownership types. Table 2 reports export value marketshares by rm ownership type and product-destination group in 2004 and 2005 (top and middle panels,respectively), as well as the change in market share between these two years (bottom panel). SOEs havea substantially greater presence in MFA-UEC than in the other ve product-destination groups priorto quota removal, but that gap drops markedly once quotas are removed. As indicated in the table,SOEs possess 54 percent of the MFA-UEC market in 2004 versus 26 percent for overall exports and 44percent for OTC-UEC. Once quotas are removed, SOEs MFA-UEC market share falls 16 percentagepoints, to 38 percent, bringing it closer to the 36 percent for OTC-UEC. So, while SOEs lose marketshares across all groups, their decline is most pronounced in MFA-UEC.

Together, Tables 1 and 2 highlight three facts about MFA-UEC exports following quota removal.First, post-quota export growth in MFA-UEC is large relative to other groups, particularly its closetcomparator, OTC-UEC. Second, MFA-UEC export growth is accompanied by a similarly large relativeincrease in the number of MFA-UEC exporters. Third, the disproportionately high market share heldby SOEs under quotas disappears quickly once quotas are removed. The rst fact indicates that thequotas imposed on Chinese exports by the United States, EU and Canada were binding. (Indeed, inunreported results, we nd even greater growth in exports and exporters among product-country pairswhose ll rates exceed 95 percent.) The second and third facts suggest that export growth followingquota removal is at odds with the ecient-licensing model discussed above, which has export growthfollowing quota removal being concentrated among large and productive incumbents.

4.3 Margins of Adjustment

We nd that export growth after quota removal is due disproportionately to the extensive margin, andfavors privately owned rms at the expense of incumbent SOEs.

Export growth can be decomposed into one intensive and two extensive margins. The intensivemargin is populated by incumbents by which we mean eight-digit HS products exported by the samerm to the same country in both 2004 and 2005. The extensive margin is comprised of entrants andexiters. Entrants are rm-product-country triplets which appear in 2005 but which were not presentin 2004. Exiters exhibit the opposite pattern.22 As illustrated in the top panel of Figure 2, 73 percentof the 10.7 billion dollar growth in MFA-UEC export value between 2004 and 2005 is due to net entry.This contribution is large compared to the 49 percent extensive-margin share observed in the smallerOTC-UEC export value increase over the same period.

The relative contributions of the intensive and extensive margins with respect to growth in exportquantity are similar, but due to the fact that HS codes vary in terms of the units used to record quantitywe cannot report quantity growth for the MFA-UEC and OTC-UEC groups as a whole. Instead, werst compute and decompose quantity growth for each product-country pair in MFA-UEC, and then,in the bottom panel of Figure 2, report the mean growth and mean contribution of each margin across

21A rm may appear in more than one group in Table 1if it exports in multiple product classes or if it exports toboth ROW and UEC. We nd that less than 2 percent of MFA-UEC exporters representing an even smaller fraction ofMFA-UEC export value are active only in that group. Indeed, depending on the year, 85 to 90 percent of MFA-UECexporters also export in MFA-ROW. Overlap with other groups, e.g., OTC-UEC is lower, on the order of 80 percent. Inour model, we treat multiple-product rms as single-product rms that manufacture dierent varieties.

22Note that multiple-product exporters may be counted in more than one margin of adjustment, e.g., they may exitone product-country and enter another.

11


product-country pairs, excluding outliers.23 As indicated in the gure, on average 86 percent of MFA-UEC quantity growth between 2004 and 2005 is driven by the extensive margin, versus an average of52 percent for OTC-UEC. This relative contribution of the extensive margin of 34 percent is a keydeterminant of the degree of ineciency in the quota allocation, and one that we make use of in thecounterfactual analysis in Section 6.

SOEs are the largest incumbent exporters. In 2004, MFA-UEC export value per rm averaged 2.1,0.5 and 0.7 million dollars for SOEs and privately owned domestic and foreign rms, respectively. Underecient allocation, export growth following quota removal should be concentrated among the largestincumbents, i.e., SOEs, due to their assumed greater productivity. Instead, we nd in Figure 3, thatSOEs exhibit the sharpest relative declines in market share during quota liberalization. In the gure,incumbents change in quantity-based market share between 2004 and 2005 is plotted against theirmarket share in 2004 using a lowess smoother. Separate relationships are plotted for each ownershiptype. Although the largest rms in 2004 experience the greatest declines in market share in both MFA-UEC and OTC-UEC, which could simply reect mean reversion, the declines are more pronounced inMFA-UEC, and most dramatic for SOEs within MFA-UEC.

Table 3 decomposes the average change in quantity-based market share for MFA-UEC between 2004and 2005 by margin of adjustment and type of rm ownership. It is constructed by determining themarket share of each margin within each product-country pair in 2004 and 2005, taking their dierenceand then averaging these dierences across the pairs. The rst column summarizes the overall shiftin market share from incumbents to net entrants, where the latter now distinguishes between newexporters and adders. New exporters are rms that did not export at all in 2004, while adders arerms that exported one or more other (potentially MFA-UEC) products in 2004 prior to adding anMFA-UEC product.24 The rst column of the table reveals that incumbents' quantity-based marketshare declines an average of 21 percentage points across MFA-UEC product-destination pairs in theyear quotas are removed. This decline is (necessarily) oset by a 21 percentage point gain by netentrants. Of this gain, adders and new exporters contribute 65 and 6 percent, respectively, whileexiters account for -50 percent.

The remaining columns of Table 3 decompose these overall changes by type of rm ownership;in each row, the sum of the nal three columns equals the value in the rst column. Three trendsstand out. The rst, contained in the nal row of the table, is a net reallocation of export activityaway from SOEs: their quantity-based market share declines an average of 22 percentage points acrossproduct-country pairs between 2004 and 2005, with 13 percentage points of this market share beingpicked up by privately owned domestic rms and 8 percentage points by privately owned foreignrms.25 Second, there is substantial gross reallocation of market share within rm types. This grossreallocation is highest among SOEs, where exiters and adders contribute -32 and 26 percent marketshare, new exporters hardly contribute, and the overall negative contribution of net entry reinforcesthe loss of market share by incumbents. SOEs that contribute to the add margin likely reect the largeheterogeneity within SOEs that we document below. Among privately owned domestic and foreignrms, by contrast, net entry makes a positive contribution that more than osets incumbents loss.

23We exclude observations outside the 5th and 95thpercentiles, which can exhibit very negative or very positive growthrates. The share of growth due to the extensive margin is 87 percent if these observations are included.

24According to Table 1, there are 18,628 rms that exported in MFA-UEC in 2005. A given rm may contribute toboth the intensive and adder extensive margins if they both continue exporting at least one MFA-UEC product between2004 and 2005 and add another MFA-UEC product during that interval. We nd that 4,668 rms satisfy these criteria;they account for 74 percent of 2005 MFA-UEC export value and experience a near doubling of their average number ofMFA-UEC products, from 9.8 in 2004 to 17.3 in 2005. We nd that an additional 1,132 rms are pure incumbents,i.e., they continue exporting at least one MFA-UEC product between 2004 and 2005 but do not add any new MFA-UECproducts after quotas are removed. These pure incumbents account for less than 3 percent of 2005 MFA-UEC exportvalue and exhibit a decline in their average number of MFA-UEC products from 2.2 in 2004 to 1.3 in 2005. A relativelylarge number of rms 12,828 add at least one MFA-UEC product between 2004 and 2005 without having exportedin this group (or at all) in the year before. These rms represent 23 percent of 2005 MFA-UEC export value and pickup an average of 4 MFA-UEC products.

