Protectionist Lobbying and Strategic Investment

Protectionist Lobbying and Strategic Investment*

RICHARD DAMANIA

School of Economics, University of Adelaide,Adelaide, Australia

Why are some uncompetitive industry sectors so effective in lobbyingfor greater protection and support? This paper attempts to explain thelobbying success of these industries in terms of the strategic role ofinvestment in technology as a credible commitment device. By es-chewing potentially profitable investment opportunities firms crediblysignal to the government that the cost of a tariff reduction will besubstantial. This enables the firms to lobby more effectively for policyconcessions. Political considerations may therefore provide a signifi-cant incentive for firms to reject investment in newer technologies, evenwhen these lower production costs.

I IntroductionA growing body of literature suggests that some

older and uncompetitive industries often formhighly effective lobby groups, which resist reformssuch as the elimination of trade barriers.1 More-over, while successful in lobbying, such industrieshave at times been slow to adopt newer and moreefficient technologies.2 Theoretically, this findingseems paradoxical. Rapidly expanding �sunrise�industries with more resources at their disposal,ought to be better placed to lobby effectively andgarner more favourable treatment, than theirdeclining counterparts.

This paper attempts to address this issue byexploring the interaction between a firm’s invest-ment strategy and its lobbying influence ongovernment policy. It is shown that the level ofprotection received by domestic firms is influencedby their prior investment decisions. By precom-mitting to (older) relatively high cost productiontechniques, firms can tilt the political game withpolicy makers in their favour.The analysis is based on a domestic oligopoly

which competes with foreign producers. Thedomestic firms are protected by a tariff andtherefore have an incentive to lobby for greaterprotection. Lobbying is introduced into thisframework by drawing on the well establishedcommon agency model of political support devel-oped by Grossman and Helpman (1994), whichhas been widely applied in a number of contexts(see Fredriksson 1997; Damania 2001). Accord-ingly, it is assumed that a self-interested govern-ment cares about both, aggregate welfare and thepolitical contributions it receives from lobbygroups.3 Firms seek to influence government

*The author acknowledges with gratitude the helpfuland incisive comments of two referees. The usualdisclaimer applies.Correspondence: Dr Richard Damania, School of

Economics, University of Adelaide, Adelaide 5005,Australia. Tel.: + 61 8 8303 4933; Fax: + 61 8 82231460. Email: richard.damania@adelaide.edu.au

1 El-Agraa (1987) provides evidence based on interindustry studies. Some industry specific cases include:textiles in the USA (Dixit & Londregan 1995),agriculture in developed countries. Baldwin (1993) andGrossman and Helpman (1996) provide a generaldiscussion of this issue.

2 Discussion of the evidence is provided in Section IV.

3 Political donations influence the government’sdecisions because of their many uses, including fundingelection campaigns, retiring debt from previous electionsand deterring rivals.

57

� 2003. The Economic Society of Australia. ISSN 0013–0249.

THE ECONOMIC RECORD, VOL. 79, NO. 244, MARCH, 2003, 57–69

policy by offering political contributions to thegovernment, in the expectation of securing greatertariff protection. The government in turn, selectsthe policy that maximises its own welfare.4

It is shown that when lobbying occurs firmshave an incentive to underinvest in technology.Specifically, in a political equilibrium, the tariffwhich is set by the government depends on thelevel of political contributions that it receives, andthe welfare costs of the chosen policy. By adoptinga less efficient technology, the firm credibly signalsto the government that a reduction in tariffs willresult in substantially lower profits. In the polit-ical equilibrium, lobby group contributions arelinked to profits. Hence, a decline in profits leadsto a fall in political donations. A governmentwhich values political contributions is thereforeinduced to adopt a policy which mitigates thedecline in profits and contributions. In essence, byunderinvesting in technology, the firms need tospend less on political contributions, so thatlobbying for protection becomes more productive.In deciding on whether to invest in more

efficient equipment, the firm will trade off theusual cost and benefits of investment, against theneed to spend more on lobbying, as investment intechnology increases. Consequently, lobbyingdiminishes the net benefits from investment. Theanalysis therefore predicts that when governmentsare receptive to lobby group demands, higherlevels of protection may be associated withrelatively lower levels of investment in technology.That is firms adopt the �puppy dog� strategy(Fudenberg & Tirole 1984).This paper is related to two distinct strands of

literature: political economy models of protectionin declining industries and the strategic tradepolicy literature. The political economy literaturehas paid close attention to the role of specialinterest lobbying on trade policy. However, thesestudies ignore the effects of a firm’s investmentdecisions on lobbying incentives and policy out-comes. Long and Vousden (1991), Grossman andHelpman (1996) and Baldwin (1993) are examples

in this vein. In contrast, the strategic trade policyliterature focuses upon oligopolistic competitionand the strategic investment incentives of firms.These models ignore the influence of specialinterest group lobbying on policy decisions (e.g.Krugman 1984; Bouet 2001).This paper combines the political economy

approach, with the strategic trade policy models.The analysis is most closely related to Damania(2001) who explores environmental policy andlobbying in the context of a simple monopoly.5

However, to our knowledge all the existingliterature has thus far assumed that lobby groupformation and individual firm contributions toa lobby group are predetermined. This paperextends the literature by allowing lobby groupcontributions to be endogenously determined.Grossman and Helpman (1994) argue that theneglect of collective action problems is a signifi-cant weakness in the literature, similarly Perssonand Tabellini (2001, p. 175) note that the commonagency approach to lobbying �lacks an explan-ation of the process whereby some groups getpolitically organised…�. In what follows weinvestigate whether free riding on political contri-butions undermines efforts to form a lobby group.The paper derives an important result whichshows that lobbying remains both feasible andeffective, in a non-cooperative equilibrium. Theresults are therefore robust to the usual collectiveaction problems that make high levels of cooper-ation difficult to sustain.It is useful to note that the conclusions of this

paper contradict those of the strategic trade policymodels, when firms compete using quantities. Thestrategic trade literature predicts that investmentin technology rises with the level of protection.6

This occurs because higher levels of tariff protec-tion, shift profits from the foreign firm to thedomestic industry. Under Cournot competition,the adoption of a more efficient technology by onefirm, lowers its marginal costs and allows it to

4 As the focus of this paper is upon the effects oflobbying by producers, the role of an opposingconsumer lobby group is suppressed. This may bejustified by assuming that the benefits of a tariff areconcentrated, but the costs of protection are so thinlyspread that they do not provide sufficient incentive forindividuals to organise a lobby group, or make politicaldonations.

