Promoting Trade: The Importance of Market Protecting Institutions

Promoting Trade: The Importance of MarketProtecting Institutions

Mark Souva Florida State University

Dale L. Smith Florida State University

Shawn Rowan Florida State University

What set of domestic institutions most affect bilateral trade flows? Extant research emphasizes security institutionsor democratic political institutions. We agree that domestic institutions matter, but not the ones previouslyidentified. Rather, the key set of institutions influencing trade flows are those that protect property rights. Thesemarket-protecting institutions, such as the establishment of banking and insurance laws and common standards ofmeasurement, promote trade by lowering the transaction costs of commercial exchange. An empirical analysis ofbilateral trade flows employing a number of estimation methods and operationalizations of the key concept supportour market-protecting institutions argument.

In recent years political scientists have attemptedto enhance standard economic models of trade byasserting that political institutions have an im-

portant influence on the pattern of bilateral trade.Gowa (1989), Gowa and Mansfield (1993, 2004), andMansfield and Bronson (1997) emphasize the roleof security externalities and claim that alliance tiesinfluence trade flows. Alternatively, Morrow, Siverson,and Taberas (1998, 1999), Bliss and Russett (1998),and Lektzian and Souva (2001) argue that democ-racies are more likely to trade with other democra-cies. We agree that political institutions affect trade,but not precisely the way previous research has claimed.The primary influence of politics is not related toalliance ties or the political systems that two statesshare. Rather, it revolves around the cultivation ofmarket-protecting institutions.

We begin with the following conception of thedeterminants of trade flows: firms trade while govern-ments determine the broader policy environmentwithin which firms’ trading decisions are made. Thelevel of trade between two states is an aggregation ofthousands of decisions by firms in one country tochoose firms in another country to be their sources ormarkets for tradable goods. Economic theory tells usthat those decisions are based on price, as determinedby factor endowments and economies of scale, andtransaction costs. Governments are usually not directly

involved in firm-level decisions of who to export toor import from. They do, however, determine thebroader policy environment affecting firm decisions.Governments can make it easy or difficult for itsfirms to engage in international trade by affecting thetransaction costs, the ex ante and ex post costs ofnegotiating, implementing, and enforcing transactionsof commercial activity. Governments that protect pri-vate property, enforce contracts, have transparentadministrative procedures and low levels of graft andbribery, policies which we cluster together under thelabel of market-protecting institutions, enhance afirm’s ability to profit by decreasing transaction costs.In focusing on market-protecting institutions, we de-velop a different perspective on the impact of politicaland economic institutions on trade. Institutions mat-ter, but not the ones previously identified. In brief,firms aim to maximize profit, and governments affectbilateral trade flows by choosing policies that affectfirms’ ability to profit. The most important of thesepolicies, we contend, is the cultivation of market-protecting institutions.

The empirical analysis supports the central argu-ment of this paper: market institutions promote trademore than political institutions, such as democracyor alliances. Although this finding is novel in thepolitical determinants of trade research program, itis consistent with other research that finds that

The Journal of Politics, Vol. 70, No. 2, April 2008, Pp. 383–392 doi:10.1017/S0022381608080377

� 2008 Southern Political Science Association ISSN 0022-3816

383

market-protecting institutions matter more thandemocratic institutions in other economic areas,such as attracting foreign direct investment (Li andResnick 2003).

The Political Determinants of Trade

Research on the political determinants of trade can besummarized in terms of three important strands. Thefirst argues that the foreign policy behavior of statesinfluences commercial relations, with the generaltenor of relations, whether cooperative or conflictual,affecting firms’ decisions regarding trading partners.If conflict between two states either exists or isanticipated, firms will shift their suppliers and mar-kets to other states to avoid the disruption, whichheightened levels of conflict can produce. This argu-ment was pioneered by Polachek (1980) and Pollins(1989a, 1989b) and extended in recent work by Liand Sacko (2002). While this research emphasizes theeffect of foreign policy behavior on trade, our focusis on the institutional determinants of trade and soin our review we emphasize the second and thirdstrands within this research program.