25Price changes explain the dierence between the 21 percent decline in SOEs average quantity-based market sharesin Table 3 and their 16 percent decline in value-based market share in Table 2.

12


Third, while net entry by new exporters is negligible among SOEs, it accounts for 5 and 1 percentagepoints of the overall 13 and 8 percentage point gains of privately owned domestic and foreign rms.

Comparison of MFA-UEC to the other textile and clothing groups OTC-UEC in particular aids our assessment of whether the margin adjustments observed in Table 3 above are related to quotaremoval or other factors common to textile and clothing products, e.g., the removal of entry barriersand the declining importance of SOEs. Figure 4 displays the deviation between MFA-UEC and OTC-UEC incumbent and net entry margins by rm ownership type. This gure is constructed by creatinga table analogous to Table 3 for OTC-UEC, and then reporting the dierences between the rst andsixth lines in columns three to ve. As indicated in the gure, incumbent SOEs loss of market shareis 12 percentage points greater in MFA-UEC than in OTC-UEC. The corresponding dierences forprivately owned domestic and foreign rms are -2 and 0 percentage points, respectively. At the sametime, net entry by privately owned rms is 8 percentage points higher in MFA-UEC than in OTC-UEC.We note that t-tests reveal that all dierences except for the incumbent foreign rms are statisticallysignicant at conventional levels.

Together, the data in Figure 4 and Table 3 show that even though incumbents exports grewfollowing quota removal, they lost market share to entrants, and that this loss of market share isconcentrated among SOEs. These results provide further support for the idea that quota licenses wereallocated ineciently both across and within rm ownership types prior to their removal in 2005.

Our measurement of the extensive margin in this section and the price and productivity trends inthe next two sections is potentially sensitive black-market licensing, by which we mean the illegal sub-contracting of quota-holders' allocation to rms whose names do not appear on customs documents.We address this concern in detail in Section 5.1.

4.4 Prices

Chinese MFA-UEC export prices fall relative to the export prices of all other groups the year thatquotas are removed. In contrast with the ecient allocation model developed above, however, thesedeclines are due disproportionately to entrants rather than incumbents.

Figure 5 displays the mean percent change in groups' export prices between 2004 and 2005. Thesechanges are computed in two steps. First, for each product-country (hc) pair in each year (t), wecalculate a weighted-average export price (Phct) across all exporting rms using their quantity marketshares (θfhct) as weights,

Phct =∑f

θfhctpfhct. (11)

Then, for each product-country pair, we compute the percent change between years t and t − 1,∆Phct =

(Phct − Phct−1

)/Phct−1. Each bar in Figure 5 displays the mean of ∆Phct across all product-

country pairs in the group, excluding outliers.26 As indicated in the gure, export prices in MFA-UECfell 8 percent on average between 2004 and 2005. In OTC-UEC, by contrast, average export prices grew14 percent. Thus, relative to its closest comparator group, MFA-UEC export prices fell 22 percent.The MFA-UEC price decline in 2005 is also sharp relative to the group's average price growth of 16percent between 2003 and 2004 (not shown).

Figure 6 compares the export prices of entrants to exiters to incumbents. To facilitate comparisonacross products using dierent quantity units, we normalize incumbent and entering rm's exportprices in 2005 (pfhc2005) by their across-year, product-country mean (Phc), where

Phc =1

2

(Phc2004 + Phc2005

). (12)

26Extreme price changes are found for some product-country combinations, e.g., HS 62101030, garments of felt ornonwovens, of man-made bers, to Suriname, which grew 70,000 percent between 2004 to 2005. In Figure 6 we dropproduct-country pairs whose price changes are either below or above the rst and ninety-ninth percentile, respectively.Though excluding these product-country pairs lowers average export price growth in all groups, it does not undermineany of the substantive patterns discussed in this section.

13


For exiters, where a 2005 price is not observed, we divide rm's 2004 export prices by the same mean, pfhc2004/Phc. As in Figure 5, we exclude rms whose relative prices are below and above the rstand ninety-ninth percentiles of each distribution, respectively.

A key feature of Figure 6 is that the price distribution for exiters lies to the right of that for entrants.This ordering indicates that rms exiting MFA-UEC in 2004 have relatively high prices compared torms entering the group in 2005. By comparison, Figure 7 reveals that we do not nd a similarordering of entrants' and exiters' prices either contemporaneously in OTC-UEC or in MFA-UEC theyear before. Indeed, exiters' prices are lower than entrants' prices in MFA-UEC in 2004 and are almostindistinguishable from entrants' prices in OTC-UEC in 2005. A second notable feature of Figure 6 isthat MFA-UEC incumbents export prices in 2005 have a thin left tail compared to entrants, i.e., theyhave a lower proportion of very low prices. To the extent that incumbents relatively high prices do notreect variation in quality (more on this below and in Section 5.3), they provide intuition for the lossof market share by incumbents discussed in the last section. On the other hand, incumbents abilityto retain as much market share as they did given their relatively high prices may be due market orpolicy asymmetries such as long-term contracts or better marketing information that give high-pricedincumbents an advantage over low-priced entrants.

We quantify the relative importance of each margin in the overall growth of MFA-UEC prices usinga productivity decomposition proposed by Foster et al. (2008) and Griliches and Regev (1995):

∆Phct =1

Phct−1

∑f∈I

θfhc (pfhct − pfcht−1) +∑f∈I

(θfhct − θfhct−1)(pfhc − Phc

) (13)

+1

Phct−1

∑f∈N

θfhct(pfhct − Phc

)− 1

Phct−1

∑f∈X

θfhct−1

(pfhct−1 − Phc

) .Here, p and θ represent export unit values and quantity-based market shares, while f , h and c indexexporters, eight-digit HS categories and countries. I, N and X correspond to the sets of incumbent,entering (new exporters plus adders) and exiting rms, respectively. (We do not break entrants intoadders versus new exporters given the relatively small market share of new entrants noted in Table 3.θfhc is the average market share of rm f in hc across 2004 and 2005, i.e., θfhc =

(θfhct + θfhct−1

)/2.

Finally, pfhc is the average price of rm f in product-country hc across years t and t − 1. Like θfhc,it can be computed only for incumbents.

The rst term in square brackets captures the intensive margin. Its within component, the rstterm inside the brackets, measures the price change of incumbent exporters holding their market sharexed. The second, across component accounts for changes in incumbents market shares, weightingthose changes by the dierence between the rm's average across-year price and the overall averageacross-year price

(pfhc − Phc

). If incumbents prices fall (due, for example, to the elimination of the

license fee), the within component is negative. If incumbents prices are relatively high and their marketshares tend to decline, the across component is also negative and both components contribute to areduction in ∆Phct. The second term captures the entry margin; this term is negative if entrants'prices are lower than the across-year average price. The third term captures the exit margin, andits interpretation is analogous to the entry term: it is positive if exiters have relatively high pricescompared to the across-year average; because it is subtracted, a positive margin implies a negativecontribution to price changes.

Tables presenting separate decompositions for MFA-UEC and OTC-UEC by ownership type areavailable upon request. Here, we present in Table 4 the dierence between these decompositions, termby term. We assume that this dierencing controls for ination our data are nominal as well asthe other factors such as changes in technology and exchange rate movements that aect the pricesof all Chinese textile and clothing exports equally. Each column of the table sums to the nal row,while each row sums to the rst column. The top panel reports changes in price, while the bottom

14


panel expresses these changes as a percent of the overall 21.7 percent average relative MFA-UEC pricedecline displayed in the bottom of the rst column (as well as Figure 5). Results in Table 4 dropthe same product-country outliers as Figure 6, and all of the dierences between the two groups arestatistically signicant at the 1 percent level with the exception of the exit term for foreign rms.