5 Damania (2001) ignores complications that arisefrom the strategic interactions between firms and theanalysis is thus analogous to that of the politicaleconomy literature.

6 There is usually a restriction for this result to occur(which is often implicit in the second order conditions).It is the requirement that demand not be too convex.This condition ensures that the rent shifting benefitsfrom the foreign to the domestic industry are sufficientlylarge.

ECONOMIC RECORD58 MARCH

credibly commit to higher output levels. There istherefore a strategic incentive for each firm toover-invest in cost reducing technology. In con-trast, the results presented in this paper suggestthat if firm lobbying plays a significant role indetermining protection levels, the conclusions ofthe strategic trade policy models of Cournotcompetition may be reversed.The analysis is based on the following sequence

of events. In the first stage the firms simultaneouslychoose their production equipment, from a con-tinuum of available production technologies. Thesecond stage defines the political equilibrium, inwhich the tariff is determined in a lobbying game.In the final stage the firms choose output levels.The remainder of this paper is organised as

follows. Section II outlines the basic structure ofthe model, derives the political equilibrium anddescribes the manner in which investment influ-ences political contributions. Section III dealswith the problem of investment and outlinesthe circumstances under which lobbying dimin-ishes the incentive to invest in new technology.Section IV concludes the paper.

II The ModelThe aim of this paper is to examine the

lobbying incentives of established industries whichface competition from foreign producers. More-over, when protected by tariffs or quotas, theseindustries may be characterised by imperfectcompetition. We therefore assume that a domesticduopoly which is protected by a tariff, facescompetition from a foreign rival.7 The duopolymay be sustainable even with free trade due to�home bias� and other factors (Bloningen &Wilson 1999). For simplicity it is assumed thatthe industry produces a homogenous product.Let X¼ xi + xj be the output of the domestic

firms i and j. Let y be the output of the foreignfirm. Then total industry output is Q ¼ X + y.The inverse demand function is defined asP(Q)¼P(X + y); with P¢ < 0, P¢¢ < 0. The firmscompete using quantities as the strategic variable.8

The cost to each firm of producing output levelq (q¼ xi, xj, y) is given by the cost functionCi(w(si), q), where w(si) is the input price and si

defines the type of production technology used byeach firm. For given input prices and technology,production costs are increasing and convex inoutput (i.e. @Ci/@q > 0, @2Ci/@q2 > 0). SectionIII defines the properties of the productiontechnology in more detail. However, at this stagewe note that the technologies defined as si 2 [1,T],are distinguished by the fact that higher values ofsi correspond to technology with lower produc-tion costs (i.e. dw/dsi < 0). Hence, the technol-ogies with larger values of si may be regarded asmore cost-effective and efficient. These assump-tions imply that total and marginal productioncosts decline with more efficient technologies. Wefurther assume that Ci(w(si), q) satisfies all theusual properties of a cost function.9

The analysis is based on the following sequenceof events. In stage 1 the firms simultaneouslydetermine investment levels in equipment (i.e.choice of s). The next stage defines the politicalequilibrium where each firm determines its polit-ical contributions (Si) independently, while thegovernment sets the tariff (t) to maximise itspayoffs. In the final stage, the firms compete usingquantities as the strategic variable. As usual, themodel is solved by backward induction.In an attempt to influence the level of protec-

tion, each domestic firm offers political contribu-tions Si to the government. Thus, domestic firm i’sprofits are defined as:

Pi ¼ PðQÞxi � CðwðsiÞxiÞ � Si ði ¼ 1; 2; i 6¼ jÞð1aÞ

In keeping with the existing literature weassume that the foreign firm, being located over-seas, has no leverage on domestic policy issuesand therefore does not engage in lobbying (seeGrossman & Helpman 1995). The foreign firm’sprofits are given by:

Py ¼ PðQÞy � CðwðsyÞ; yÞ � ty ð1bÞ

where t is the tariff.

(i) Stage 3: Output CompetitionWe begin by solving the final stage of the game

in which output levels are determined. Takinginvestment levels, the tariff and contributions as

7 The results can readily be generalised to the case ofan n >2 firm oligopoly at both home and overseas.

8 The central conclusions are not affected by theassumption of Cournot competition. This issue isaddressed in Section IV.

9 That is, the cost function Ci(w(si), q) is homogenousof degree one in input price, convex in output andconcave in input price.

PROTECTIONIST LOBBYING AND INVESTMENT2003 59

given, equilibrium output levels are given by thesolutions to the first order conditions:

P þ @P

@xixi � @CðwðsiÞ; xiÞ

@xi¼ 0 ði; j ¼ 1; 2; i 6¼ jÞ

ð1cÞ

P þ @P

@yy � @CðwðsyÞ; yÞ

@yð1dÞ

Let xi¼ xj¼ xn denote the symmetric equilibriumsolution of the domestic firms.10

For future reference, the following well knowncomparative static properties of the equilibriumare briefly described. The proofs are relegated tothe Appendix.

dxn

dt> 0;

dy

dt< 0;

dQ

dt< 0 ð2aÞ

Equation (2a) reveals that with greater protection(higher tariffs) the domestic firms expand theiroutput levels, while the foreign firm’s outputcontracts. This occurs because a higher tariff shiftsdemand from the foreign to the domestic industryand results in an increase in domestic output levels.However, overall, domestic consumption of thegood is declining in the tariff (i.e. dQ

dt < 0).

dxi

dsi> 0;

dy

dsy> 0 ði; j ¼ 1; 2; i 6¼ jÞ (2b)

Equation (2b) suggests that adoption of a moreefficient technology by a firm, lowers its produc-tion costs and thus induces an expansion in itsown output level.

dxi

dsj< 0;

dy

dsj< 0;

dxi

dsy< 0 ði; j ¼ 1; 2; i 6¼ jÞ

ð2cÞ

Finally, equation (2c) summarises the wellknown result that investment in an oligopoly hasstrategic effects.11Ceteris paribus, the adoption of amore efficient technology by one firm, lowers itsmarginal costs, and allows the firm to crediblycommit to more aggressive (i.e. higher) outputresponses. As quantities are strategic substitutes,the commitment to a higher output by one firminduces its rival to lower its production levels inresponse.