The Security Institutions Argument

The security institutions understanding of bilateraltrade focuses on two causal mechanisms that flowfrom military alliances. Military alliances reducesecurity externalities and the risk of expropriation.Given that trade produces welfare, states are cautiousabout permitting trade with an adversary state, fearingthat the gains from trade may be used for militaryproduction (Gowa 1989; Gowa and Mansfield 1993).By the same logic, trading with allies is to be en-couraged. Trade results in economic gains and a statewill always prefer that those gains go to its alliesrather than its adversaries. Military alliances may alsoreduce the risk of opportunistic behavior, such as onestate expropriating the assets of firms from anotherstate, that is, alliances reduce the transaction costs ofexchange, thereby enhancing trade between allies(Gowa and Mansfield 2004; Mansfield and Bronson1997). Security alliances reduce opportunism because‘‘governments have less incentive to behave oppor-tunistically toward their allies’ firms than towardfirms of other states’’ (Mansfield and Bronson 1997,95). This argument has also been extended to accountfor intra-industry trade between major powers. Thevalue of this trade ‘‘can be large enough to attract the

attention of both home and destination governments.If the latter are allies, their stake in joint-welfaremaximization endows them with an interest in deter-ring efforts to renegotiate the surplus that bilateralmonopolies create. Adversaries, in contrast, may seekto expropriate this surplus’’ (Gowa and Mansfield2004, 776; emphasis added). In brief, military alli-ances reduce security externalities and function as thecentral institutional feature limiting the likelihoodof opportunistic behavior; as a result, trade betweenallies should be greater than trade between otherstates.

The Democratic Institutions Argument

In response to the security institutions argument,Morrow, Siverson, and Taberas (1998, 1999) andBliss and Russett (1998) extend the democratic peaceargument to trade, claiming that states which sharedemocratic political institutions—not alliances—willhave higher levels of bilateral trade. One important wayin which democratic institutions differ from non-democratic institutions is in the amount of account-ability and flexibility for their leaders, with leaders indemocracies being more accountable and having lessflexibility to change policies (Leeds 1999). ‘‘The effectof the increased accountability in democratic insti-tutions is a decrease in uncertainty and risk, whichleads to a decrease in transaction costs . . . .These fac-tors all contribute to promote trade between coun-tries with democratic institutions’’ (Lektzian andSouva 2001, 64). Where the security argument expectsalliances to reduce opportunistic behavior, the democ-racy argument says democratic institutions reduceopportunistic behavior. ‘‘Entrepreneurs are also likelyto be more confident in the continuity of businesspractices and the rule of law in another democracythan in an autocracy, where such capricious acts asexpropriation may threaten their interests’’ (Bliss andRussett 1998, 1129).

Complementing the influence of democratic in-stitutions on reducing transaction costs is its influ-ence on reducing aggregate trade barriers. Mansfield,Milner, and Rosendorff argue and find ‘‘a greatertendency for pairs of democratic countries to agreeupon lower trade barriers than pairs comprised of ademocracy and an autocracy’’ (2000, 305). Similarly,Bueno de Mesquita et al. (2003) contend that de-mocracies are more likely to provide public goodsthan nondemocracies and since trade is a publicgood, democratic states are likely to engage in moretrade. In brief, the democracy argument expectsmore trade between democracies because democratic

384 mark souva, dale l. smith, and shawn rowan

institutions decrease uncertainty and tariffs, therebyreducing both transaction costs and product costs.

The Market Institutions Theory

Central to all the political determinants of trade re-search is the concept of transaction costs. We argue,however, that neither of the institutionally basedresearch streams reviewed above specifies the mostimportant set of domestic institutions affecting trans-action costs: market-protecting institutions.

Before disentangling security and political insti-tutions from market institutions, it is important toclarify the role of politics in promoting trade. Thepolicies governments choose make it more or lessdifficult for trade to occur. Facing any governmentis a menu of choices. If a certain set of policiesare chosen, the ‘‘political’’ barriers to trade will benonexistent, and trade can occur in the ‘‘frictionless’’environment found in economics texts. However, ifanother set of policies are chosen, trade will bechoked off. Government chooses first, and then firmsmust make choices within the policy environmentthat has been determined by the government. Whenstudies find that alliances or democracy are signifi-cant determinants of trade flows, what they are reallyfinding is that dyads sharing these institutions aremore likely to choose the set of policies which lowertransaction costs and minimize barriers to trade.Security and political institutions, then, are not theexplanatory mechanism, but only a proxy for thoseeconomic policies and institutions that lower trans-action costs.