The rst column of Table 4 reveals that the entrant and exit er margins account for 37 (-0.080/-0.217) and 30 (-0.065/-0.217) percent of the decline in MFA-UEC relative prices between 2004 and2005. This 67 percent contribution from the extensive margin is twice the 34 (-0.072/-0.217) per-cent contribution of the intensive margin. Here, as well, this dominance of the extensive margin isinconsistent with our model of ecient allocation of quotas prior to their removal.

Examination of the within and across terms for the incumbent margin indicate that changes inprice as well as changes in market share drive its contribution. The negative within term reveals thatMFA-UEC incumbents experienced larger price declines than OTC-UEC incumbents. The negativeacross term indicates that MFA-UEC incumbents with high prices lost relatively more market sharethan high-priced OTC-UEC incumbents.

The remaining columns of Table 4 highlight the inuence of SOEs on price trends. Almost half(47 percent) of the overall decline in MFA-UEC relative prices is due to SOEs. Moreover, SOEscontribution is weighted heavily towards the incumbent and exit margins, whereas entry plays thestrongest role among privately owned rms.

We nd very similar results using a more stringent comparison of price changes within HS categories.This comparison exploits variation in the sets of products subject to quotas in the three countries,thereby allowing us to control for demand shocks in destination markets and product-specic supplyshocks that might inuence price changes. As noted above, products subject to a quota in one of theUEC countries are not necessarily subject to quotas in the other two countries. This feature of thedata permits the following OILS dierence-in-dierences specication:

∆Phct = αh + αc + β1MFAhc + εhct, (14)

where αh and αc represent HS product and country xed eects andMFAhc is an indicator of whetherthe product-country pair is subject to a quota. The dependent variable is either the overall price decline(∆Phct) or the net price decline associated with the intensive or extensive margin. The regression isrestricted to MFA-UEC and OTC-UEC product-country observations, so that β1 picks out their averagedierence.

Results are displayed in Table 5. The rst three columns of the top panel report the results ofestimating equation (14) without xed eects. These columns reproduce the results shown in the rstcolumn of Table 4 and, as indicated in the nal row of the panel, reproduce the result that the extensivemargin accounts for 67 percent of the total price decline.

The next three columns of the top panel report results inclusive of product xed eects. Asindicated in the table, we continue to nd a sizable and statistically signicant average price declineeven when these declines are identied solely across UEC countries within products. The principaldierence between these results and those without xed eects is the smaller contribution of theextensive margin, which indicates that HS products subject to quotas in all three countries of theUEC (and therefore excluded from this regression) experience relatively greater net entry. One reasonfor this might be that products subject to a quota in all three countries may be more threateningto domestic competitors than those subject to a quota in only one or two countries. To the extentthat this threat is a signal of their labor intensity, entry may be easier. The nal three columnsof the top panel include both product and country xed eects, where the latter help control fordemand shocks common to all products within a country. The estimated total price change remainsstatistically signicant but declines in magnitude; the relative contribution of the extensive marginfalls to 44 percent.

The bottom panel of Table 5 partitions the product-country observations subject to a quota accord-ing to whether they are binding , i.e., whether their ll rates are above 95 percent. Unsurprisingly,the estimated total price price declines are larger for binding quotas across the left, middle and right

15


panels. We also nd that the relative contribution of the extensive margin is higher for binding quo-tas. Even when product and country xed eects are added to the regression, the extensive marginaccounts for the majority (55 percent) of the total price decline. The substantial inuence of entryacross the specications in Table 5 is inconsistent with ecient allocation.

As noted in Section 2, an alternative interpretation of the relative price declines documentedin this section is quality downgrading. Because quotas exert a relatively large per-unit penalty onlow-price, low-quality goods, rms may have an incentive to reduce export quality when quotas areremoved. Under this interpretation, the price declines associated with the extensive margin between2004 and 2005 merely reect the entry of low-productivity, low-price rms in 2005 and the exit of high-productivity, high-price rms in 2004. This interpretation, however, is inconsistent with the apparentrelative productivity of entrants and exiters. In particular, as noted above, we nd that entrantsare drawn predominantly from the private sector and that they enter predominantly at the expenseof SOEs which, as discussed below, are generally estimated to be of low average productivity. InSection 5.3 below we provide additional evidence against quality downgrading by estimating changesin export quality explicitly. Using a method based on Hummels and Klenow (2005), Khandelwal(2010) and Hallak and Schott (2011) we nd that although quality in MFA-UEC fell relative to thatof OTC-UEC as quotas were removed, the decline is not statistically signicant. In any case, even ifhigh-productivity entrants are entering with high-quality products, that outcome remains inconsistentwith ecient allocation, which predicts entry by low-productivity rms once quotas are removed.27

4.5 Productivity

Our model of ecient allocation predicts that entering rms are less productive than incumbents.Unfortunately, we are unable to estimate the productivity of entrants directly due to diculties asso-ciated with matching trade and production data.28 We do, however, observe a key rm characteristic ownership type that can be used to make indirect inferences about productivity.

Existing estimates of Chinese rm's productivity indicate that state-owned enterprises are sub-stantially less ecient than privately owned domestic or foreign rms. Using the Annual Survey ofIndustrial Production (ASP), which provides information on output and inputs for all SOEs and non-SOEs with revenue greater than 5 million RMB. Brandt and Zhu (2010) estimate the aggregate totalfactor productivity (TFP) of SOEs to be half the level of non-SOEs. Hsieh and Klenow (2009) usethe same data to estimate TFP at the rm level and nd that, on average, SOEs' productivity is 41percent that of domestic private rms.

Neither of these studies report dierences between exporting and non-exporting rms. Here, weuse the ASP to compare the TFP of SOE vs non-SOE and exporters, making use of a variable in theASP that identies whether or not the rm exports. We restrict our comparison to exporters whosemajor line of business in 2005 is textiles or clothing (industry codes 17 or 18), though we note that thiswill exclude many exporters of these goods whose main line of business lies outside this industry. Ofthe 15,214 textile and clothing exporters found in the ASP, 142 are SOEs, 7,992 are privately owneddomestic rms and 7,080 are privately owned foreign rms. Following Brandt, Van Biesebroeck andZhang (2009), we estimate TFP using a Tornqvist index number approach,

ln(TFPf ) = (vaf − va)− sf (lf − l)− (1− sf )(kf − k), (15)

27It is of course possible that high-productivity entrants choose to export low-price, low-quality goods in 2005. Thischoice hinges on the relative costs and benets of quality upgrading (see, for example, Baldwin and Harrigan 2009, Kuglerand Verhoogen 2010 or Johnson 2009). In fact, empirical evidence in Khandelwal (2010) or Kugler and Verhoogen (2010),suggest that the clothing may be an example of such an industry where, on average, the pursuit of high quality by highproductivity rms is unlikely.

28In principle one might merge the trade data, which identies entrants versus incumbents at the HS-country level,with China's Annual Survey of Industrial Production, which provides information on output and inputs. In practice,match rates from this merge are quite low given the diculties of matching rms based on their names rather thannumerical identiers. Of the 37,986 rms that we observe in the trade data exporting textile and clothing products in2004, we have succeeded in matching 7,157 rms (19 percent) to the Annual Survey.