(ii) Stage 2: The Political EquilibriumHaving defined equilibrium output levels, we

now consider the manner in which politicalcontributions are determined. For given technol-ogy (s), each domestic firm will choose itspolitical contributions (Si) to maximise profits,taking account of the impact of its choices onoutput market competition in stage 3. Each firmsolves:

Maxsi

Pi ¼ PðQÞxi � CðwðsiÞxiÞ � Si

ði ¼ 1; 2; i 6¼ jÞ ð3aÞ

where xi is defined as the solution to (1c) and (1d).The associated first-order condition is12

@Pi

@t

@t

@Si� 1 ¼ 0 ði ¼ 1; 2; i 6¼ jÞ ð3bÞ

Note that since @Pi

@t ¼ xi @P@Q ð@y

@t þ @xj

@t Þ > 0, then

(3b) defines an interior solution only if @t@Si > 0.

This implies that lobbying will occur only ifhigher political contributions (Si) induces thegovernment to set a higher tariff (t).Following Grossman and Helpman (1994), the

government is assumed to maximise a weightedsum of the political contributions it receives andaggregate social welfare. Define the sum ofconsumer surplus and tariff revenues as:

WC �ZQ

0

PðQÞdQ � PðQÞQ þ ty ð4aÞ

Aggregate social welfare gross-of-contributionsis given by the sum of expression (4a) and gross-of-contributions profits (denoted ~PP)

W � WC þ ~PP; ð4bÞ

where: ~PP¼ ~PPi þ ~PPj; ~PPi ¼ Pi þ Si; ði ¼ 1; 2; i 6¼ jÞ.The government’s objective function is given by

a weighted sum of political contributions andsocial welfare (Grossman & Helpman 1994):

G ¼ S þ aW ð4cÞ

where: a is the weight given to aggregatesocial welfare relative to political contributionsS¼Si + Sj.

10 The arguments of functions are ignored fornotational brevity when not essential.

11 See, for example, Tirole (1990), p. 323.

12 The second order conditions are specified in theAppendix.

ECONOMIC RECORD60 MARCH

This specification of government utility is widelyused in the political support literature. It is basedon the assumption that political donations arevalued by governments because of their many uses,such as funding election campaigns, retiring debtfrom previous elections and deterring rivals.However, social welfare is included to capturethe notion that the prospect of retaining powermay be linked to average welfare in the economy(Grossman & Helpman 1994). By this interpreta-tion, the weight (a) given to social welfare isdetermined by factors such as the level of politicalcompetition, political stability, and the policies ofrivals. There are several alternative models of self-interested government behaviour. The most signi-ficant of these are models of political competition(see Persson & Tabellini 2001 for a survey).However, the formulation of government utilityused in (4c) subsumes many of the importantfeatures of lobbying and political competition thatare captured in the alternative models of govern-ment behaviour and is therefore used in this paper(Grossman & Helpman 1994; Persson & Tabellini2001).For future reference let the welfare maximising

level of domestic output be defined as:

t ¼ Argmax W ð4dÞ

where t*¼ P0ðXð@xi=@tÞþyðð@X=@tÞþð@y=@tÞÞÞ�y@y=@t � 0 is the

tariff required to achieve the welfare maximisingoutput level Q* > 0. Define W* as the resulting(maximal) level of welfare at Q*.13 The secondorder condition for a maximum requires that@2W/@t2 < 0, which is assumed to hold.A subgame perfect Nash equilibrium for this

game is a contribution (Si) for each firm and atariff (tL), such that: (i) the contribution is feasibleand maximises each firm’s payoffs, taking theother firm’s contributions as given; (ii) the policytL maximises the government’s welfare, G, takingthe contributions as given.From Lemma 2 of Bernheim and Whinston

(1986) the following necessary conditions yield asubgame perfect Nash equilibrium {S, tL}:

tL 2 Argmax G ¼ S þ aW; ðSIÞtL 2 Argmax PðtÞ þ G ðSIIÞ

Condition (SI) asserts that the equilibriumtariff tL must maximise the government’s payoff,given the contribution offered by the industrylobby group. Condition (SII) requires that tL

must also maximise the joint payoff of the firmsand the government. If this condition is notsatisfied, the lobby group will have an incentiveto alter its strategy to induce the government tochange the tariff, and capture more of thesurplus. Maximising (SI) and (SII), and perform-ing the appropriate substitutions, yields thepolitical equilibrium contribution of the lobbygroup which satisfies:

@PL

@t¼ @SL

@t: ð5aÞ

SL¼ contributions at equilibrium tariff tL, PL¼profits at equilibrium tariff tL.Equation (5a) implies that in equilibrium, the

change in the industry’s political contribution(i.e. @SL

@t ), equals the effect of the tariff onindustry profits (i.e. @PL

@t ). Thus, as noted byGrossman and Helpman (1994), the politicalcontribution schedule is �locally truthful�. As inBernheim and Whinston (1986), this conceptcan be extended to a contribution schedule thatis globally truthful. This yields a functionwhich accurately mirrors the preferences of thelobbyist’s at all policy points. The Appendixprovides a discussion of the existence of thisequilibrium.Having determined the slope of the contribu-

tion schedule, it is necessary to derive anexpression for the level of contributions in apolitical equilibrium. Grossman and Helpmandemonstrate that with one lobby group, theequilibrium contribution to the government isdefined by the difference in social welfare,when the tariff is set at the welfare maximis-ing rate t* and at the political equilibrium rate tL.Specifically:

SL ¼ aðW � WLÞ ð5bÞ

where: W* is the level of social welfare whicheventuates when the tariff is set at the welfaremaximising level t* and WL is the level of socialwelfare when the tariff is set at the politicalequilibrium level tL.

13 We implicitly rule out an import subsidy andassume that some level of domestic production isoptimal and can be achieved with a non-negativetariff. This assumption is made to simplify the proofs,but is not essential to the central results.

PROTECTIONIST LOBBYING AND INVESTMENT2003 61

Observe that a(W* ) WL) defines the loss ofutility to the government when the tariff deviatesfrom the welfare maximising level. Equation (5b)reveals that political contributions perfectly com-pensate the government for the welfare lossassociated with participation of the lobby groupin the political process. The welfare loss is weigh-ted by the factor a in order to adjust forits importance in the government’s objectivefunction.Clearly, the equilibrium defined in equation (5b)

can be sustained only if the individually rationalcontributions of firms as given in (3b) are sufficientto compensate the government for its utility lossfrom raising the tariff (i.e. 5b). If this conditiondoes not hold the government has no incentive toraise the tariff above the welfare maximising levelso that there can be no effective lobbying. Lemma1 explores this issue in greater detail and outlinesan important property of the equilibrium that hasbeen overlooked in the literature.

