a contest model of liberalizing government procurements

European Journal of Political Economy 25 (2009) 479–488

Contents lists available at ScienceDirect

European Journal of Political Economy

j ourna l homepage: www.e lsev ie r.com/ locate /e jpe

A contest model of liberalizing government procurements

Ngo Van Long a,b,c,⁎, Frank Stähler c,d

a Department of Economics, McGill University, Montreal, Quebec, Canada H3A 2T7b CIRANO, Montreal, Canadac CESifo, Munich, Germanyd Department of Economics, University of Würzburg, Sanderring 2, D-97070 Würzburg, Germany

a r t i c l e i n f o

⁎ Corresponding author. Department of Economics,E-mail address: [email protected] (N.V. Long).

0176-2680/$ – see front matter © 2009 Elsevier B.V. Adoi:10.1016/j.ejpoleco.2009.02.003

a b s t r a c t

Article history:Received 12 January 2008Received in revised form 8 February 2009Accepted 8 February 2009Available online 1 March 2009

This paper models liberalization of government procurements as admitting entry of foreignfirms in a contest among potential rent seekers. It contributes to the literature on howinstitutions influence socially desirable outcomes. Liberalizing procurements reduces wastefuldomestic lobbying but also increases the likelihood that a foreign firm will capture the rent. Amain result is that the domestic welfare change is not monotonic in the foreign firms' abilities.Furthermore, we show that domestic liberalization policies can be inefficient from the globalpoint of view because foreign lobbying costs can outweight productive efficiency.

© 2009 Elsevier B.V. All rights reserved.

JEL classifications:F12F13H57

Keywords:Trade liberalizationTrade in servicesGovernment procurements

1. Introduction

While liberalization of trade in goods has been progressing for decades, liberalizing trade in services seems to face substantialreluctance, in particular when the services relate to government contracts. Governmental procurements are quite substantial in moderneconomies: theOECDestimates that public procurement bygovernments andby state-ownedenterprises amounts to about 15%ofGDP inOECDcountries (seeOECD,2005). There is a lotof evidence indicating a strict preference fordomestic agents orat leastmoreprotection forpublic procurement agencies than for private firms (see for example Trionfetti, 2000). The WTO had established a Working Group onTransparency in Government Procurement in order to enhance transparency in procurement decisions and prepare an internationalagreement. However, members could not agree on the launch of negotiations in 2003 and referred this issue back to the WTO GeneralCouncil. The Council decided in 2004 that this issue should not be taken any further and should not form part of the Doha RoundProgramme. This decision basically allows a country to set up its own rules and procedures unless they are in conflict with supranationalprovisions, such as rules set up by the European Union, or if the country has joined the WTO Agreement on Government Procurement.

This agreement was negotiated during the Tokyo Round and is meant to set up non-discriminatory rules for governmentalprocurements as to allow competition by foreign firms. It has 28 members, including the U.S., Canada and the EU. Developing countriesare obliged to comply with the rules as a condition of receiving foreign aid. However, it is questionable whether this agreementreally bites in practice. For example, Europe's EADS hadwon a USD 35 billion deal with the U.S. Defense Department on the delivery of airforce tankers, but this deal was revoked after EADS's rival, Boeing, and Boeing supporters in the U.S. Congress complained about the deal.

In this paper, we take the view that procurement outcomes are largely influenced by the interests of powerful groups and by theinstitutional set-up of the countries concerned. It is well known that bidding for a governmental contract does not resemble an

McGill University, Montreal, Quebec, Canada H3A 2T7. Tel.: +1 514 398 4844; fax: +1 514 398 4938.

ll rights reserved.

mailto:[email protected]

http://dx.doi.org/10.1016/j.ejpoleco.2009.02.003

http://www.sciencedirect.com/science/journal/01762680

480 N.V. Long, F. Stähler / European Journal of Political Economy 25 (2009) 479–488

auction in the usual sense, because government officials take into account many additional criteria beyond the bids. For example, arecent survey shows that firms often have to offer a substantial share of the contract value as additional illicit payments to secure aprocurement contract (see Transparency International, 2007, p. 320.) These observations suggest that governmental procurementdecisions may be strongly influenced by partisanship or even corruption. Although this practice is to be contained by the OECDAntibribery Convention, this convention seems to fail to a large extent. Auriol (2006) quotes aWorld Bank estimate that about USD200 billion are spent on public procurement bribery per year, which amounts to about 3.5% of public procurement spending. Shealso demonstrates that the deadweight loss is substantial.