16


where va, l, and k are in logs and denote value added, wages and xed assets (net of depreciation) foreach rm, and where a bar over a variable denote an average across all textile and clothing exporters.The weight on wages is sf = (sf + s)/2, where sf is the share of wages in total value added by eachrm and s is the average across rms. The TFP measure for a given rm is relative to a hypotheticalrm with the average output and inputs. Following Brandt et al. (2009), wages are dened asreported rm wages plus employee benets (unemployment insurance, housing subsidies, pension andmedical insurance), and the capital is reported capital stock at original purchase price less accumulateddepreciation.

Figure 8 plots the distribution of textile and clothing exporter's TFP relative to the hypotheticalaverage rm by type of ownership. As indicated in the gure, while some SOEs exhibit relatively highproductivity, their distribution lies substantially further to the right than the distributions of privatelyowned exporters. On average, SOEs are 18 percent less productive than the hypothetical mean, whileprivately owned domestic and foreign exporters are 76 and 54 percent more productive, indicating thatthe average privately owned domestic and foreign rms are 88 and 72 percent more productive thanSOEs.

In Table 6, we combine these estimates with the market share changes by ownership type reportedin Table 2 to provide a coarse, back-of-envelope estimate of the productivity gain associated withthe replacement of SOEs by privately owned rms. The rst column of the table reports each typeof rm's average TFP relative to the hypothetical mean. The second column reports the change inmarket share for each rm type in MFA-UEC between 2004 and 2005. . Assuming that all rms withinan ownership type have the same relative TFP, and that rms' TFP remains constant as the MFA ends(both of which are conservative), the growth in aggregate MFA-UEC TFP following quota removal is18.5 percent (column 3). This increase is substantial relative to the 9 percent growth implied by thecorresponding market share changes in OTC-UEC over the same period reported in columns 4 and 5.It is also large relative to the average 4 percent TFP growth of all non-agricultural rms in China overthe 1998 to 2007 period found in the growth accounting exercise conducted by Brandt and Zhu (2010).

In Section 6, we discuss how such an overall gain in TFP might be decomposed into the part dueto the removal of quotas versus the part due to the dismantling of the inecient quota allocationinstitution.

5 Caveats and Alternate Explanations

5.1 Subcontracting by Producing Firms

As noted in Section 4.3, our measurement of the extensive margin is potentially sensitive to subcon-tracting if the quota-holding rm and the ultimate producer of the export are dierent and customsdocuments list the name of the former rather than the latter. In that case, our estimates of extensive-margin activity following quota removal will be biased upwards, as we would erroneously interpret theultimate producers' ability to export under their own name in 2005 as entry, and the disappearanceof the former quota holders that used subcontractors as exit. To the extent that subcontracts wereawarded to subcontractors on the basis of eciency, unobserved subcontacting also interferes with ourability to identify ecient allocation via margins of adjustment.

Five trends in the data suggest that subcontracting exerts a limited inuence on our results. First,if subcontracting were rampant and driven by eciency, one would expect entry to be concentratedamong a small set of large, highly productive former subcontractors that had not previously appearedon customs documents and which could now export freely. As noted in Section 4.3 (footnote 24),however, we nd that MFA-UEC entrants in 2005 that had not exported in that group in 2004 wererelatively large in number but small in export value. Indeed, 76 percent of MFA-UEC exports in2005 are accounted for by the minority (40 percent) of rms that are either incumbents or adders ofadditional MFA-UEC products between 2004 and 2005. This is reassuring since there is ample evidencefrom existing research on entry into export markets that new exporters, or new products exported by

17


existing rms, are small relative to incumbents (Bernard et al. 2010).Second, we nd that the vast majority of rms exporting MFA-UEC products in 2004 are also active

in MFA-ROW. Here, too, if subcontracting of MFA-UEC production were substantial, one would expectthe opposite, i.e., that MFA-UEC quota holders would outsource production of MFA-UEC goods toexporters of those goods that did not possess quotas, i.e., the rms exporting in MFA-ROW. Likewise,it is hard to understand why exporters of MFA-ROW products would require subcontractors to meettheir quota in MFA-UEC.29 We nd that 86 percent of MFA-UEC exporters, accounting for 96 percentof MFA-UEC export value, are present in both groups in 2004. Furthermore, we nd even higher jointparticipation among SOEs, the set of rms for which subcontracting might be most likely, given theirlow average productivity.

Third, examining the ASP, we nd little evidence that producers' exports exceeded their production,as might be expected if they were on-exporting subcontractors output. Using ASP data from 2004and 2005, we compute the export-to-production ratio among exporters that report textile and apparelas their main line of business. In each year, we nd that production exceeds exports for 95 percent ofexporters, and that this fraction is even higher among SOEs. While this evidence is consistent with lowlevels of subcontracting, it must be interpreted in light of two limitations of our data. First, the valuesin the Annual Survey reect the total exports and total production of rms whose predominant line ofbusiness is textiles and clothing; as a result, we are unable to narrow in on exports versus productionof just MFA goods for all rms that produce them. Second, information revealed by the exports toproduction ratio depends on the relative importance of the export market; rms selling large quantitiesdomestically might nevertheless export a relatively small amount of sub-contracted production.

Fourth, if subcontracting were the only way a rm with a quota license was able to fulll itsquota, the rms relying on subcontractors in 2004 would exit or shrink substantially once quotas wereremoved. In fact, we nd that the 2004-to-2005 exit and shrink rates are relatively low in MFA-UECcompared to OTC-UEC across all ownership types (results available upon request). We note that thistrend, too, must be interpreted with caution: to the extent that (potentially less-productive) rmsthat subcontracted production in 2004 were able to overcome the costs of exporting once the MFAended, rms that previously used subcontractors might decide to produce their own exports, reducingthe expected exit and shrink rates.

Finally, we nd a relatively strong contribution by the extensive margin in processing versusordinary exports, where the former refers to exports that are assembled in an export processingzone with a disproportionate share of raw materials that are imported at reduced or often zero tarirates. Subcontracting of processed exports is more dicult, especially for subcontractors that lieoutside the processing zone, given that the rules governing this class of exports must be obeyed bythe subcontractor. Processed exports account for 19 and 20 percent of MFA-UEC exports in 2004 and2005, respectively. Table 7 compares the relative contribution of the extensive margin in MFA-UECversus OTC-UEC exports for processed versus all exports. We nd that incumbents lose more marketshare in processing exports (-16 percent) than in all exports (-14 percent), and a similar reallocationaway from SOEs.

5.2 Subcontracting by Intermediaries

Subcontracting by intermediaries (i.e., non-producing trading rms) presents a dierent challenge toidentication than subcontracting by producers: while the latter had no reason to continue once thequota institution ended, there is no reason for the former to disappear. In any case, even if the numberof intermediates remained constant between 2004 and 2005, the number of producing rms with whichthey contracted and, therefore, their inuence on the true adjustment of China's extensive andintensive margins would be unknown because we do not observe the set of producers from which anintermediary sources.

29As discussed in Section 3, virtually all MFA products had full trading rights so all rms could directly export anMFA product to ROW if they so chose .

18


One might expect trading rms to be replaced by producers in 2005 if quota-rich trading rmswere an important conduit for quota-poor producers' goods. In fact, we nd relatively strong entry bytrading rms, dened as in Ahn et al. (forthcoming) as rms with the words importer, exporteror trader in their title, in MFA-UEC versus OTC-UEC between 2004 and 2005. One reason for thisgrowth that is consistent with our conclusions above but which contributes to an under-estimation

of the inuence of the extensive margin, is that intermediaries helped a new set of low-productivityentrants overcome the xed costs of exporting once quotas were removed (Ahn et. al, forthcoming).One caveat associated with this conclusion is that our classication of rms as trading companies isimperfect, and, in particular, might result in rms that have both production and trading arms beingclassied as traders. A large fraction of the textile and clothing apparel SOEs that export, for example,are classied as traders, which is at odds with the evidence presented above that virtually all SOEs inthe ASP have higher production output than exports. Indeed, according to our classication, tradingcompanies account for 48 and 46 percent of OTC-UEC and MFA-UEC exports in 2004, which is quitelarge relative to the 24 percent share of intermediaries in China's overall exports. If we reclassify allSOEs as producers, the export share of the remaining rms classied as traders falls to 13 and 11percent, respectively.