Lemma 1 The individually rational contributionsof the firms (as defined in (3b)), equals the amountthat is necessary to induce the government to raisethe tariff above the welfare maximising level (asdefined in (5b)).

Proof: From equation (3b) the profit maximisingpolitical contributions offered by each domesticfirm to the government satisfy the first-ordercondition

@Pi

@t

@t

@Si� 1 ¼ 0 ði ¼ 1; 2; i 6¼ jÞ ðIÞ

Note that since @Pi

@t ¼ xi @P@Q ð@y

@t þ @xj

@t Þ > 0, then (I)

defines an interior solution only if @t@Si > 0. The tax

schedule t(S) is thus monotonic and hence itsinverse exists.14 By the property of inverse func-tions (I) can therefore be rearranged as:

@Pi

@t¼ @Si

@t: ði ¼ 1; 2; i 6¼ jÞ ðIIÞ

Observe that (II) is the local truthfulnesscondition, necessary for a subgame perfect Nashequilibrium in equation (5a).

As both domestic firms are symmetric, thenaggregating equation (II):

@P@t

� 2@Pi

@t¼ @S

@t

� 2@Si

@t: ðIIIÞ

The first-order-condition corresponding to con-dition (SI) of the political equilibrium requiresthat:

@P@t

¼ �a@W

@t: ðIVÞ

Substitute (IV) in (III), and integrate:

SL ¼ 2

ZtLt

@Pi

@tdt ¼ �a

ZtLt

@W

@tdt

¼ �aðWL � WÞ ¼ aðW � WLÞ: ðVÞ

where WL¼welfare at political equilibrium tarifftL and W*¼welfare at the welfare maximisingequilibrium tariff t*.Expression (V) defines the equilibrium level of

lobby group contribution payments. It is equal tothe equilibrium contribution amount defined in(5b), which is necessary to compensate thegovernment for raising the tariff above the welfaremaximising level. Q.E.D

Lemma 1 reveals that the individually rational(Nash) contributions which maximise a firm’sprofits (i.e. (3b)), are equal to the contributionsnecessary to support a subgame perfect equilib-rium of the political game with higher tarifflevels (i.e. (5b)). Observe that, as the benefitsfrom paying political contributions accrue to theentire (domestic) industry rather than the indi-vidual firm, the problem is analogous to that ofthe private provision of a public good. Henceeach firm has an incentive to �undersupply�lobbying contributions, relative to the amountthat would be paid in a fully cooperativeequilibrium with no free-riding. Lemma 1 showsthat the lobbying equilibrium in this modeldoes not require contributions (i.e. cooperation)between lobbyists beyond that which obtainsfrom independent firm optimisation. Hence, theindividually rational Nash contributions of eachfirm are sufficient to compensate the govern-ment for its utility loss from raising the tariffabove the welfare maximising level. Accordingly,

14 This assumption is implicit in the Grossman–Helpman model which assumes that @S/@t > 0 for allfeasible t > t*.

ECONOMIC RECORD62 MARCH

lobbying is not undermined by collective actionproblems that require levels of cooperationbeyond the individually rational level.15

Recall that technology levels are chosen in Stage1 and are therefore taken as given (in Stage 2)when political contributions are determined. How-ever, in choosing its technology levels each firmwill take account of the impact of its investmentdecisions on political contributions. It is thereforeinstructive to examine the consequences of varyingtechnology levels (in Stage 1) on the politicalequilibrium in (Stage 2).

Lemma 2 The adoption of more efficient tech-nology si in Stage 1, increases the aggregatepolitical contributions paid to the government.

i.e.dS

dsi¼ dSi

dsiþ dSj

dsi> 0 ði ¼ 1; 2; i 6¼ jÞ

� �:

Proof: See Appendix.

Intuitively, this result follows from the localtruthfulness property (equation (5a)), which statesthat political contributions mirror the marginalprofitability of any policy change. Ceteris paribus,the adoption of a more efficient technology byfirm i, lowers its production costs and thus raisesprofits. Since a given tariff level now yields higherprofits, by local truthfulness, political contribu-tions rise.16 Hence investment in technology

makes lobbying for protection more expensivefor the firms.

Lemma 3 The adoption of more efficient tech-nology by domestic firms in Stage 1 leads to alower tariff being set in the political equilibrium, ifthe demand function is not too convex.�

i.e.dtL

dsi< 0 if R > �1=Qt;

where R ¼ P00Q=P0 ði; j ¼ 1; 2; i 6¼ jÞ�:

Proof: See Appendix.Intuitively, this result may be explained as

follows. From equation (5a) we know thatpolitical donations are truthful, in the sensethat they reflect variations in payoffs whichresult from a change in the tariff. Suppose thata firm chooses not to invest in a more efficienttechnology. Ceteris paribus, production costswill be relatively higher, so that a given levelof protection yields lower profits. By Lemma 2,political contributions will decline. A govern-ment that values political contributions, has anincentive to adopt policies which raise profitsand political donations. To maintain contribu-tion levels, the government therefore raises (ordoes not lower) the tariff level. The requirementthat demand not be too convex is a regularitycondition that is widely used in much of thestrategic trade literature. It ensures that highertariffs improve domestic payoffs. This require-ment is always satisfied for a linear demandcurve. Lemma 3 therefore reveals that when thiscondition holds, then in a lobbying equilibriumthe adoption of a more efficient technology willlead to lower levels of protection. In keepingwith much of the literature, it is assumed thatthis convexity requirement holds.This finding has important implications for

firms� investment strategy. If firms can crediblycommit to higher production costs in earlierstages of the game, they can lower the politicalcontributions that will be paid in the ensuingpolitical equilibrium. The next Section deals withthe circumstances in which technology can beused as a credible commitment device.

III Technology ChoiceThis Section investigates the manner in which

political lobbying influences the firm’s choice of

15 Note that the non-cooperative profit maximising(Nash) contributions of each firm are non-zero whenthere is an interior solution to (3a). A corner solutionwith zero contributions occurs when the costs oflobbying always exceed the benefits of lobbying. In thiscase both the fully cooperative and the non-cooperativeequilibrium contributions are zero. Another propertythat is worth stating is that the political equilibrium isidentical whether the lobbyists are assumed to be�groups� representing an entire industry or simply thefirms acting individually. Intuitively, this follows directlyfrom the local truthfulness condition (5a). Formalproofs are available from the author upon request.