What are the domestic institutions that mostaffect transaction costs? The most important institu-tions affecting transaction costs are not security ordemocratic institutions, but domestic institutionsthat protect private property, establish banking andinsurance laws, and create common standards ofmeasurement. We label these market-protectinginstitutions. More generally, ‘‘Economic rules [insti-tutions] define property rights’’ (North 1990, 47).When economic institutions embrace the market,they enhance property rights and commerce byreducing information asymmetries and enforcementproblems. One of the ways by which institutionsdo this is by establishing a reliable and trustworthybanking system. A trustworthy banking system isespecially important because there ‘‘are markets inwhich economic actors make exchanges requiringsignificant and irreversible commitments in the present,

whether in the form of goods manufactured and shippedor fixed investments made, in the expectation ofpayment or a stream of returns in the future . . . . Thesemarkets are less likely to exist when institutionsfor the protection of property rights and contractenforcement are absent’’ (Clague et al. 1999, 186).In summary, market-protecting institutions enhanceproperty rights and contract enforcement, therebyreducing transaction costs and permitting trade toflourish.

This account offers a clear response to the centralmicrofoundations question at issue: what motivatesfirms to trade with businesses in some countries morethan other countries? Simply put, firms seek profit,and international trade happens because differenttypes of firms in different countries will have a com-parative advantage. The principle factors affectingcomparative advantage are economies of scale (newtrade theory) and differences in factor endowments(classical trade theory).1 But the ability of a firm toprofit is also influenced by the level of transactioncosts, which are lower when property rights are pro-tected by the institutions created by governments.

Market institutions have a stronger affect onfirms’ decisions to trade than political or securityinstitutions, and those arguments point to the causalinfluence of market institutions. The security argu-ment, for instance, suggests a firm is more likely totrade with firms in allied states in order to reduceconcerns about expropriation. That is possible, butwe believe expropriation is more likely a function of astate’s economic institutions than alliance ties. Thedemocracy argument contends that democratic in-stitutions are the key to reducing transaction costs.The democracy connection, however, is indirect.Firms care less about the political regime in powerthan about having their property rights protected andcontracts enforced. In general, states with democraticinstitutions will have market institutions, but thismay not always be the case, especially with newdemocracies. Similarly, some nondemocratic regimesmay also have market-oriented institutions, with theFour Tigers in Southeast Asia as the leading exam-ples.2 Theoretically, then, market-protecting institu-tions not only directly measures the concept at issue,it also encompasses democracy as a measure for

1See for example, Krugman (1980), Helpman and Krugman(1985), and Helpman (1999).

2If one defines democracy as having a polity index score of six orgreater, then South Korea becomes a democracy in 1988,Singapore is a democracy from 1959 to 1962 but not thereafter.Hong Kong and Taiwan are not part of our analysis as they arenot independent states according to the United Nations.

promoting trade: the importance of market protecting institutions 385

property rights and contract enforcement. This leadsto the central hypothesis for this research.

H1: Dyads with stronger market-protecting insti-tutions will have more trade than dyads that affordless protection of the market, ceteris paribus.

While traders are more likely to choose partnerswith high levels of market-protecting institutions,under some circumstances the type of governmentcan influence that choice. When market-protectinginstitutions are weak, trade is risky, and thereforemore costly. This increased risk holds regardless ofwhether the government is autocratic or democratic.However, when market-protecting institutions arestrong, the type of government should have a greatereffect. In this case, traders favor a market-protectingdemocratic government over a market-protectingautocratic government because in autocracies, rela-tive to democracies, it is much easier to change therules of the game (see e.g., Tsebelis 2002). An autoc-racy could decide to do away with those marketingprotecting institutions whereas this sort of majorpolicy shift would be much more difficult in democ-racies. In addition, since democratic dyads on averageshould have lower tariffs (Mansfield, Milner, andRosendorff 2000), two democratic states with strongmarket-protecting institutions are likely to have moretrade than two autocratic states with strong market-protecting institutions. In brief, the influence of demo-cratic political institutions is conditional on thestrength of market-protecting institutions. This leadsto our second hypothesis:

H2: Democratic institutions will have a positive im-pact on the level of dyadic trade only when market-protecting institutions within that dyad are strong.

Research Design: Variables and Data

Within both economics and political science, thegravity model has been the workhorse of quantitativetrade studies. The traditional gravity model specifiesbilateral trade as a function of two general factors, thesize of each state’s economy and a resistance term.The size of a state’s economy is a proxy for nationalincome, indicating that one can only purchase,import, what one can afford. It is measured as astate’s Gross Domestic Product (GDP) and is ex-pected to be positive; more money means moreimports. In the basic gravity model, the resistanceterm is operationalized as the distance between twostates. As distance increases, transportation costs, andtherefore the cost of the goods increase. Bergstrand

(1985, 1989) and Helpman and Krugman (1985)have provided microfoundations for the gravitymodel, and both include GDP per capita to take intoaccount intra-industry arguments from the new tradetheory. Essentially, GDP per capita is a proxy for astate’s capital-to-labor ratio, with a higher ratioindicating that a state is more capital-intensive. Animportant insight of the new trade theory is thatcapital-intensive states increase their trade with othercapital-intensive states, that is, they engage in intra-industry trade; as a result, we expect that dyads withhigher capital-to-labor ratios should have more trade,holding other factors constant.