5.3 Quality Downgrading

An alternative interpretation of the price declines observed in Section 4.4 is that following the removalof the quotas, rms lowered product quality. We investigate quality downgrading by embedding con-sumer's preference for quality (λ) in the CES utility employed in our model of ecient allocation:

U =(´

ω∈Ω(λ(ω)q(ω))

(σ−1)/σdω)σ/(σ−1)

. The demand for a particular rm f 's export of product h

to country c at time t is given by:

qfhct = λσ−1fhctp

−σfhctP

σ−1ct Yct (16)

We remove the common destination-year price index and market size by dividing by the averagedemand within a country-product-time triplet, and solve for the rm's relative quality:

λfhctλhct

=

(qhctqhct

) 1σ−1

(phctphct

) σσ−1

. (17)

Using the elasticity of substitution σ = 4 described above, we employ data on quantity and unit valueto estimate the relative quality of each exported variety. The intuition behind this approach is similarto Hummels and Klenow (2005), Khandelwal (2010) and Hallak and Schott (2011): conditional onprice, a variety with a higher market share is assigned higher quality. By imposing an elasticity ofsubstitution, we avoid having to estimate demand before inferring quality.

We use our estimates of rm-level quality to assess the extent to which quality grows dierentiallyin MFA-UEC vs OTC-UEC following quota removal. As with prices (equation 11), we dene aggregatequality to be

Λhct =∑f

θfhctλfhctλchct

, (18)

where, as above, θfhct denotes the quantity market share of rm f in product-country pair hc in yeart. Then, for each product-country pair, we compute the percent change in quality between betweenyears t and t− 1 as ∆Λhct =

(Λhct − Λhct−1

)/Λhct−1.

Each bar in Figure 9 displays the mean of ∆Λhct across all product-country pairs in the notedgroup, excluding outliers.30 As indicated in the gure, MFA-UEC export quality rose an average of

30As with prices, we drop product-country pairs whose quality changes are either below or above the rst and ninety-ninth percentile, respectively.

19


13 percent between 2004 and 2005, versus 17 percent for OTC-UEC. Regression results analogousto equation (14), however, reveal this dierence to be statistically insignicant at conventional levels.Our within-product method for inferring quality downgrading diers from the across-product approachadopted by Harrigan and Barrows (2009). They dene quality downgrading as a shift in consumptionfrom high- to low-priced product HS categories. That method, also used by Aw and Roberts (1986) andBoorstein and Feenstra (1991), compares quantity-weighted average prices to value-weighted averageprices. If the former decreases by more than the latter, exports have shifted towards cheaper categoriesand average quality is said to have fallen. We follow our approach for two reasons. First, across-productevidence of quality downgrading does not account for quality changes within HS categories or withinrms, which our data can address directly. Second, lower prices may also reect increases in a rm'seciency, which must be disentangled from quality. Analysis of a rm's price and quantity informationcan be used to infer its quality. We note that our lack of evidence in favor of quality downgradingwithin products is not inconsistent with Harrigan and Barrows' (2009) nding of a a shiftby U.S.consumers toward lower-price textile and clothing categories once the MFA ends.

6 Decomposing Productivity Gains

Given the evidence against ecient allocation presented in Section 4, we construct a model of inecientallocation to decompose the overall productivity gain following the end of the MFA into the part thatis due to the removal of the quotas versus the part that is accounted for by the removal of the exportlicensing institution that managed it. This model relies on the same basic structure, assumptions andparameters as the ecient-allocation model, including asymmetric iceberg transportation costs andPareto-distributed rm productivity, that we used in Section 2.2.

We rst solve the model under the scenario that rms do not face quotas in the foreign market. Inthis quota-free equilibrium, we nd that aggregate TFP among China's exporters is 10.5.

We next solve the model under ecient quota allocation. We rst need to determine the restrictive-ness of the quota. The median country-product export quantity growth for MFA-UEC between 2004and 2005 was 162 percent versus 25 percent in OTC-UEC. Thus, relative to OTC-UEC, MFA-UECexports grew 137 percent. This translates into a quota restrictivness of 58 percent (1-1/2.37).31 Asdescribed in Section 2.2, rms choose whether or not to pay the common license fee implied by thisrestrictiveness. While the most productive rms do pay the fee and enter the export market, the feekeeps their market shares relatively low compared to the quota-free equilibrium, constraining aggregateTFP to 5.6. As illustrated in the right panel of Figure 10, this result implies aggregate TFP growth of4.9, or 87 percent (4.9/5.6), in moving between an equilibrium in which quotas are allocated ecientlyto one which is quota-free. This large gain is driven by the highly skewed distribution of exporterproductivity: as quotas are removed, exports by the highest-productivity incumbents surge. As notedin Irarrazabal et al. (2010), the large gains associated with the removal of specic taris, or, in ourcase, licensing fees, exceed those implied by traditional trade models that solely consider the removalof iceberg transportation costs (i.e., the class of trade models discussed in Arkolakis et al., 2010).

To solve the model under inecient allocation, we impose the same quota restrictiveness as underecient allocation, but assume the government allocates market share to N exporters based uponpolitical connections that are potentially uncorrelated with productivity. Thus, to solve the model,we need to choose a value for N , draw a political connection for each rm, and then use the draws toallocate quota among the N rms.

Two pieces of information determine N . The rst is the relative growth of MFA-UEC versus OTC-UEC rms between 2004 and 2005 which, as reported in Table 1, is 57 percent. The second piece

31Observed export growth in 2005 is due both to the removal of the quotas and to the elimination of their embeddedinstitutions. In using this overall observed export growth to solve the model under ecient allocation, we thereforelikely overstate export growth due solely to the removal of the quotas. Over-statement of export growth implies ourunder-estimation of TFP under eciently alloated quotas and, therefore, under-estimation of the contribution of theremoval of the embedded licensing institution in the inecient-allocation equilibrium below. Thus, our estimate of thecontribution of the removal of the embedded institution is conservative.

20


of information needed to choose N is the number of exporters under the no-quota equilibrium. Thisnumber is generated in the quota-free solution discussed above. We choose N to be 57 percent of thisnumber; as a result, when we remove ineciently allocated quotas, the growth in exporting rms inthe model will match the data perfectly.

We draw rms' political connections from a standard uniform distribution in a way that allows fora correlation between political connectedness and productivity.32 To determine the amount of quotaallocated to each of the N rms, we use the empirical distribution of export values observed in 2004.That is, we divide the actual export transactions in 2004 all of which by denition were exportedunder a quota allocation into 1000 bins and compute the market share of each bin. We then dividethe N rms in the model into 1000 bins, and give each of them the value market share needed to matchthe total value market share of their respective bin in the empirical distribution. (We use value- ratherthan quantity-based market because we cannot compare quantities across products).