16 As lower costs give firm i a greater share of themarket, firm j’s profits and (by local truthfulness)political contributions decline. However, the increasein i�s contributions outweighs the decline in j�scontributions, so that aggregate industry politicalcontributions rise. Formally, this reflects the fact thatthe slope of the reaction functions are less than unity inabsolute value. Hence the decline in j’s contributions donot offset the increase in i’s contributions.

PROTECTIONIST LOBBYING AND INVESTMENT2003 63

production technology. We begin by defining theproperties of the available technologies.Let si 2 [1,T]R +

(i¼ 1, 2; i „ j) be the con-tinuum of existing production technologies. Thetechnologies in si are distinguished by theirassociated production costs. Specifically, thereexists a one-to-one mapping from the set oftechnologies (si) to the costs associated with eachtechnology (w(si)). It is assumed that @wðsiÞ

@si < 0;@2wðsiÞ@si2 < 0. Thus, higher values of si correspond

to equipment which embodies lower productioncosts. The cost of purchasing equipment asso-ciated with a given technology of type si 2 [1,T] isgiven by K(si). It is assumed that K(si) is a sunkcost and that @KðsiÞ

@si > 0; @2KðsiÞ@si2 > 0. This implies

that the efficient technologies, with correspond-ingly lower production costs, are more expensiveto purchase.In Stage 1 each firm will choose a type

of technology (si) to maximise profits, takingaccount of is technology choice on all the otherdecision variables (i.e. output, political contribu-tions and the tariff). Thus:

Maxsi

P̂Pi ¼ PðQÞxn � Cðwðsi; xnÞÞ � Si � KðsiÞ

ð6aÞ

The first order condition is:17

@KðsiÞ@si

¼ P0 @xj

@siþ @y

@si

� �� @C

@w

@w

@si� dSi

dsið6bÞ

Equation (6b) reveals that the type ofequipment (si) adopted depends on a variety offactors, which encompass political, strategic andcost considerations. Each of these is discussedbriefly below. The firm acquires the type ofequipment at which the marginal cost ofpurchasing a more efficient technology (i.e.� @KðsiÞ

@si ), is set equal to the net marginal benefitsof the improved technology. These include themarginal benefits in the form of cost savingsfrom this technology (i.e. � @C

@w@w@si). In addition,

from Lemma 2 we know that adoption ofa more efficient technology raises political

contributions by the local truthfulness property.Thus, the need to lobby more intensively partlydiminishes the benefits of acquiring a more cost-effective technology and lowers the level ofinvestment in technology. These effects arecaptured in the term � dSi

dsi . Finally, as notedearlier, investment in cost saving technologieshas strategic output effects (equation (2c)).Adopting a more efficient technology allows afirm to credibly commit to more aggressive (i.e.higher) output responses. As quantities arestrategic substitutes, the commitment to a higheroutput induces a rival to lower its productionlevels, thereby raising the expected profits of thefirm that invests. Oligopolistic output competi-tion therefore induces firms to increase invest-ment levels. The strategic output effects aresummarised by the terms P0ð@xj

@si þ @y@siÞ.

Observe that in the absence of lobbying, firmswould simply equate the marginal cost ofacquiring a more efficient technology to themarginal benefits in the form of cost savings andstrategic effects from the equipment. It is there-fore important to investigate whether lobbyingresults in (higher) lower investment in technol-ogy. This issue is dealt with in the followingProposition.Define the choice of technology under lobbying

as:

siL 2 Argmax P̂Pi ¼ PðQLÞxn

L � CLðwðsiLÞ; xn

L� Si

L � KðsiLÞ

where: QL ¼ QðsiL; t

LÞ; tL ¼ tariff in the lobbyingequilibrium, xn

L is the corresponding output offirm i¼ 1, 2, subscript L on variables denotedterms in the lobbying equilibrium.Define the choice of technology in the absence

of lobbying as:

siu 2 Argmax Pi

u ¼ PðQuÞxnu

� Ciuðwðsi

uÞ; xnuÞ � Kðsi

uÞ

where: Qu ¼ Qðsiu; t

Þ is industry output whenthe tariff is at the welfare maximising level t* andthere is no lobbying and xn

u is the correspondingfirm output level, subscript u on variables denotesterms in the absence of lobbying.

Proposition 1 If the production costs associatedwith less efficient technologies are sufficientlyhigh, then lobbying lowers the level of investmentin technology.

17 There are a number of additional terms in the firstorder condition which are zero. Observe that: @Pi

@xi@xi

@si ¼ 0since by (1c) @Pi

@xi ¼ 0. In addition, ð@Pi

@t � @Si

@t Þ @t@si ¼ 0, since

by the local truthfulness condition in (5a) @Pi

@t ¼ @Si

@t .Using these results, yields the first order condition in(6b). The first order condition for the foreign firmis @KðsyÞ

@sy ¼ P0ð@xj

@sy þ @xi

@syÞ � @Cy

@w@w@sy.

ECONOMIC RECORD64 MARCH

�i.e. si

L < siu if

@CL

@wL

@wL

@siL

� @Cu

@wu

@wu

@siu

� �

> P0xiL

@xjL

@siL

þ @yL

@siL

!� P0xi

u

@xju

@siu

þ @yu

@siu

� ��

Proof: The first order condition when there is nolobbying is given by:

@KðsiuÞ

@siu

¼ P0xiu

@xju

@siu

þ @yu

@siu

� �� @Cu

@wu

@wu

@siu

(I)

When firms lobby from equation (6b) theassociated first order condition is:

@KðsiLÞ

@siL

¼ P0xiL

@x jL

@siL

þ @yL

@siL

!� @CL

@wL

@wL

@siL

� dSi

dsiL

(II)

Suppose that siu > si

L, then@Kðsi

uÞ@si

u>

@KðsiLÞ

@siL

(since

by assumption @KiðsÞ@si > 0; @2KiðsÞ

@s2 > 0). From (I)and (II) this implies that the right hand side of (II)must be less than that of (I):