In addition to solid theoretical foundations,another advantage of the gravity model is its flexi-bility. The flexibility comes from unpacking the re-sistance term, which can be thought of broadly astransaction costs. The contribution of political sci-entists to international trade has been in articulatinghow political factors affect transaction costs. Forexample, both the security and democracy argumentsfocus on opportunism. If opportunism is high, thenthe cost of implementing and enforcing a transactionincreases. As discussed earlier, we agree with theemphasis previous political studies have placed ontransaction costs, but believe they have missed themost important set of institutions affecting suchcosts, namely market-protecting institutions.

The dependent variable for this research is thenatural log of the level of dyadic trade in a given year.The trade data come from Rose (2004), who drew thedata from the International Monetary Fund’s Direc-tion of Trade statistics. The advantage of the Rosedata over the raw IMF data is that he deflates tradeby a base Consumer Price Index. Although these datacover all countries over the period 1948–99, data forthe market institutions variable are only availablefrom 1960 onwards.3

The primary theoretical variable of interest isMarket Protecting Institutions. When market institu-tions are relatively more developed, transaction costsrelated to property rights, contract enforcement, andstandards of measurement are lower so firms are freerto conduct commercial exchanges. In recent years,economists have developed a variety of measures ofmarket institutions. The vast majority of these meas-ures are based on expert opinion and have a limitedtemporal and cross-sectional domain. Economistsat the World Bank, however, have created a moreobjective measure of market institutions that also has

3The Rose data may be obtained at http://faculty.haas.berkeley.edu/arose/RecRes.htm#Trade

386 mark souva, dale l. smith, and shawn rowan

a larger temporal and spatial domain. This measure isbased on the amount of contract-intensive money(CIM) in circulation for a given state. As discussedearlier, the market values secure property rights andenforceable contracts. Correspondingly, CIM is ‘‘anobjective measure of the enforceability of contractsand the security of property rights’’ (Clague et al. 1999,186). The argument behind CIM is that if propertyrights are secure and contract enforcement is strong,then there is less reliance on currency. ‘‘Thus theextent to which societies can capture not only thegains from self-enforcing transactions but also thosepotential trades that are intensive in contract enforce-ment and property rights can be approximated by therelative use of currency in comparison with contract-intensive money’’ (188). In other words, societies thatrely more on currency have more extensive blackmarkets, which are a sign of less secure propertyrights and contract enforcement. Operationally, CIMis ‘‘the ratio of non-currency money to the totalmoney supply’’ (188).4 Data for this variable comefrom the IMF’s International Financial Statistics. Tocapture dyadic effects, we use the weak link approachand include the lower of the two states’ CIM scores.5

To evaluate our second hypothesis, that the im-pact of democratic institutions is conditional on thestrength of market institutions, we need a measure ofdemocratic institutions. We measure democracy usingthe Polity IV dataset (Marshall and Jaggers 2002). ThePolity project uses a number of institutional indicatorsto measure the level of both autocracy and democracy

in a state. From this one can obtain an overalldemocracy score by subtracting a state’s autocracyindex score from its democracy index score.6 Thevariable Democracy ranges from –10 to +10, and againwe use the weak link approach and include the lowerof the two states’ scores. As our hypothesis is condi-tional, we also include a variable interacting marketinstitutions and democratic institutions. This allows usto test if democratic institutions influence trade flowsmore when market institutions are robust than whenmarket institutions are weak.

Our transaction cost approach also expects mili-tarized conflict to influence trade flows. Violent con-flict between states increases the risk of doing business,which increases the cost of trade (Keshk, Pollins, andReuvany 2004; Li and Sacko 2002; Pollins 1989a,1989b).7 However, states have many disputes unlikelyto affect aggregate trade flows because they are bothmild and short in duration. Since we only expectserious conflict to affect trade, we measure violentdyadic conflict as the presence of a fatal militarizedinterstate dispute. The variable Fatal MID equals onewhen a militarized dispute results in fatalities andzero otherwise. Data on fatal MIDs comes from theCorrelates of War project.8

Next, we include a variable, Allies, to control forthe security institutions argument. This variable equals1 for any dyad with a defense pact, entente, or non-aggression agreement, 0 otherwise. Data come from theCorrelates of War project (Gibler and Sarkees 2004).