As reported in Figure 11, we compute TFP and the net contribution of the extensive marginto export growth following quota removal for various levels of correlation between productivity andpolitical connectedness. Each point in the gure represents a dierent correlation between the twodraws. Intuitively, we nd that aggregate TFP is lower, and the contribution of the extensive marginis higher, as the correlation between the two draws falls: the stronger the relationship between politicalconnections and productivity, the lower the penalty for inecient allocation. We seek a correlation thatmatches the 34 percent growth in the MFA-UEC versus OTC-UEC extensive margin following quotaliberalization reported in Figure 2, a key statistic in identifying resource misallocation. As indicatedin the gure, extensive-margin growth of 34 percent along the x -axis corresponds to a TFP of 1.80along the y-axis and implies a 24 percent correlation between the two draws.

As illustrated in Figure 10, these results imply that removal of China's inecient quota-allocationinstitution for a TFP gain of 3.8 is close to the 4.9 TFP gain associated with the removal ofeciently allocated quotas. That is, of the overall gain associated with the removal of inecientlyallocated quotas, 44 percent is due to the institutional reform embedded in the liberalization.

We note that while the absolute gain from quota removal is sensitive to the skewness of the pro-ductivity distribution, the relative importance of institutional reform is not. We nd that inecientallocation accounts for 40 to 60 percent of the overall gain as the Pareto shape parameter varies be-tween 5.5 and 3.2 and all other parameters are held xed. Results are also insensitive to calibrationof export size and fraction of exporters to a lognormal distribution (mean 0.4 and standard deviationof 0.8). Under that assumption, the contribution of inecient licensing is also 44 percent. Finally, wealso examine the sensitivity of our numerical solutions to the particular net extensive margin growthwe use to calibrate the model. As discussed in Section 4, unobserved subcontracting may under certaincircumstances lead us to over-estimate this key statistic. As noted in Figure 11, if the true contri-bution of this margin were 25 rather than 34 percent, aggregate productivity in the inecient modelwould be 2.04, implying inecient allocation accounts for 42 percent of the overall productivity gainfrom removing the quotas.

The large absolute productivity gains in the model are driven by the skewed productivity distri-bution we observe in the data. Applying the implied breakdown of gains to the growth implied bythe back-of-envelope calculation in Table 6, we nd that the removal of the inecient licensing insti-tution increased textile and clothing exporters' productivity by 3.9 percentage points (8.8*.44) whileremoving the quotas increased productivity their productivity by the remaining 4.9 percentage points.

7 Conclusion

We evaluate productivity gains from a specic trade liberalization episodethe removal of textile quotason Chinese exporters. Following liberalization, we observe substantial reallocation away from inecientincumbent rms towards ecient entrants which implies large productivity gains among these textile

32We can change the correlation between a rm's productivity (ϕ) and its political connection (pc) draw by pc =

ρϕ+√

(1− ρ2)ε, where ρis the correlation and ε is a iid random variable.

21


exporters. These patterns of adjustment are at odds with predictions of a model under which quotalicenses were allocated based on rm productivity.

This episode highlights two key themes that have been under-studied in the literature. First, whilemany studies have emphasized misallocation of resources among the set of active rms, we observemisallocation along the extensive margin. Explicit government policy kept quota licenses out of thehands of the most productive textile exporters and once this institution was dismantled, these rmsenter the export market. The removal of the licensing institution highlights the second key implicationof our analysis. Theoretical models in international trade typically presume an ecient allocation ofresources, irrespective of trade barriers. However, institutions that evolve to manage trade barriersare often corrupted by government bureaucrats which impose additional distortions in addition to thetrade barrier itself. Trade liberalization that dismantles such institutions delivers additional gains fromtrade beyond just the removal of the trade barrier. Our counterfactual analysis suggests that movingfrom an inecient quota allocation to an ecient one delivers almost twice the TFP gains than theremoval of the trade barrier itself. That is, the eciency cost of the quota on the Chinese exporterscould be been reduced by more than half through internal reforms of the licensing system.

Our results provide one explanation for why empirical ndings of the gains from trade, for instancein Feyrer (2010) or Pavcnik (2002), are often large compared to the gains predicted in standard modelsof international trade (e.g., Arkolakis et. al 2010). These models ignore the fact that countries mustcreate institutions to manage the trade barriers they impose, and that these institutions may imposeineciencies in addition to those directly caused by the trade barrier. The results in this paper suggestthat an interesting avenue for future research would be incorporate these types of distortions withintrade models which may enhance our understanding of the gains from trade.

References

[1] Ahn, JaeBin, Amit K. Khandelwal and Shang-Jin Wei. The Role of Intermediaries inFacilitating Trade, forthcoming Journal of International Economics.

[2] Alchian, Armen A. and William R. Allen (1964). University Economics. Belmont, CA:Wadsworth Publishing Company.

[3] Alfaro, Laura., Andrew. Charlton, and F. Kanczuk (2008). Firm-Size Distribution andCross-Country Income Dierences, NBER Working paper 14060.

[4] Anderson, James (1985). The Relative Ineciency of Quotas: The Cheese Case, Amer-

ican Economic Review, 75(1), 178-90.

[5] Arkolakis, Costas, Arnaud Costinot and Andres Rodriguez-Clare (2010). New TradeModels, Same Old Gains?, NBER Working Paper 15628.

[6] Aw, Bee Yan and Mark J. Roberts (1986), Measuring Quality Change in Quota-Constrained Import Markets: The Case of U.S. Footwear, Journal of International Eco-nomics 21:45-60.

[7] Baldwin, Richard and James Harrigan (2011). Zeros, Quality, and Space: Trade Theoryand Trade Evidence, American Economic Journal: Microeconomics, forthcoming.

[8] Banerjee, Abhijit V. and Esther Duo (2004). Do Firms Want to Borrow More: TestingCredit Constraints Using a Targeted Lending, mimeo, MIT.

[9] Banerjee, Abhijit V. and Esther Duo (2005), Growth Theory Through the Lens of De-

velopment Economics, in Philippe Aghion and Steven N. Durlauf, eds. Handbook ofEconomic Growth, Volume 1A. Cambridge: Elsevier B.V.

22


[10] Banerjee, Abhijit V. and Benjamin Moll (2010). Why Does Misallocation Persist?,American Economic Journal: Macroeconomics, 2(1), 189-206.

[11] Berman, Nicolas, Philippe Martin and Thierry Mayer (2009). How Do Dierent ExportersReact to Exchange Rate Changes? Theory and Empirics, CEPR Discussion Paper 7493.

[12] Brambilla, Irene., A. K. Khandelwal, and P. K. Schott (2010). China's Experience Underthe Multiber Arrangement (MFA) and the Agreement on Textiles and Clothing (ATC),in China's Growing Role in World Trade, Ed. by R. Feenstra and S.-J. Wei, Chicago:University of Chicago Press.

[13] Brandt, Loren, Johannes Van Biesebroeck, and Yifan Hang (2009). Creative Accountingor Creative Destruction: Firm-Level Productivity Growth in Chinese Manufacturing,NBER Working Paper 15152.

[14] Brandt, Loren and Xiaodong Zhu (2010). Accounting for China's Growth, mimeoToronto University.

[15] Buera, F., J. Kaboski, and Y. Shin (2008). Finance and Development: A Tale of TwoSectors, mimeo, UCLA.

[16] Buera, F. J. and B. Moll (2009). A Simple Model of the Wealth Distribution and theMacroeconomy. mimeo, UCLA.

[17] Boorstein, R. And R. Feenstra (1991). Quality Upgrading and Its Welfare Costs in U.S.Steel Imports, 1969-1974, in E. Helpman and A. Razzin, eds. International Trade andTrade Policy. Cambridge: MIT Press, 1991, 157-186.

[18] Chaney, Thomas (2008). Distorted Gravity: The Intensive and Extensive Margins ofInternational Trade, American Economic Review, 98(4), 1707-1721.