P0xiL

@x jL

@siL

þ @yL

@siL

!� @CL

@wL

@wL

@siL

� dSi

dsiL

< P0xiu

@xju

@siu

þ @yu

@siu

� �� @Cu

@wu

@wu

@siu

ðIIIÞ

Recall that by Lemma 2 � dSiL

dsiL

> 0. It thenfollows that the RHS of (I) always exceeds that of(II) if:

P0xiL

@x jL

@siL

þ @yL

@siL

!� @CL

@wL

@wL

@siL

< P0xiu

@xju

@siu

þ @yu

@siu

� �� @Cu

@wu

@wu

@siu

(IV)

Which holds if

@CL

@wL

@wL

@siL

� @Cu

@wu

@wu

@siu

> P0xiL

@x jL

@siL

þ @yL

@siL

!� P0xi

u

@xju

@siu

þ @yu

@siu

� �

Q.E.D

Proposition 1 formalises the condition thatunderinvestment in technology acts as a crediblecommitment device, only if less efficient tech-nologies are associated with sufficiently high

production costs. Intuitively, when the costincrease from rejecting a more efficient technologyis sufficiently large, underinvestment provides acredible signal to the government that a reductionin tariffs will result in substantially lower profits.As political contributions are linked to profits, adecline in profits leads to a fall in politicaldonations. A government that values politicalcontributions is therefore induced to adopt a morefavourable policy towards firms. Thus there willbe higher levels of tariff protection associated withlower levels technological investment.Stated differently, when the cost saving from

adopting a more efficient technology is sufficientlyhigh, underinvestment in the first stage of thegame provides a credible signal to the governmentthat lower tariffs will result in lower politicalcontributions. Underinvestment therefore tilts thepolitical game in the domestic industry’s favour.18

Finally, we note that the underinvestment equi-librium is based on Nash conjectures by each firmand therefore does not involve any cooperationbeyond the individually rational levels.

IV Conclusions and ImplicationsThis paper has examined the interaction

between investment, lobbying and protectionistpolicy decisions. The central message is that whengovernments are receptive to special interestgroup pressures, political considerations mayprovide an incentive for firms to reject cost savinginvestments. If the costs associated with lessefficient technologies are sufficiently high, under-investment in technology provides a crediblesignal to the government that profits and politicaldonations will decline if tariffs are lowered. Agovernment that values political contributions istherefore induced to adopt a more favourablepolicy towards firms. Hence, industries with morecostly technologies are better placed to securepolicy concessions.The main findings of this paper conflict with the

conclusions of the strategic trade policy modelswhich predict that, under Cournot competition,protection induces greater investment in costsaving technology. Hence, the validity of theresults presented here must rest on the empiricalevidence. However, the mechanisms identified inthis paper are new, hence econometric support for

18 This is an example of Fudenberg and Tirole’s (1984)�puppy dog� strategy.

PROTECTIONIST LOBBYING AND INVESTMENT2003 65

the conclusions is hard to find. However, there issome indirect empirical evidence which is consis-tent with the predictions of the model.A number of studies have attempted to test the

infant industry argument. At its simplest level theinfant industry hypothesis asserts that newlyformed industries may require time to establishand become competitive. There is therefore a needfor temporary protection to allow the industry tomature, so that costs can fall to the level ofinternational competitors. A number of empiricalstudies have tested this hypothesis. It has beenfound that infant industries continue to be pro-tected many decades beyond the anticipatedperiod of protection. More importantly, increasedprotection has been associated with higher pro-duction costs (Baldwin (1992); Krueger & Tuncer(1982; 1988), Lucas (1984)). These findings appearto be consistent with a key prediction of themodel: underinvestment in cost saving initiativescan be credibly used to sustain high levels ofprotection.There is further support for the results from a

number of industry based studies. In an econo-metric study of lobbying in the agricultural sector,Eliste and Fredriksson (1999) find that users ofolder and more damaging technology obtaingreater net policy support from the government.Similarly, studies of the metal industry in Korea(Truett & Truett 1997), electronics in Brazil(Luzio & Greenstein 1995), engineering in Indo-nesia (Braadbaart 1996), vehicle manufacture inSouth and South-East Asia (Okamato 2000) notethat these industries are heavily protected. How-ever, they have higher production costs than theirinternational rivals and produce goods that lagbehind the technological frontier. These conclu-sions once again appear to be consistent with thecentral conclusion of this paper. It is perhapsuseful to note that the results of this papersimply indicate that less efficient industries maybe more successful in securing concessions. Theanalysis does not suggest that tariffs will neverdecline.19

There are a number of other issues that havenot been considered so far. Most important ofthese is the assumed form of competition in theoutput market. It is well known that results basedon strategic interactions are highly sensitive to theassumed form of competition in oligopolisticmarkets. It is therefore important to determinewhether the main result summarised in Proposi-tion 1 is reversed under the assumption that firmscompete in prices in the final stage of the game.Consider the problem when the duopolists com-pete in prices.20 Recall from Proposition 1 andLemma 2, that underinvestment raises costs andlowers political contributions. The credible threatof lower contributions induces the government toprovide greater protection. This link betweenlobbying and investment is unaffected by pricecompetition, so long as underinvestment raisesproduction costs sufficiently. This is because theunderinvestment equilibrium arises as a conse-quence of the interaction between the governmentand each firm, rather than as a result of thestrategic interaction between firms. While thequalitative effects are unaffected by the form ofproduct market competition, the quantitativeimpacts may differ. However, we are unable tosay anything about the relative sizes at this level ofgenerality.Another important issue is the assumed

sequence of events. The credible commitmenteffects stem from the assumption that firmsdetermine their investment first and the govern-ment chooses its policy taking the investmentdecision as given. This seems reasonable if it issupposed that investment in technology is a long-run decision variable, while the details of govern-ment policies are influenced by lobby grouppressures and more immediate (short-term) polit-ical concerns.21 If, however, firms delay theirinvestment decisions so that the sequence of eventsis reversed, then investment can no longer have acredible commitment effect. Clearly, delayinginvestments would be the rational strategy forfirms if they expect a regime change that brings in a

19 For instance a referee provided a counter-exampleof trade policies in the Thatcher era, where trade barriersin the older industries were systematically lowered. Ourmodel does not suggest that such policy initiatives areimpossible to introduce, but that in the high costindustries lobbying would be more effective and hencereforms are harder to introduce if the government valuespolitical donations from these industries.