The three basic variables in the gravity model areGross Domestic Product (GDP), GDP per capita, andthe distance between states. Following standard prac-tice, we include the product of the two states’ GDPand GDP per capita. Both GDP and GDP per capitaare measured in constant dollars. Distance is the great-circle distance between capital cities or nearest majorcities. Data for these variables come from Rose (2004).

Finally, our transaction cost story of dyadic tradeleads us to include two other factors that influencetransaction costs. First, it is far easier, that is less costly,to conduct commercial exchanges when two actorsspeak the same language. Language may also serve as aproxy for a variety of cultural factors that influencepreferences and therefore trade flows. Not surprisingly,

TABLE 1 Average Level of Trade by Dyad Type

Type of DyadAverage Level of Trade(exports and imports)

Joint Market Protecting andJoint Democratic

446,859

Joint Market Protecting butnot Joint Democratic

110,194

Joint Democratic but notJoint Market Protecting

80,017

Neither Joint MarketProtecting nor JointDemocratic

20,952

4More specifically, CIM is the ratio of (M2-C)/M2, where M2 ismoney supply and C is currency held outside banks. In theinternational financial statistics, M2 is the sum of lines 14a, 24,15, 25, and C is line 14a.

5The essence of the weak link argument is that trade flows (or anappropriate dependent variable) are constrained by the set ofinstitutions in the dyad that are less fully developed. The statewhose institutions are less developed is the weak link.

6This is the ‘‘polity2’’ variable in the Polity IV dataset.

7For an alternative view of the impact of conflict on trade,Barbieri and Levy (1999) find that even states involved in violentconflicts continue to have some trade though at less thanpreconflict levels.

8We use Bennet and Stam’s (2000) Eugene program (version3.040) to generate the MID data.

promoting trade: the importance of market protecting institutions 387

quantitative analyses consistently find that languagesimilarity between states strongly influences tradeflows (e.g., Bliss and Russett 1998; Glick and Rose2002; Rose 2004). The variable Language equals 1 ifthe primary language spoken in each state is thesame. Data come from Rose (2004). Second, regionaltrade agreements should reduce transaction costs, andin some cases, tariffs, thereby increasing trade flows.Indeed, Frankel, Stein, and Wei (1995) and Mansfieldand Bronson (1997) have found that regional tradeagreements lead to a statistically significant increase inbilateral trade. The variable Regional Trade Agreementequals 1 if both states are members of any of thefollowing regional trade agreements: ASEAN, EEC/EC/EU, US-Israel FTA, NAFTA, CARICOM, PATCRA,ANZCERTA, CACM, SPARTECA, and Mercosur.Data on regional trade organizations comes from theWorld Trade Organization.9

Based on the above, we posit the following modelof bilateral trade flows.

lnTradeijtþ1 ¼ b0 þ b�1 ln ðGDP�i GDPjÞþ b�2 lnðGDPPC�i GDPPCjÞþ b�3lnDistanceij

þ b�4Market Protecting Institutions

þ b�5Democracy

þ b�6Market-Democracy Interaction

þ b�7Allinesþ b�8Fatal MID

þ b�9Language

þ b�10Regional Trade Aggrementþ u

The parameters b1, b2, b3 represent the standardgravity influences on trade. The value of primaryinterest to us is (b4 + b6*Democracy), which weexpect to be positive regardless of the value ofDemocracy. The marginal impact of democracy(b5 + b6*MPI), however, should only be positiveat higher levels of MPI. That is, democracies that donot offer significant protection to property rightsare not likely to see greater levels of trade, but thosethat do protect property rights should see an increasein trade.

Beck and Katz show that ‘‘the temporal and spatialproperties of TSCS (time-series cross-sectional) datamake the use of ordinary least squares (OLS) prob-lematic’’ (1995, 634). To address the problem, theyrecommend the use of panel-corrected standard errors

(PCSE). Unless otherwise specified, the analyses belowemploy PCSE.10

Empirical Results

We begin the analysis by examining particular sets ofdyads. If the argument we advance is accurate, thendyads that afford greater protection of property rightsshould have more trade, on average, than otherdyads. In addition, dyads with both well developedmarket-protecting and democratic political institu-tions should have more trade than dyads with onlyone set or none of these institutions. To examine thesedifferent subsets, we distinguish between democraticand nondemocratic states and market-protecting statesand nonmarket-protecting states. A state is democraticif its democracy index score is six or greater. A state ismarket protecting if its market institutions score isgreater than the 90th percentile.11 Based on these twovariables we have four types of dyads: joint market-protecting and democratic dyads, market-protectingdyads that are not jointly democratic, democraticdyads that are not jointly market protecting, anddyads that are neither jointly democratic nor marketprotecting.