[19] Chari, A. V. (forthcoming). Identifying the Aggregate Productivity Eects of Entryand Output Restrictions: An Empirical Analysis of License Reform in India, American

Economic Journal: Economic Policy.

[20] Demidova, Svetlana, Hiau Looi Kee and Kala Krishna (2009). Do Trade Policy Dier-ences Induce Sorting? Theory and Evidence from Bangladeshi Apparel Exporters, mimeoPennsylvania State University.

[21] Djankov, Simeon, Caroline Freund and Cong S. Pham (2010). Trading on Time, Reviewof Economics and Statistics, 92(1), 166-173.

[22] Dollar, David, and Shang-Jin Wei (2007). Das (Wasted) Kapital: Firm Ownership andInvestment Eciency in China, NBER Working Paper 13103.

[23] Falvey, Rodney E (1979), The Composition of Trade within Import-Restricted Cate-gories, Journal of Political Economy, 87(5): 1105-1114.

[24] Feenstra, Robert C. (1988). Quality Change under Trade Restraints in Japanese Autos,Quarterly Journal of Economics 103(1), 131-146.

[25] Feyrer, James (2009). Trade and Income: Exploiting Time Series in Geography, NBERWorking Paper 14910.

[26] Foster, Lucia, John Haltiwanger and Chad Syverson (2008). Reallocation, Firm Turnover,and Eciency: Selection on Productivity or Protability?, American Economic Review,98(1), 294-425.

23


[27] Griliches, Zvi and Haim Regev (1995). Firm Productivity in Israeli Industry: 19791988,Journal of Econometrics, 65(1), 175203.

[28] Hallak, Juan C. and Peter Schott (2011). Estimating Cross-Country Dierences in Prod-uct Quality , forthcoming Quarterly Journal of Economics.

[29] Harrigan, James and Georey Barrows (2009). Testing the theory of trade policy: Ev-idence from the Abrupt End of the Multiber Arrangement, Review of Economics and

Statistics, 91(2): 282-294.

[30] Hummels, David and Peter Klenow (2005). The Variety and Quality of a Nation's Ex-ports, American Economic Review 95, 704-723.

[31] Hummels, David, and Alexandre Skiba (2004). Shipping the Good Apples Out? AnEmpirical Conrmation of the Alchian-Allen Conjecture, Journal of Political Economy,

112(6), 1384-1402.

[32] Hsieh, Chang-Tai and Peter J. Klenow (2009). Misallocation and Manufacturing TFP inChina and India, Quarterly Journal of Economics, CXXIV(4), 1403-1448.

[33] Irarrazabal, Alfonso, Andreas Moxnes and Luca David Opromolla (2010), The Tip of theIceberg: Modeling Trade Costa and Implications for Intra-Industry Reallocation, mimeo,Dartmouth College.

[34] Johnson, Robert (2010). Trade and Prices with Heterogeneous Firms, mimeo, Dart-mouth College.

[35] Kaufmann, Daniel and Shang-Jin Wei (2000). Does `Grease Money' Speed Up the Wheelsof Commerce, IMF Working Paper WP/00/64.

[36] Khandelwal, Amit K. (2010). The Long and Short of Quality Ladders, Review of Eco-

nomic Studies, 77(4), 1450-1476.

[37] Krishna, Kala M., and Ling Hui Tan. Rags and Riches: Implementing Apparel Quotasunder the Multi-Fibre Arrangement. Ann Arbor: University of Michigan Press, 1998.Laird, Sam, and Alexander Yeats, 1990, Quantitative Methods for Trade barrier Analysis,New York: NYU Press.

[38] Kugler, Maurice and Eric Verhoogen (2010). Prices, Plant Size and Product Quality,mimeo, Columbia University.

[39] Mandel, Benjamin (2009). Heterogeneous Firms and Import Quality: Evidence fromTransaction -Level Prices, mimeo, Federal Reserve Board of Governors.

[40] Midrigan, Virgiliu and Yi Xu (2010). Finance and Misallocation: Evidence from Plant-level Data, NBER Working Paper 15647.

[41] Melitz, Marc J. (2003). The Impact of Trade on Intra-Industry Reallocation and Aggre-gate Industry Productivity, Econometrica, 71(6), 1695-1724.

[42] Moore, T. G. (2002). China in the World Market: Chinese Industry and InternationalSources of Reform in the Post-Mao Era. Cambridge: Cambridge University Press.

[43] Restuccia, D. and R. Rogerson (2008). Policy Distortions and Aggregate Productivitywith Heterogeneous Plants, Review of Economic Dynamics, 11(4), 707720.

[44] Pavcnik, Nina (2002). Trade Liberalization, Exit, and Productivity Improvements: Evi-dence from Chilean Plants, Review of Economic Studies, 69(1), 245-276.

24


[45] Petrin, Amil and Jagadeesh Sivadasan (2010). Estimating Lost Output from AllocativeIneciency, with Application to Chile and Firing Costs, mimeo, University of Michigan.

[46] Schott, Peter K. (2004). Across-Product Versus Within-Product Specialization in Inter-national Trade, Quarterly Journal of Economics, 119(2), 647-678.

[47] Sequeria, Sandra and Simeon Djankov (2010). An Empirical Study of Corruption inPorts, mimeo, London School of Economics.

[48] Tang, Man-Keung and Shang-Jin Wei (2009). The Value of Making Commitments Ex-ternally: Evidence from WTO Accessions, Journal of International Economics, 2009,78(2): 216-229.

[49] United States International Trade Commission (USITC) The Economic Eects of Signif-icant U.S. Import Restraints. Third Update 2002, June 2002.

25


! "#$ $#$ %#& %'#( )# "#$ &!#" ! !&#% $#' &"#& '%#! )#) )# $(#$ !(&# $#' &%# !!#" %#' )#( &(#) & "#% !!# ")#! !(%#& !!# %#' "&$#( " $&#) !&#' ((#$ !%#! !"#& $#' ('&#) ( &$&#) !)#) ($# %'#$ !$#$ !'#) %%)#)*+,- .( )) $$ !!& ( ( & % &)*+,- ." $ ! ' & !% %*+,- ".( &( !' " ' & !!' &!

/01/2

&%3( ! 3( !!3'%& "3 "" (3'$ &3(&) "%3"!& ! "%3 '& !!3%%$ !"3$%$ & 3%" )3"$ "3(& ('3(% )!3($& !"3""% !'3!)' &'3& ' '3)( (3'!! %$3)! & %"3') !%3) $ &3 '% "'3 "' !3% ! %3%'& '(3)$$ " '"3'!' 3("$ '3!) )&3 $( !)3((" '3(& ! 3(&( ( !!3"$$ )3$% &&3$"$ %%3 $ &3 ! !$3)$ !"&3)%)*+,- .( !(% !$& &&" "% &*+,- ." ) ( % && $ )*+,- ".( !' !% !) &' ') !'24526 021727/012088179772#372..7.9-:3-79-:37;.92779-::7232968 2#1221,73,-71221-2-92#,219-9:,-1/ "71/ " (32968#

Table 1: Export Value and Number of Exporters, by Product and Destination

26


! " # $%&'( "# ! "! ) "*'+ # !# # ! #* " " " " " " "

#

! * " ! ! $%&'( " ! ! "# ! ! ")'+ !# * ! ! #) " " " " " " "

$',,(' #

- - -# -! - -" -$%&'( # ) '+ " -" * ./0&10-2%&/330,,'% 04(44&''%'4# &5&(+'%&/54#

Table 2: 2004 versus 2005 Export Value Market Shares, by Type of Firm, Product and Destination