Given this background, we propose to model government procurements as an imperfectly discriminating contest in whichfirms lobby for winning (Tullock,1980,1991; Rowley et al., 1988; Hillman and Riley, 1989; Nitzan,1994; and several other papers inCongleton et al., 2008). The winner may run the project and will realize a surplus from the project. The size of this surplus,however, may depend on the type of the winning firm. In particular, foreign firms may have different abilities to run the projectcompared to domestic firms. Wewill explore conditions under which the inclusion of foreign firms will improve domestic welfare.We also include an analysis of global welfare effects as we would like to learn whether a bilateral liberalization of governmentprocurement will increase world welfare. For this purpose, we use an asymmetric contest model in which domestic and foreignfirms have different abilities.1 Concerning domestic welfare, we note two important points. First, rents that accrue to domesticcontestants form part of social welfare, while those that accrue to foreign firms are not. Second, the real domestic resources used inlobbying by domestic contestants are socially wasteful, but real domestic resources used by foreign contestants are domesticexports. We determine conditions under which a country is better off by allowing foreign contestants to participate. Our modelhighlights the fact that institutions influence socially desirable outcomes.

Although the literature on procurement discrimination is vast, this paper is to our knowledge the first which discussesgovernment procurement policies in an asymmetric contest model.2 McAfee and McMillan (1989) show in an auction model thatexcluding foreign firms may enhance competition among domestic firms and can thus be welfare-improving. Branco (1994) andVagstad (1995) extend the Brander and Spencer (1981) analysis to procurement and show that the profit-shifting incentive maycall for the exclusion of foreign firms. In our setting, however, domestic and foreign firms do not compete by bids on a level-playingfield. Instead, foreign and domestic firms are asymmetric from the outset, and lobbying by each firm uses up real resources andenhances its probability of winning.

The remainder of the paper is organized as follows. Section 2 explains the contest model, Section 3 discusses the domesticwelfare effect of liberalizing government procurements and Section 4 deals with the global welfare effects. Section 5 concludes.The details of our proofs are relegated to the Appendix.

2. The model

We assume two countries, a domestic country and a foreign country which will be denoted by a star. A project is available in thedomestic country. Let n (n⁎) denote the number of domestic (foreign) contestants for this project. Domestic and foreign contestantsdiffer in terms of their abilities: a winning domestic (foreign) firm will generate a gross surplus V (V⁎) from the project. Notecarefully that differences between V and V⁎ may also stem from different productivities across borders to run the project. Forexample,VNV⁎may include the casewhere the revenues from this project are the same for both domestic and foreign firms, but theforeign firms may have a comparative cost advantage in running this type of projects due to past experience in this business.Furthermore, the size of V⁎ will also depend on the institutional setup of the contest. Foreign firms may have to comply withcountry-specific regulations, may have tomeet local content requirements or may be required to take on board a local partner, andthese requirements are likely to reduce V⁎.Wewill therefore also discuss howchanges in V⁎will affect domestic and globalwelfare.

We take both n and n⁎ as exogenous. To focus on the interesting case where foreign firms have interest in participation, wemake the following assumption:

Assumption 1. V⁎N(n−1)V/n.

As will be seen below, Assumption 1 implies that foreign firms will actively lobby. Let si (si⁎) denotes individual domestic(respectively, foreign) lobbying. Let wi (wi

⁎) denote the probability that a domestic (foreign) firm i wins the contest. A generalformulation of the contest success function is

with h

1 Asy2 For

framewprefere

3 If hstrategiresults.

wi =h sið ÞP

n h sið Þ + Pn⁎ h si⁎

� � ; wi⁎ =h si⁎ð ÞP

n h sið Þ + Pn⁎ h si⁎

� �

(0)=0, h′(0)N0 and wi=1/(n+n⁎)=wi⁎ if si=si⁎=0 for all firms.3

mmetry in abilities amongcontestantswere dealtwith inNti (1999) andStein (2002). For a general paperon asymmetric contests, see Cornes andHartley (2005).an overview of the literature, see for example Evenett and Hoekman (2005). Procurement has also been modelled in an asymmetric informationork; see for example Mougeot and Naegelen (2005) who consider a foreign and a domestic firm whose costs are private information and discussntial treatment under different political economy assumptions.″(si)≤0, the second order condition for each player's payoff maximization problem will be satisfied. In this case, with some additional mild assumption,c substitutability will obtain (that is, if one contestant increases lobbying, the others will reduce their lobbying). Strategic substitutability drives our main

481N.V. Long, F. Stähler / European Journal of Political Economy 25 (2009) 479–488

Tullock (1980) proposed that h(s)=sα where αN0. Hillman and Riley (1989, Section 6) considered the case in which α=1because of greater analytical tractability. For the same reason, we follow the Hillman–Riley specification.4

Let S=Σsi (respectively, S⁎=Σsi⁎) denote aggregate domestic (foreign) lobbying. The expected profit of each domestic (foreign)contestant, denoted byΠi (Πi

⁎), is the probability ofwinning the contest times the surplus net of taxes,minus the lobbying effort:

4 If wfirm anform soaggregathis cremore gesimulat

5 In Cof firms

Πi =si

S + S⁎1− tð ÞV − si;

Πi⁎si⁎

S + S⁎1− tð ÞV ⁎ − si⁎

: ð1Þ

x rate t is exogenous and equal to the corporate profit tax rate in the country under consideration. Assume 0b tb1. Profits are
The tataxed at their source. The first-order conditions and symmetry within each group of firms lead to
si = max 0; n + n⁎ − 1ð Þ 1− tð ÞVV⁎ n⁎V − n⁎ − 1ð ÞV ⁎ð Þn⁎V + nV ⁎ð Þ2

� �if s⁎i N 0;

si =n − 1ð Þ 1− tð ÞV

n2 if si⁎ = 0;

si⁎ =n + n⁎ − 1ð Þ 1− tð ÞVV⁎ nV ⁎ − n − 1ð ÞVð Þ

n⁎V + nV ⁎ð Þ2 if si N 0;

si⁎ =n⁎ − 1ð Þ 1− tð ÞV ⁎

n⁎2if si = 0:

ð2Þ

n lobbying — if allowed — is always positive due to Assumption 1 but domestic lobbying may not be profitable.
ForeigLiberalization is defined as a move from autarky (i.e., when only domestic firms are allowed to contest) to a regimewhere all n⁎
foreign firms are allowed to contest. We distinguish two cases, which depend on parameter values. In Case 1, where the foreignsurplus V⁎ is very large, foreign contestants will lobby very strongly and thus render unprofitable any lobbying effort of domesticcontestants. According to Eq. (2), this will happen if V⁎≥Ve≡n⁎V /(n⁎−1). In Case 2, V⁎bVe, and consequently domestic lobbyingwill continue to be profitable after liberalization. We call Ve the foreign-dominance threshold level of gross surplus of foreign firms,beyond which domestic firms will not lobby.

Definition 1. The foreign-dominance threshold level of gross surplus of foreign firms is

Vun⁎V = n⁎ − 1� �

:

Note that the greater is the number n⁎, the lower is the threshold level Ve. The foreign firms are said to be dominant if V⁎≥Ve. Anecessary condition for foreign firms to be dominant is V⁎NV.

Domestic welfare is equal to the sum of expected net profits of domestic firms and expected tax revenues. Withoutliberalization, n⁎=si⁎=S⁎=0, and this will serve as our benchmark. As tax revenues cancel out in this case, the domestic welfareunder autarky is the difference between the domestic gross surplus and aggregate domestic lobbying. Denote this by WAU wherethe superscript AU means autarky, or pre-liberalization.

WAU = V − S = V − n − 1ð Þ 1− tð ÞVn

=1 + n − 1ð Þtð ÞV

n: ð3Þ

3. The effects of liberalization on domestic welfare

Let us now calculate domestic welfare under liberalization. For expositional clarity, we treat Case 1 and Case 2 separately.5

Case 1 (the case of dominant foreign firms) is the easiest to deal with, because domestic lobbying is zero. Foreign lobbying is onlywasteful for the foreign country. If foreign firms use domestic resources for lobbying, such resources are exports of services from

e set h(s)=sα where α≠1, closed form expressions cannot be obtained, except in the special case where n=n⁎=1 (so that there is only one domesticd one foreign firm). Nti (1999) was able to find closed form solution only for the two-contestant case. Cornes and Hartley (2005) point out that closedlution is not available in the case of an arbitrary number of contestants. They can nevertheless show that a contestant's effort, expressed as a share ofte effort, is always decreasing in aggregate effort, if α≤1, or more generally, with decreasing returns. (If αN1, the payoff function is not single-peaked, andates serious difficulties, see Cornes and Hartley, 2005, Section 5). This leads us to believe that our results would not change qualitatively when α≤1, ornerally, with decreasing returns. However, the analysis of changes in domestic welfare would be very cumbersome and would rely heavily on numericalions.ase 1, domestic welfare is linear in V⁎ since domestic firms drop out of the contest, while in Case 2, domestic welfare is quadratic in V⁎ since both groupsare active contestants.


the domestic to the foreign country. Accordingly, without domestic lobbying, domestic welfare is equal to the tax revenues, tV⁎.Comparing tV⁎ with welfare under autarky, WAU in Eq. (3), leads to:

Proposition 1. Assume dominant foreign firms, i.e., V⁎≥Ve. Then domestic lobbying falls to zero after trade liberalization, and domesticwelfare after liberalization is linear and increasing in V⁎. Given V⁎≥Ve, liberalization improves domestic welfare if and only if

which

with

Our n

6 Bot

V⁎Nnt + 1− tð Þ

ntVuV Vu

WAU

tð4Þ

is equivalent to the condition

t N tVuV

nV⁎ − n − 1ð ÞV ð5Þ

Proposition 1 demonstrates that, given any positive tax rate t, dominant foreign contestants will improve domestic welfare, ifthey are substantially outnumbered by potential domestic contestants. As can be seen from Eq. (4), since V⁎NV, it follows that forany givenpositive tax rate t, V⁎NV′ if n is sufficiently large before liberalization. Alternatively, for any given n≥1, trade liberalizationin the case of dominant foreign firms will improve domestic welfare if the tax rate is sufficiently high. Since V⁎≥Ve in Case 1,

tVbn⁎ − 1

n + n⁎ − 1: ð6Þ

, a sufficient condition for liberalization to improve domesticwelfare in Case 1 is that the tax rate exceeds (n⁎−1)/(n+n⁎−1).
HenceThe welfare gain comes from both the elimination of wasteful domestic lobbying and the tax revenue from the foreign firm.
Note that a high tax rate also implies low level of lobbying under autarky, hence the benefits from liberalization do not mainlycome from avoiding wasteful lobbying but from taxing the high gross surplus of a foreign firm.

In Case 2, where foreign firms are not dominant, domestic lobbying and foreign lobbying co-exist under liberalization. Theforeign gross surplus is smaller than the threshold Ve. Similar to Case 1, liberalization has two opposite effects on a country's welfare.First, it improves welfare because wasteful domestic lobbying is reduced and replaced by foreign lobbying. Second, it reducesdomestic (expected) welfare as a foreign contestant may win the contest, resulting in the surplus net of taxes being transferredabroad. Appendix A.1 deals with details on the welfare gains when both domestic and foreign firms actively lobby. Let Δ denotethe domestic welfare change due to liberalization. It is shown in Appendix A.1 that, for V⁎ in the non-empty range [(n−1)V /n,n⁎V /(n⁎−1)] (so that both groups of firms actively lobby), Δ can be expressed as a cubic function of V⁎:

Δ V⁎� �

= n2tA V⁎� �

X V⁎� �

; ð7Þ

X V⁎� �

= V⁎2 −2n 1− tð Þ− n⁎n + n2tn2t

V⁎V −n⁎ 1− tð Þ + nn⁎

n2tV2

: ð8Þ

A V⁎� �

=n⁎ nV⁎ − n − 1ð ÞV� �n n⁎V + nV⁎� �2 N 0 forV⁎

Nn − 1ð ÞV

n: ð9Þ

ext proposition shows that the foreign surplus must be sufficiently large as to make liberalization worthwhile.

Proposition 2. Assume that n − 1ð ÞVn bV⁎b n⁎V

n⁎ − 1u V , so that both groups of firms lobby. (1)Liberalization results in a negative domesticwelfare change if foreign contestants produce the same surplus (V=V⁎), (2) If Ve≤V′, then liberalization results in a negative domesticwelfare change. (3)If VeNV′, then there exists a critical surplus level V

—⁎, where VbV—⁎bVe and (a) liberalization results in a negative

domestic welfare change if V⁎a n − 1ð ÞVn ;V

⁎� i

, (b) liberalization results in a positive domestic welfare change if V⁎∈ [V—⁎, Ve].

Proof. See Appendix A.1. □

The intuition for part 1 is simple: domestic welfare declines because the reduced waste of lobbying resources from lowerdomestic lobbying is small relative to the risk of rent being shifted to foreign firms. The other two parts demonstrate that welfaredeclines with liberalizationwhen domestic and foreign contestants co-exist and V⁎ is not very large, but a sufficiently large V⁎willguarantee a positive welfare effect.

The results stated in Propositions 1 and 2 can be illustrated best by two numerical examples. These two examples highlight theU-shaped relationship between foreign firms' productivity and domestic welfare. Both examples use the same parameter valuesexcept for the number of foreign contestants.6 In the first (respectively, second) simulation, we set n⁎=30 (respectively, n⁎=5) so

h simulations assume n=20, V=100 and t=0.3; n⁎=30 implies VebV′ (see Fig. 1) and n⁎=5 implies VeNV′ (see Fig. 2).

Fig. 1. Domestic welfare change for V~b V′.


that the threshold level Ve is below (respectively, above) V′. Due to our parameter specification, foreign contestants become active ifV⁎ is larger than 95. The value of V′ is equal to 111.67 in both simulations.

Fig.1 applies to Proposition 1 (for V⁎≥Ve=103.45) and also to part 2 of Proposition 2 asVe=103.45bV′. In this case, coexistence ofboth domestic and foreign contestants (i.e., when V⁎ is in the range (95, 103.45)) will unambiguously deteriorate domestic welfarerelative to the pre-liberalization level. The change in welfare is U-shaped and non-monotonic, however. If the foreign surplus islarger than Ve, foreign firms dominate and domestic welfare increases linearly with V⁎, but the welfare effect of liberalization is stillnegative until V⁎ reaches V′. It is positive only for V⁎ greater than V′.