As shown in Table 1, dyads that afford the mostprotection to property rights, that is jointly demo-cratic and market-protecting dyads, have the highestlevels of trade. Market protecting dyads that are notboth democratic have the second highest amounts oftrade. Representative examples of this group includeTrinidad-Swaziland (1999) and Brazil-Lesotho (1992).Democratic dyads that are not market-oriented havethe third highest amount of trade, and nondemocratic,nonmarket-protecting dyads have the least amount oftrade. There are quite a few democratic states withweak market-protecting institutions, and the followingare jointly democratic dyads but with weak market-protecting institutions: South Korea-Philippines (1997),Greece-Mongolia (1993), Ireland-Poland (1991), andHonduras-India (1996). While both market and dem-ocratic institutions lead to more commerce, Table 1provides preliminary evidence that market institutionsmatter more than democratic institutions. Further,difference-of-means tests indicate that the differencesbetween each category are statistically significant,including the difference between market-protecting

9www.wto.org/english/tratop_e/region_e/region_e.htm

10Specifically, we use Stata’s xtpcse command.

11We do not claim that only those states with a marketinstitutions score in the top 10% are market oriented. We usethe top 10% simply as a convenient focal point for identifyingtypes.

388 mark souva, dale l. smith, and shawn rowan

dyads without democracy and democratic dyads with-out well-developed market-protecting institutions.

Next, we examine whether these institutionalfactors are still relevant once we control for economicdeterminants of trade. To ensure that new data arenot driving the results in our analysis, we begin witha baseline model similar to that found in previousresearch (see e.g., Bliss and Russett 1998). In Model 1a(see Table 2), we analyze the effect of democratic andsecurity (alliance ties) institutions on bilateral tradeflows while controlling for the basic gravity variables.First, all of the gravity variables perform as expected.

Bigger and wealthier dyads have more trade thanother dyads, and the distance between states reducestrade. Second, we find that democratic institutionsexercise a statistically significant and positive influ-ence on dyadic trade flows. Third, the variable Allies,after controlling for the gravity variables and democ-racy, does not have a statistically significant effect ontrade. These results are similar to those reported byBliss and Russett (1998).

In Model 1b, we add the market-protectinginstitutions and market-protecting institutions-democracy interaction variables to the model. Turn-ing to the variables of interest for our hypotheses, weshow in Figure 1 that the marginal effect of market-protecting institutions increases dyadic trade flowsfor most values of democracy.12 While the impact ofthis variable is stronger in more democratic dyads, evenin dyads with weakly developed democratic institutions,increases in market-protecting institutions will increasebilateral trade. Nevertheless, contrary to Hypothesis 1,in dyads with at least one strongly nondemocratic state,the effect of market-protecting institutions on trade isnot significant. As we discuss below, the nonsignificanceof market-protecting institutions in the presence ofstrongly nondemocratic dyads is confined to thisparticular estimation method. Figure 2 offers an assess-ment of Hypothesis 2. When market institutions donot afford significant protection of property rights,democratic institutions do not positively affect bilateraltrade. However, as market institutions are strengthened,we observe an increasingly positive impact of demo-cratic institutions on trade flows. Where previousresearch has suggested that democracy unequivocallypromotes trade, we find that democratic politicalinstitutions do not increase trade flows when market

FIGURE 1 Marginal Effect of Market ProtectingInstitutions on Trade across theRange of Democarcy Scores

-0.5

0

0.5

1

1.5

-10 -8 -6 -4 -2 0 2 4 6 8 10

Democracy

Marg

inal E

ffect

of

MP

I o

n T

rad

e

Marginal Effect = β4 + β6(Democracy)

95% confidence interval

TABLE 2 Regression Analysis of MarketInstitutions and Dyadic Trade Flows,1960-1999

Model 1a:Base Model

Model 1b:Full Model

Market ProtectingInstitutions (MPI)

.532** (.115)