27

Embedded Institutional Reform !"# $ $%% $ $ & '!#( $ ) ! !" * ) + ,!) !%(!&!!$!-!!!#(!."!/!-!"!!0!%1!2!#3!!4'!!!$!!!'!! !'!!!0!'!!%!!! !!!0!! !Table 3: Decomposition of 2004 to 2005 Changes in MFA-UEC Market Share

28


!" #$#%& #$#'( #$#(# #$##) #$#%* #$#'' #$##& #$##*+ #$#)' #$#*' #$#(& #$#((

, + #$#&# #$#'- #$#%' #$#''

./ + #$#0- #$#%- #$#'& #$##'

+1, / #$'() #$(#' #$#)& #$#%0#

+

!" () (# - % (0 (# * '+ %% '# & -

, + %) (' (- (#

./ + %# (0 (% (

+1, / (## *) %0 (0,2+3343'##*'##- + 3"54647"3$833"""3""9$$93.$.434 537"3"74"45 "3935$

Table 4: Decomposition of MFA-UEC vs OTC-UEC Export Price Declines Between 2004 and 2005

29


!"#

$%

&'$

(

(

(

(

(

(

(

(

(

%

))*

+

+

+

,-

,-,-

,-(.

#$/

,-(.

#$/

,-(.

#$/

$0

.

$'#

! 1

! 2

!"#

$%

&'$

(

(

(

(

(

(

(

(

(

%

))*

+

+

+

,-

,-,-

,-(.

#$/

,-(.

#$/

,-(.

#$/

$0

.

$'#

+3

%0

3%

0

+

4

55$5

$&

6-$0

$

)

$

5$%

#**

#$

/5$**

70

$7

*70%#

7

$

8

$0

%*

$)

977

$7

5$%

#**

#$

/#

':*

"#

3

85

$5

$$

#)

8

$$0$

'

#97

5$

%

*$

$

'

)977

$7

5$%

#**

#$

/#

':*

'%

0"#

((97

7$7

)$

*%

5

$*

%%5

*#%

5$

%#*) %

))*

%$07

5

*#

%5$

%#*

%*

#$/

) %

))*

7

)

$9)

7

5%

'

85

$5

$7

*

$'#

)7

%

8$0

7

8

%

$05$

**7

0!0$

85

$$

*%

;.

%&.;.

5$%#

**#

$/5

$<8

)7

5$%$

5#

97

%%

05$

%#*%

*#

$/)

%))*

-%

$%$$

$$

*#

$%

7

5$%

#*

-0)*

*4<

<

Table 5: Relative MFA-UEC Export Price Declines Between 2004 and 200530


!"

#$!#%

##$!""

#!%&#'$

!'($

!#)

#!

"$!#

$!'

!#"

!(%*)+,

-).) ) /

-0,0) 1)

23

2'3!4

1) - ))1) )

/-0./ --

/ !1 -)/.

) . )

/ 0.1) -

) 0!

-) ) )

) -)

) / !

-.

1 -)/.)1

' 1 )

.,!-)

-)/.

-./0-

.

) /-0./

!1 )-

/.)11

)!

Table 6: Aggregate TFP Gain Following Quota Removal

31


!"

#

$#

%%

#$

"

#"

&!'

(

#$

(

)$

* !

' !'

+(% '

",-

./,(/

./!' ,0

- 1%2

3!4%1 &

-' '5-'

% '

!' 3!45 &)

" & &

1 &%

1

Table 7: Aggregate TFP Gain Following Quota Removal

32


0.5

11.

5C

hang

e in

Wei

ghte

d-A

vera

ge P

rodu

ctiv

ity

-.05

-.04

-.03

-.02

-.01

0P

erce

nt C

hang

e in

Ave

rage

Pro

duct

ivity

0 .2 .4 .6 .8Quota Restrictiveness

Average Productivity Weighted-Average Productivity

Productivity Change Between 'No Quota' and 'Efficient Quota'

Figure 1: Numerical Solution: Change in Exporters' Average Productivity

33


02

46

810

Bill

ions

of U

SD

OTC-UEC MFA-UEC

Note: percentages record portion of overall growth accounted for by net entry

By Group and Margin2005 Export Value Growth

Incumbants Net entry

0

100

200

300

400

Per

cent

OTC-UEC MFA-UEC

Note: percentages record portion of overall average growth accounted for by net entry.

By Group and Margin2005 Average Export Quantity Growth

Incumbants Net entry

Figure 2: Export Growth by Year, Group and Margin

34


-.6

-.4

-.2

0

2004

-200

5 C

hang

e in

Mar

ket S

hare

0 .2 .4 .6 .8 12004 Market Share (Among All Firms)

OTC-UEC SOEs OTC-UEC Domestic OTC-UEC Foreign

MFA-UEC SOEs MFA-UEC Domestic MFA-UEC Foreign

Note: Lines generated by lowess smoothing.

Among All FirmsChange in Market Share vs Initial Level

Figure 3: MFA-UEC Incumbents's 2004-5 Change in Market Share vs Initial 2004 Level

35


Figure 4: Average 2004 to 2005 Change in Quantity-Based Market Share, MFA-UEC vs OTC-UEC

36


-10

010

2030

40P

erce

nt

OTC-ROW MFA-ROW OTC-UEC MFA-UECNote: Product-countries in first and ninety-ninth percentiles are dropped from each distribution.

By GroupAverage 2004 to 2005 Price Change

Figure 5: Average Export Price Growth Across Product-Country Pairs, by Group

37


0.2

.4.6

.8D

ensi

ty

.125 .5 1 2 8Ratio

Incumbents Entrants Exiters

First and ninety-ninth percentiles are dropped from each distribution.

By MarginMFA-UEC Firm Prices Relative to 2004-5 Average

Figure 6: MFA-UEC Export Prices Relative to the Average Export Price Across All Firms in 2004and 2005, by Margin

38


0.2

.4.6

.8D

ensi

ty

.125 .5 1 2 8Ratio

2004 Incumbents 2004 Entrants 2003 Exiters


By MarginMFA-UEC Firm Prices Relative to 2003-4 Average

0.2

.4.6

.8D

ensi

ty

.125 .5 1 2 8Ratio

2005 Incumbents 2005 Entrants 2004 Exiters


By MarginOTC-UEC Firm Prices Relative to 2004-5 Average

Figure 7: Exiters versus Entrants in 2005 OTC-UEC and 2004 MFA-UEC

39


0.5

11.

52

Den

sity

.0625 .125 .5 1 2 8 16 32Ratio

SOE Domestic Foreign

First and ninety-ninth percentiles are dropped from each distribution. Collective firms are excluded.

by Ownership2005 TFP, Textile & Clothing Exporting Firms

Figure 8: Textile and Apparel Producers' TFP, 2005

40


010

2030

Per

cent

OTC-ROW MFA-ROW OTC-UEC MFA-UECNote: Product-countries in first and ninety-ninth percentiles are dropped from each distribution.

By GroupAverage 2004 to 2005 Quality Change

Figure 9: Average Export Quality Growth Across Product-Country Pairs, by Group

41


Figure 10: TFP Gains Implied by Ecient and Inecient Allocation

42


0.690.660.630.600.570.530.500.470.430.40

0.360.33

0.290.26

0.22

0.19

0.15

0.11

0.07

0.040.00

0.2

.4.6

.81

Net

Ext

ensi

ve M

argi

n C

ontr

ibut

ion

to G

row

th

1 2 3 4 5Aggregate Productivity

Inefficient Allocation

Figure 11: Counterfactual TFP under Inecient Allocation

43

when institutional reform is embedded in trade ... · when institutional reform is embedded in...

Documents