Fig. 2 applies to Proposition 1 (for V⁎≥Ve=125) and also to part 3 of Proposition 2 as Ve=125NV′. For V⁎N125, foreign firmsare dominant, and in this case liberalization will unambiguously improve domestic welfare. However, domestic welfare improvesalso for foreign surplus levels in the range (115, 125) where both domestic and foreign firms lobby. Once again, the welfare gain isU-shaped in the coexistence range, and liberalization leads to domestic welfare losses for low levels of V⁎, i.e., for V⁎ in the range(95, 115).

While domestic welfare level is non-monotone in V⁎, the following Corollary shows that there is a monotone relationshipbetween policy preference and the excess of V⁎ over V. Let E≡V⁎−V.

Corollary. Let E denote the productivity difference between the representative foreign firm and the representative domestic firm. Then(1) If E is negative, the domestic country does not want to liberalize trade. (2) If E is positive, the domestic country (a)does not want toliberalize trade if E is below a certain positive threshold E

_, and (b) wants to liberalize trade if E is above E

_.

Proof. This follows from Propositions 1 and 2. □

Note that the corollary determines the preference of the domestic country in terms of a specific procurement only. The countryunder consideration may have credibly committed to liberalize trade, including governmental procurements. In this case, thepotential loss may be overcompensated by potential gains from trade in governmental procurements for which the domesticwelfare effect is positive and/or by gains from trade in other areas.

Note also that the imposition of additional requirements for foreign firms like local content requirements, etc., will potentiallyeat up any gains from liberalization. These requirement will reduce V⁎ and unless the regulation itself increases domestic welfare,they will make countries even more reluctant to allow foreign contestants.

Fig. 2. Domestic welfare change for V~N V′.


4. The effects of liberalization on global efficiency

In this section, we want to discuss how the impact of unilateral liberalization on domestic welfare will differ from impact onglobal welfare. From the viewpoint of global efficiency, both the (wasteful) lobbying efforts of the foreign firms and the expectedpayoff for foreign firms have to be taken into account. Note carefully that — compared to domestic welfare without tradeliberalization— the global expected surplus depends on thewinning probabilities unless V=V⁎. The global expected surplus is theexpected surplus of all domestic and foreign firms, i.e.,

Condi

so thadominwere t

These

nsiVS + S⁎

+n⁎si⁎V ⁎

S + S⁎:

gain, we have to distinguish between dominant foreign firms and co-existence of domestic and foreign contestants. For the
Once acase of dominant foreign firms, global welfare increases (decreases) if
1 + n⁎ − 1ð Þtð ÞV⁎

n⁎N bð Þ 1 + n − 1ð Þtð ÞV

nfV ⁎ N bð Þn⁎V 1 + n − 1ð Þtð Þ

n 1 + n⁎ − 1ð Þtð Þ : ð10Þ

tion (10) follows from the welfare autarky level (3); in the case of dominant foreign firms, global welfare is equal tosion (3) with domestic values replaced by foreign values. Two observations are noteworthy.
expres
First, both a negative and a positive welfare change are possible, since the foreign-threshold level which renders foreign firmsdominant does not have any bearing on expression (10). In fact, we can easily construct cases under which

n⁎V 1 + n − 1ð Þtð Þn 1 + n⁎ − 1ð Þtð Þ N V ⁎ N

n⁎Vn⁎ − 1

:

eans that a global welfare loss is possible evenwhen foreign firms dominate. Both inequalities hold if V⁎ is neither too large
This mnor too small and n(1− t)+1bn⁎(1− t). The latter condition is plausible: if there are too many foreign firms, their aggregatelobbying is excessive from the global efficiency point of view.
Second,when comparingexpression (10)withV′of Proposition1 (see Eq. (4)),we can easilyfindparameter values such that

V⁎ NWAU

tu1 + n − 1ð Þt

ntV N

n⁎ 1 + n − 1ð Þtð ÞVn 1 + n⁎ − 1ð Þtð Þ

t both world welfare and domestic welfare improve with domestic liberalization. Similarly, one can find parameter values
so thasuch that
V⁎bn⁎ 1 + n − 1ð Þtð ÞVn 1 + n⁎ − 1ð Þtð Þ b

WAU

t

t both world welfare and domestic welfare decrease after domestic liberalization. However, we can show that, in the case ofant foreign firms, if liberalization improves domestic welfare, it must also improve world welfare. Suppose the oppositerue, implying the existence of parameter values for which

n⁎ 1 + n − 1ð Þtð ÞVn 1 + n⁎ − 1ð Þtð Þ N V⁎

NWAU

tu

1 + n − 1ð Þtð Þnt

V :

values would imply that tn⁎N1+(n⁎−1)t or 0N1− t, which is not possible.