20 To avoid the discontinuity problems caused byBertrand competition let the goods produced by theduopolists be imperfect substitutes, sold at prices Pi andPj (i¼ 1, 2; i „ j).

21 This is one of the central assumptions of theGrossman–Helpman model. It defines the short-runpolitical equilibrium, taking longer term considerationsas given.

ECONOMIC RECORD66 MARCH

government which places no weight on politicalcontributions from the industry lobby group.Finally, it worth noting that the results in this

paper are consistent with those of Wright (1995),who explores the time consistency of future tariffpolicies. Wright demonstrates that a policy oftariff removal can be rendered time inconsistent ifa firm increases its costs. However, Wright doesnot explicitly model lobbying or the politicalprocess, and abstracts from investment and cre-dibility issues. Thus, even though the objectiveand framework of Wright’s analysis differs fromthis paper, the results lend further support to thebasic conclusions of this model.

APPENDIX

Proofs of Equations (2a)–(2c): Totally differen-tiating (1a) and (1b), gives the comparative staticssystem of equations:

Piii Pi

ij Piiy

Pjji Pj

jj Pjjy

Pyyi Py

yj Pyyy

2664

3775

dxi

dxj

dy

264

375

¼ �0

0

Pyyt

264

375dt �

Piisi

0

0

264

375dsi �

0

Pjjsj

0

264

375dsj ðA1Þ

where: subscripts denote partial derivatives, Piii ¼

@2P@xi2 x

i þ 2 @P@xi � @2Ci

@xi2 < 0; Piij ¼ @2P

@xi@xj xi þ 2 @P

@xj < 0;

Piit¼ 0; Pi

isi ¼ � @2Ci

@xi@wi@w@s > 0; Py

yt ¼ �1.Note thati and j are symmetric.By the SOCs it is assumed that the determinant

denoted D < 0 and that the following usualassumptions are maintained:

jPiiij> jPi

ijj; jPjjjj> jPj

jij; jPyyyj> jPj

jij; jPyyyj> jPy

yij¼ jPy

yjj ðA2Þ

Solving (A1) (using symmetry between i and j):

dxi

dt¼

PjjyP

yytðPj

jj � PiijÞ

D> 0 ðA3Þ

dy

dt¼

�Pyyt ðPj

jjÞ2 � ðPi

ijÞ2

� D

< 0 ðA4Þ

dxi

dsi¼ �

PiisiðPj

jjPyyy � Pj

jyPyyjÞ

D> 0 ðA5Þ

dxj

dsi¼ �

PiisiðPj

jiPyyy � Pj

jyPyyiÞ

D< 0 ðA6Þ

dy

dsi¼ �

PiisiðPj

jiPyyj � Pj

jjPyyiÞ

D< 0 ðA7Þ

Lemma 2 Totally differentiating (3b), gives thecomparative static system of equations:

PiSiSi Pi

SiSj

PjSiSj Pj

SjSj

� �dSi

dSj

� �¼ � Pi

sisi

0

� �dsi � 0

Pjsjsj

� �dsj

ðA8Þ

Let D > 0 be the determinant of the systemwith jPi

SiSi j > jPiSiSj j; jPi

SiSi j > jPjSiSj j; jPi

SiSi j < 0;

PiSjSj < 0; Pj

Sjsi ¼ @y@si P

0ð@y@t þ @xj

@t Þ > 0; PiSisj ¼ @xi

@sj P0ð@y

@t þ @xj

@t Þ < 0.Solving:

dSi

dsi¼

PiSisiP

jSjSj þ Pi

SisiPiSjSj

D> 0 ðA9Þ

dSj

dsi¼

�PjSjsiPi

SiSi þ PiSisiP

jSjSi

Dl0 ðA10Þ

dS

dsi� dSi

dsiþ dSj

dsi

¼ðPi

SjSi � PiSiSiÞðPj

Sjsi þ PiSisiÞ

D> 0 ðA11Þ

Lemma 3 By condition (SI) in the text, the tariffis determined by the government to maximiseits welfare. The tariff thus satisfies the first ordercondition:

Gt¼P0XðxitþytÞð1þaÞþaðyþtyt�P0QtQÞ¼0

ðA12Þ

where subscripts denote partial derivatives.Totally differentiate (A12) and rearrange:

dt

dsi¼ �Gtsi

GttðA13Þ

By the SOC it is assumed that Gtt < 0. Thussign of dt/dsi¼ sign of Gtsi.Further differentiating (A12):

Gtsi ¼ P0Xsiðxi

t þ ytÞð1þ aÞþ aðysi � P00QsiQtQ � P0QsiQtÞ: ðA14Þ

PROTECTIONIST LOBBYING AND INVESTMENT2003 67

Rearranging (A14) it can be verified thatGtsi < 0 if RQt>)1 where R¼P¢¢Q/P¢ which isthe usual measure of convexity of the demandfunction that is used in the literature.Thus, if R > )1/Qt then

dtdsi ¼ �Gtsi

Gtt< 0.

Existence of a Political EquilibriumShapiro (1986) discusses the necessary condi-

tions for a Cournot equilibrium in the outputmarket to exist in a game such as that outlined inSection II. It is shown that if the first and secondorder conditions are satisfied a Cournot equilib-rium will exist and be stable.Showing existence of the political equilibrium

is, however, somewhat more complicated andhence a brief discussion of this issue is providedhere. To establish the existence of the politicalequilibrium it is necessary to show that the gameplayed between the firms and the governmentsatisfy Kakutani’s fixed point theorem. To do sowe introduce some further notation.Let Si 2 ri be the strategy space of firm i (i¼ 1,

2; i „ j) where Si denotes contributions of firm i.Let t 2 rG be the strategy space of the govern-

ment where t is the tariff.Define r¼ ri · rj · rG, as the Cartesian prod-

uct. It defines the strategy space of the gameplayed between the firms and the government.Define Pi(r, Q) as firm i�s payoffs and G(r, Q)

as the government’s payoffs.

Assumption 1 ri is compact. Note that it is closedas 0 £ Si £ Pi(r, Q)

Assumption 2 rG is compact. It is closed ast � t � t̂t, where t* is the welfare maximisingtariff and t̂t is defined by the condition thatPyðt̂tÞ¼ 0 (that is, the height of the tariff issuch that the foreign firm earns no profits in thedomestic market).