Democracy .017** (.002) 2.018** (.009)MPI*Democracy .043** (.012)Allies .038 (.041) 2.057 (.046)Fatal MID 2.280** (.083) 2.222** (.117)Ln(Distance) 21.47** (.020) 21.530** (.022)Ln(Gross Domestic

Product).884** (.007) .873** (.008)

Ln(GDP Per Capita) .272** (.013) .360** (.014)Common Language .552** (.043) .593** (.045)Regional Trade

Agreement.209** (.056) .252** (.059)

Constant 225.096** (.342)225.989** (.370)N 96844 77815R2 0.412 0.446

** 5 p-value less than .05.Panel-corrected standard errors reported in parentheses.

FIGURE 2 Marginal Effect of Democracy onTrade across the Range of MPI Values

-0.1

0

0.1

0 0.2 0.4 0.6 0.8 1

Market Protecting Institutions (MPI)

Marg

inal E

ffect

of

Dem

ocra

y o

n T

rad

e

Marginal Effect = β5 + β6(MPI)

95% confidence interval

12Figures 1 and 2 were produced using the STATA do programcreated by Brambor, Clark, and Golder (2006).

promoting trade: the importance of market protecting institutions 389

institutions are not well developed. The minimal impactof democracy on trade is most likely to be the casein weak or transitional democracies where market-protecting institutions are poorly developed. Finally,like Model 1a, the estimates of Model 1b reveal nostatistically significant relationship between securityinstitutions (alliances) and bilateral trade flows.

Robustness and Sensitivity Analysis

We undertook a variety of sensitivity analyses toassess the robustness of our results. Simply put, wewant to know if support for our hypotheses is limitedto a particular estimation method, particular measureof the central concept, or sample. Fixed-effects andGEE estimations also reveal support for our hypoth-eses (see Table 3).13 Each of these estimation methodsindicates that a uniformly positive influence ofmarket-protecting institutions on trade, that theimpact of democratic institutions is conditional onwhether market-protecting institutions are well de-veloped in the dyad, and that security institutions donot have any statistically meaningful influence ontrade flows.

Next, we examined the sensitivity of our findingsto particular subsamples. Is it the case that market-protecting and democratic institutions only positively

influence trade flows between wealthy countries ordo they influence trade in both wealthy and poordyads? To answer this question, we disaggregated thesample into OECD and non-OECD dyads. In OECDdyads, we continue to observe a positive effect ofmarket-protecting institutions, for all levels of de-mocracy, on bilateral trade (see Table 4). In non-OECD dyads, market institutions also have a con-sistently positive effect, though the marginal effect isonly significant when democracy is above zero.Democratic political institutions are not statisticallyrelated to trade flows in OECD dyads. Given that allOECD countries are democratic this is perhaps notsurprising: a constant, democracy, cannot explain thevariation in trade across these dyads. In non-OECDdyads, we observe the same relationship betweendemocratic institutions and trade as in our globalsample: as property rights become better protected,democratic institutions lead to more trade.

Finally, we reanalyzed all of the models with analternative measure of market-protecting institutions.Instead of contract-intensive money (CIM), wemeasured a state’s protection of property rights usingthe investment profile score from the InternationalCountry Risk Guide (ICRG). A state’s investmentprofile score is a function of the risk of contractexpropriation, payment delays, and profit repatria-tion. The variable ranges from 0 to 12, with higherscores representing a greater level of market protection.Data cover the period 1985 to 1999 for about 100states.

Although the ICRG data has a more limitedtemporal and spatial domain and is more subjective

TABLE 3 Regression Analyses Using Alternative Estimators: FixedEffects and GEE

Fixed Effects GEE

Market Protecting Institutions (MPI) .926** (.078) 1.017** (.107)Democracy 2.014** (.005) 2.024** (.008)MPI*Democracy .037** (.007) .040** (.011)Allies 2.008* (.041) 2.041 (.053)Fatal MID 2.432** (.169) 2.154 (.116)Ln(Distance) 21.130** (.030)Ln(Gross Domestic Product) .162** (.016) .776** (.010)Ln(GDP Per Capita) .630** (.022) 2.030 (.016)Common Language .599** (.060)Regional Trade Agreement .310** (.068) .267** (.100)Constant 28.151** (.496) 217.817** (.436)N 77815 48642R2 0.45 0.446

** 5 p-value less than .05.Standard errors reported in parentheses.

13For the generalized estimating equation (GEE), we specified thewithin-group correlation structure as an AR1 process. We alsocompared the fixed-effects estimates with random effects esti-mates. Both the Hausman and Breusch-Pagan tests rejected theappropriateness (consistency) of the random effects estimatorcompared to the fixed-effects model.