Lemma 1. In the case of dominant foreign firms, (i) trade liberalization may be globally welfare deteriorating for intermediate levels ofV⁎, and (ii) global welfare cannot decline when domestic welfare increases.

We now turn to the case of co-existence of domestic and foreign firms. Appendix A.2 has all the details for this case, andwe findthe following result:

Lemma 2. In the case where both groups of firms are lobbying, global welfare increases (decreases) if

V⁎N bð Þ n + 1− tð ÞV

n: ð11Þ


Proof. See Appendix A.2. □

Global welfare may decline or increase also in the case where both foreign and domestic contestants lobby. This can bedemonstrated by considering the special case of symmetric foreign and domestic firms, i.e., if V=V⁎. Global welfare in this case isgiven by

1 + N − 1ð Þtð ÞVN

:

N=n+n⁎ denotes the total number of domestic and foreign firms. Trade liberalization increases the number of
wherecontestants, and differentiating global welfare with respect to N yields −(1− t)V /N2b0. Thus, global welfare unambiguouslydeclines with trade liberalization (as does domestic welfare) if firms are symmetric.
Furthermore, condition (11) is also compatible with the condition of domestic welfare improvement of Proposition 2. We caneasily find parameter values that are consistent with both Eq. (11) and

n + 1− tð ÞVn

N V⁎

q. (A.5) in Appendix A.1. Hence, in the case of co-existence of domestic and foreign contestants, global welfare may decline
(see Ewhen domestic welfare improves.
When domestic and foreign firms co-exist, why can global welfare decline when domestic welfare increases at the same time?The intuition is as follows.While domestic welfare does not take the foreign rent into account, it also does not include as a negativeterm the wasteful foreign lobbying. If the foreign surplus is moderate (so that both groups of firms contest after liberalization), theincrease in productive efficiency if a foreign firmwins is relatively small. However, there can be a huge increase in foreign lobbyingefforts. From the viewpoint of the domestic country, however, the increase in tax revenue associated with higher productiveefficiency and the decrease in domestic lobbyingmay be sufficient to overcompensate the expected repatriated profit of the foreignfirm. Our final Proposition summarizes these results.

Proposition 3. From the point of view of global welfare, a domestic-welfare-enhancing liberalization policy may be either excessive(i.e. global-welfare-reducing because of its encouragement of wasteful foreign lobbying) or too restrictive in the case of a moderatelylarge foreign surplus.

Proof. This follows from Lemmas 1 and 2. □

The case of excessive trade liberalization can occur only for a moderate foreign surplus. If the foreign surplus V⁎ is sufficientlyhigh, both domestic and global welfare will improve. Although (wasteful) foreign lobbying efforts increase with foreign surplus,this effect is overcompensated by sufficiently large tax revenues to make domestic welfare improve and by a sufficiently largeexpected surplus to make global welfare improve.

5. Concluding remarks

This paper has contributed to the literature on how institutions influence socially desirable outcomes. We have shown thatif winning firms are selected on the basis of an imperfectly discriminating contest, a country may lose from liberalizinggovernment procurements if foreign contestants are not able to generate a substantially larger surplus than domestic firms.For low levels of foreign surplus, the gain from reducing wasteful domestic lobbying is more than offset by the increasedlikelihood that the rent will be captured by a foreign firm. Hence, countries can be expected to be more likely to agree toliberalization of government procurements only if the productivity of foreign firms is substantially larger than that of domesticfirms.

We have also compared the desirability of outcomes viewed from domestic efficiency perspectives and from global efficiencyperspectives. The results are mixed. For low levels of surplus genera by foreign firms, we may find that domestic liberalizationpolicies are excessive from the globalwelfare point of view.While domesticwelfare improves, globalwelfaremaygo down, becauseforeign firms excessively lobby for success in the domesticmarket. Hence, both an excessive or a too restrictive domestic policymaybe observed.

Of course, even if the welfare effect of liberalizing government procurements is negative, the overall welfare effect of tradeliberalization, including trade of goods and service on private markets, may still be positive and substantially large. Furthermore,changing the rules of the game such that governmental procurement is turned from an imperfectly discriminating contest into areal auction in which partisanship does not play any role will definitely make a positive contribution to domestic and globalwelfare.While positive gainsmay be observed in highly integrated areas such as the European Union, some empirical evidence stillpoints to the other direction. If governmental procurements continue to resemble a Tullock contest, our paper has shown thatliberalization does not necessarily qualify as a source of gains from trade.


Acknowledgments

We thank the referees for helpful comments. This paper was started when Ngo Van Long visited the Department of Economicsat the University of Otago, New Zealand, as a William Evans Fellow and was continued when Frank Stähler visited CIREQ at McGillUniversity. We also gratefully acknowledge the financial supports from the Social Sciences and Humanities Research Council ofCanada and the Fonds québécois de la recherche sur la société et la culture.