Assumption 3 Pi(r, Q) is jointly concave withrespect to Si and Q and G(r, Q) is jointly concavewith respect to t and Q.Result: By theorem 2.4 of Friedman (1980) the

game between the government and the firmssatisfies the conditions of Kakutani’s fixed pointtheorem and has one fixed point. An equilibriumpoint therefore exists.This result may be established by showing that

the following conditions hold

Condition 1 – the domain of the best replyfunctions are compact and convex

Condition 2 – the image sets of the best replyfunctions are contained in rCondition 3 – the image sets of the best replyfunctions are convexCondition 4 – the best reply functions are uppersemicontinuous.

Note that Condition 1 is satisfied by Assump-tions 1 and 2 and the fact that the Cartesianproduct of convex sets is convex.To see Condition 2 define the best reply

mapping of (say) the firms as:

riðrÞ ¼ S0 2 r0jP0ðrnS0 � P0ðrnS0iÞ�for all S0i 2 riÞ

�That is the strategy S*i is a best reply to otherstrategy combinations if it maximises the payoffsof i given the strategies of other players. Such anS*i exists because it is a maximiser of a continuousconcave function over a compact set and byconstruction it is required to be in r.To establish Condition 3 suppose that

S1, S2 2 ri(r). Define 0 < k < 1 and Sk¼ kS1þ(1 ) k)S2. Concavity of Pi implies that Pi(Sk) ‡kPi(S1) + (1 ) k)Pi(S2). Strict inequality impliesthat either or both S1, S2 =2 ri(r), hence this yieldsa contradiction.22 It follows that equality holdsand Sk 2 ri(r). So ri(r) is convex.Finally, Condition 4 follows directly from

Lemma 2.5 of Friedman (1980) which proves thatthis property holds under Assumptions 1–3.

REFERENCES

Baldwin, R.-E. (1988). Trade Policy in a Changing WorldEconomy. University of Chicago Press, Chicago.

Baldwin, R.-E. (1992), �The Case of Infant-IndustryTariff Protection� in D. Lal (ed.), DevelopmentEconomics, Vol. 3, Ashgate, Brookfield.

Baldwin, R. (1993), �Asymmetric Lobbying Effects: WhyGovernments Pick Losers FTP�, HEISUN UNIGE,CH/pub/Baldwin/asymlobbwp5. Last accessed Janu-ary 2001.

Bernheim, B.D. and Whinston, M. (1986), �MenuActions, Resource Allocation and Economic Influ-ence�, Quarterly Journal of Economics 101, 1–31.

22 To see why note that in this case Pi(Sk)>Pi(S1), orPi(Sk)>Pi(S2), hence, S1, S2 cannot satisfy thedefinition of a best reply since strategy Sk gives higherpayoffs, for given strategies of all other players.

ECONOMIC RECORD68 MARCH

Bloningen, B.A. and Wilson, W.W. (1999), �ExplainingArmington: What Determines Substitutabilitybetween Home and Foreign Goods�, Canadian Journalof Economics 32, 1–21.

Bouet, A. (2001), �Research and Development, Volun-tary Export Restraints and Tariffs�, European Econo-mic Review 45, 323–56.

Braadbaart, O. (1996), �Indonesian Engineering: Con-sumption Trends and Localisation Efforts�, Bulletin ofthe Economic Studies 32, 75–104.

Damania (2001), �When the Weak Win: The Role ofInvestment in Environmental Lobbying�, Journal ofEnvironmental Economics and Management 42, 1–22.

Dixit, A. and Londregan, J. (1995), �RedistributivePolitics and Economic Efficiency�, American PoliticalScience Review 89, 856–95.

El-Agraa, A.M. (1987) Protection, Cooperation,Development and Integration. Macmillan, London.

Eliste, P. and Fredriksson P. (1999), �The PoliticalEconomy of Environmental Regulations, Govern-ment Assistance and Foreign Trade�, in P. Fredriksson(ed.), Trade Global Policy and the Environment, WorldBank Paper, No 402, Washington D.C.

Fredriksson, P. (1997), �The Political Economy ofPollution Taxes in a Small Open Economy�, Journalof Environmental Economics and Management 33,44–58.

Friedman, J.W. (1980), Game Theory with Applicationsto Economics, Oxford University Press, Oxford.

Fudenberg, D. and Tirole, J. (1984), �The Fat Cat Effect,The Puppy Dog Ploy and the Lean and HungryLook�, American Economic Review 74, 361–8.

Grossman, G.E. and Helpman, E. (1994), �Protection forSale�, American Economic Review 84, 833–50.

Grossman, G.E. and Helpman, E. (1995), �The Politicsof Free Trade�, American Economic Review 85,667–90.

Grossman, G.E. and Helpman, E. (1996), �Rent Dissi-pation, Free Riding and Trade Policy�, EuropeanEconomic Review 40, 795–803.

Krueger, A. and Tuncer, B. (1982), �An Empirical Test ifthe Infant Industry Argument�, American EconomicReview 72, 1142–52.

Krugman, P. (1984), �Import protection as Export\Protection�, in H. Kierlowszki (ed.), MonopolisticCompetition and International Trade, Clarendon Press,Oxford; 18–24.

Long, N.V. and Vousden, N. (1991), �ProtectionistResponses and Declining Industries�, Journal ofInternational Economics 30, 87–103.

Lucas, R.E. (1984), �An Empirical Test if the InfantIndustry Argument: a Comment�, American EconomicReview 74, 1110–1.

Luzio, E. and Greenstein, S. (1995), �Measuring thePerformance of a Protected Infant Industry�. Reviewof Economics and Statistics 77, 622–33.

Okamato, Y. (2000), �Productivity in the AutomotiveIndustry�, ASEAN Economic Bulletin 17, 60–73.

Persson, T. and Tabellini G. (2001), Political Economics:Explaining Economic Policy, MIT Press, Cambridge,MA.

Shapiro, C. (1986), �Theories of Oligopoly Behavior�, inR. Schmalensee and R. Willig (eds), The Handbook ofIndustrial Organization. North Holland, Amsterdam,330–410.

Tirole, J. (1990), The Theory of Industrial Organization,MIT Press, Cambridge, Massachusetes.

Truett, L. and Truett, D. (1997), �The Korean MetalsIndustry and Trade�, Journal of Asian Economics 8,333–47.

Wright, D. (1995), �Incentives, Protection and TimeConsistency�, Canadian Journal of Economics 28,929–38.

PROTECTIONIST LOBBYING AND INVESTMENT2003 69