Appendix A.1. Domestic liberalization effects

Maximization of Eq. (1) w.r.t. si and si⁎, respectively, yields individual lobbying levels of

and T

and Slevel a

si =n + n⁎ − 1

� �n⁎V − n⁎ − 1

� �V⁎

� �1− tð ÞVV⁎

n⁎V + nV⁎� �2 ;

s⁎i =n + n⁎ − 1

� �nV⁎ − n − 1ð ÞV

� �1− tð ÞVV⁎

n⁎V + nV⁎� �2 :

ðA:1Þ

Since S=nsi and S⁎=n⁎si⁎, the probability that a given domestic [respectively, foreign] contestant wins is (n⁎V−(n⁎−1)V⁎)/(n⁎V+nV⁎) [respectively, (nV⁎−(n−1)V)/(n⁎V+nV⁎)], and the probability that the winner is a domestic [respectively, foreign]contestant is (n⁎V−(n⁎−1)V⁎)n / (n⁎V+nV⁎) [respectively, (nV⁎−(n−1)V)n⁎ / (n⁎V+nV⁎)]. Thus aggregate welfare isequal to B+T−S, where B is the gross surplus of a domestic firm multiplied by the probability that a domestic firmwins,

Bun⁎V − n⁎ − 1

� �V⁎

n⁎V + nV⁎nV ; ðA:2Þ

is the tax revenue from a winning foreign firm, multiplied by the probability that a foreign firm wins,

TunV⁎ − n − 1ð ÞV

� �n⁎V + nV⁎

n⁎tV⁎; ðA:3Þ

is the resource costs incurred by domestic firms (wasteful) lobbying. The difference between B+T−S and the autarky welfareccording to Eq. (3) yields expression (7). Consider the graph of Δ(V⁎). Now, at V⁎=V, Ωb0, hence Δ(V⁎=V)b0. This provesof Proposition 2. Observe that at V⁎=(n−1)V/n, Δ=0. Furthermore, Δ=0 if Ω=0.
part 1
Consider the following cubic equation in V⁎

Δ V⁎� �

= nt2A V⁎� �

X V⁎� �

= 0

This cubic equation has three roots, one at V⁎=(n−1)V/n, at which A(V⁎)=0. The other two roots are solutions of the quadraticequation Ω(V⁎)=0.

Let the two roots of the quadratic equation forΩ=0 be denoted by V⁎(1) and V⁎(2). According to Vieta's Theorem, the productof the roots is

V⁎ 1ð ÞV⁎ 2ð Þ = −n⁎ 1− tð Þ + nn⁎

n2tV2

b0: ðA:4Þ

shows one root, say V⁎(1), must be negative and the other root, V⁎(2), is positive. Since Ω(V⁎) tends to infinity as V⁎ tends
whichto plus or minus infinity, and Ω(V⁎=V)b0, it follows that the positive root V⁎(2) is larger than V.
Differentiating Δ at V⁎=(n−1)V/n yields

dΔ V⁎ = n − 1ð ÞV = n� �

dV⁎= −n⁎n 2− tð Þ + 2n n − 1ð Þ

n2 b0

This shows that the gain from liberalization is negative for low levels of V⁎. Furthermore,

X V⁎ = V� �

= − V2 2n + n⁎� �

1− tð Þb0:

Consider two cases:

Case I, where Ve≤V′, andCase II, where VeNV′.Under Case I, where Ve≤V′, our Proposition 1 shows that at V⁎=Ve, no domestic firm will lobby, and liberalization reduces

welfare, i.e., Δ(V⁎=Ve)b0. Note that the welfare change is continuous in V⁎. It follows that, in Case I, the root V⁎(2) must exceed Ve,and therefore is irrelevant. (Recall that for V⁎NVe, only foreign firms lobby). This proves part 2 of Proposition 1.


Next, consider Case II, where VeNV′. The welfare gain is positive for V⁎=VeNV′, when foreign contestants dominate. Accordingto Eq. (A.4), a critical level V ⁎ (which is the positive root V⁎(2) of the quadratic equation Ω(V⁎)=0) must exist with Δ(V—⁎)=0.

In fact, we may write

X V⁎� �

= V⁎2 + bV⁎ + c

with

b =nn⁎ − n2t − 2n 1− tð Þ

� �V

n2t;

c = −nn⁎ + n⁎ 1− tð Þ

� �V2

n2tb0:

and it follows that

V⁎ =

−b +ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffib2 − 4c

p

2

=V2nt

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffin⁎ + 2 + n−2ð Þtð Þ2 − 2n⁎ 3n + 2− 2t2

� �q− n⁎ − 2� �

+ n − 2ð Þt� �

N 0:

Appendix A.2. Global liberalization effects

The joint lobbying effort of both types is equal to

Si + S⁎i = nsi + n⁎s⁎i =n + n⁎ − 1

� �1− tð ÞVV⁎

n⁎V + nV⁎; ðA:6Þ

Eq. (2) has been used. The probability of a domestic [foreign] firm winning is si /(Si+Si⁎)[si⁎ / (Si+Si⁎)] which leads to an
whereexpected surplus across all firms of
nsiV + n⁎s⁎i V⁎

Si + S⁎i=

n + n⁎� �

VV⁎ + nn⁎ V⁎−V� �2

n⁎V + nV⁎: ðA:7Þ

The difference between Eqs. (A.7) and (A.6) determines global welfare, denoted by WG:

WG V⁎� �

=nn⁎ V⁎−V

� �2+ n + n⁎

� �t + 1− tð Þ

� �VV⁎

n⁎V + nV⁎: ðA:8Þ

Differentiating w.r.t. V⁎ yields

dWG

dV⁎=

n⁎ 2nn⁎VV⁎ + n2V⁎2 + n + n⁎� �

t + 1− tð Þ− n 2n⁎ + n� ��

V2� �

n⁎V + nV⁎� �2 : ðA:9Þ

Global welfare improves (deteriorates) if WG is larger (smaller) than WAU. Setting WG equal to WAU yields two solutions,denoted by V⁎(3) and V⁎(4):

V⁎ 3ð Þ =n − 1ð ÞV

n;V⁎ 4ð Þ =

n + 1− tð ÞVn

:

V⁎(3) coincides with the lower bound of Assumption 1. Computing the first derivative of global welfare w.r.t. V⁎ for V⁎=V⁎(3)

gives

dWG V⁎ 3ð Þ� �dV⁎

= −n⁎ n + n⁎ − 1

� �2− tð ÞV2

� �n⁎V + nV⁎ 3ð Þ� �2 b0: ðA:10Þ

Expression (A.10) proves that global welfare declines with an increase of V⁎ for low levels of V⁎, and improves once V⁎ haspassed V⁎(4). This proves Lemma 2.


References

Auriol, E., 2006. Corruption in procurement and public purchase. International Journal of Industrial Organization 24, 867–885.Branco, F., 1994. Favoring domestic firms in procurement contracts. Journal of International Economics 37, 65–80.Brander, J.A., Spencer, B., 1981. Tariffs and the extraction of foreign monopoly rents under potential entry. Canadian Journal of Economics 14, 371–389.Congleton, R.D., Hillman, A.L., Konrad, K.A. (Eds.), 2008. 40 Years of Research on Rent Seeking 1: Theory of Rent Seeking. Springer, Berlin.Cornes, R., Hartley, R., 2005. Asymmetric contests with general technologies. Economic Theory 26, 923–946.Evenett, S.J., Hoekman, B.H., 2005. Government procurement: market access, transparency, and multilateral trade rules. European Journal of Political Economy 21,

163–183.Hillman, A.L., Riley, J., 1989. Politically contestable rents and transfers. Economics and Politics 1, 17–39, reprinted in Congleton, R.D., Hillman, A.L., Konrad, K.A.

(Eds.), 2008. 40 Years of Research on Rent Seeking 1: Theory of Rent Seeking. Springer, Berlin.McAfee, P., McMillan, J., 1989. Government procurement and international trade. Journal of International Economics 26, 291–308.Mougeot, M., Naegelen, F., 2005. A political economy analysis of preferential public procurement policies. European Journal of Political Economy 21, 483–501.Nitzan, S., 1994. Modelling rent-seeking contests. European Journal of Political Economy 10, 41–60.Nti, K., 1999. Rent seeking with asymmetric valuations. Public Choice 98, 415–430.OECD, 2005. Fighting Corruption and Promoting Integrity in Public Procurement. OECD, Paris.Rowley, C., Tollison, R., Tullock, G., 1988. The Political Economy of Rent Seeking. Kluwer Academic Publishers, Dordrecht.Stein, W., 2002. Asymmetric rent-seeking with more than two contestants. Public Choice 113, 325–336.Transparency International, 2007. Global Corruption Report 2007. Cambridge University Press, Cambridge, U.K.Trionfetti, F., 2000. Discriminatory public procurement and international trade. World Economy 23, 57–76.Tullock, G., 1980. Efficient rent seeking. In: Buchanan, J.M., Tollinson, R., Tullock, G. (Eds.), Toward a Theory of Rent Seeking Society. Texas A&M University Press,

College Station, pp. 97–112. Reprinted in Congleton et al, (2008).Tullock, G., 1991. Economics of Special Privilege and Rent Seeking. Kluwer Academic Publishers, Dordrecht.Vagstad, S., 1995. Promoting fair competition in public procurement. Journal of Public Economics 58, 283–307.

a contest model of liberalizing government procurements

Documents