390 mark souva, dale l. smith, and shawn rowan

than CIM, it is a more common measure of marketinstitutions in extant research (see for example, Sobel1999; Knack and Keefer 1995; Li and Resnick 2003;Souva 2004). The correlation between the two meas-ures is about .60. Despite the decrease in sample size,we continue to find support for our hypotheses whenusing the ICRG measure of market-protecting in-stitutions. Hypothesis one is supported regardless ofthe estimation method, and Hypothesis two receivesunequivocal support from the fixed-effects and GEEestimations, and mild support in the panel correctedstandard errors model. In that model, we find apositive effect of democracy on bilateral trade at bothhigh and low levels of market-protecting institu-tions.14 Finally, substituting the ICRG investmentprofile measure of market-protecting institutions forthe CIM measure does not alter the findings onsecurity institutions: alliance ties continue to exhibitno significant relationship with bilateral trade flows.

Conclusion

Politics affects trade, but not precisely the way previousresearch has claimed. The primary influence of politicsis not related to alliance ties or the political system in astate. Rather, it revolves around the development andenforcement of market-protecting institutions. This isas it should be if firms, not states, are the primary agents

conducting international trade. By emphasizing micro-foundations, we argued that factors affecting the pro-tection of property rights and the cost of conductingcommercial exchange are likely to influence trade flowsmore than military alliances, and the empirical evidencesupports this argument.

Contrary to previous research, we also contendedthat political institutions, in and of themselves, areless important for understanding trade flows thanmarket-protecting institutions. Firms are less con-cerned about voting rights of citizens than they areabout the protection of their assets. To be sure, demo-cratic political systems are, on average, more likely tocultivate market-protecting institutions, but theseinstitutions may exist absent democratic institutions.And, weak democracies may have an especiallydifficult time developing market-protecting institu-tions. Trade generates wealth and wealth is crucial forconsolidating democracy. Therefore, an implicationof our research is that nondemocratic states thatcultivate market-protecting institutions may be morelikely to successfully transition to democracy thanstates implementing democratic reforms prior tosignificant market reforms.

Manuscript submitted 10 January 2006Manuscript accepted for publication 10 August 2006

References

Barbieri, Katherine, and Jack Levy. 1999. ‘‘Sleeping With theEnemy: The Impact of War on Trade.’’ Journal of PeaceResearch 36 (4): 463–79.

TABLE 4 Subsample Analyses

OECD Dyads Non-OECD Dyads

Market ProtectingInstitutions (MPI)

1.797** (.221) .626** (.226)

Democracy 2.016 (.014) 2.037** (.020)MPI*Democracy .034 (.019) .061** (.027)Allies .148** (.032) 2.009 (.087)Fatal MID 2.506** (.224) 2.329** (.178)Ln(Distance) 2.787** (.027) 21.535** (.030)Ln(Gross Domestic Product) .765** (.017) .790** (.011)Ln(GDP Per Capita) 2.186** (.030) .357** (.020)Common Language .965** (.075) .236** (.078)Regional Trade Agreement .124 (.027) 1.196** (.137)Constant 217.153** (.661) 222.441** (.553)N 6129 32195R2 0.89 0.30

** 5 p-value less than .05.Standard errors reported in parentheses.

14Since nothing significant changes and for reasons of space,results are not shown but are available from the authors. Analyseswere also run with an ICRG index instead of only the ‘‘invest-ment’’ variable and the results are similar.

promoting trade: the importance of market protecting institutions 391

Beck, Nathaniel, and Jonathan Katz. 1995. ‘‘What to do (and notto do) with Time-Series-Cross-Section Data.’’ American Polit-ical Science Review 89 (3): 634–47.

Bennett, D. Scott, and Allan Stam. 2000. EUGene: A Concep-tual Manual. International Interactions 26: 179–204. http://eugenesoftware.org.

Bergstrand, Jeffrey H. 1985. ‘‘The Gravity Equation in Interna-tional Trade: Some Microeconomic Foundations and Em-pirical Evidence.’’ Review of Economics and Statistics 67 (3):474–81.

Bergstrand, Jeffrey H. 1989. ‘‘The Generalized Gravity Equation,Monopolistic Competition, and Factor-Proportions Theory inInternational Trade.’’ Review of Economics and Statistics 71 (1):143–53.

Brambor, Thomas, William Clark, and Matt Golder. 2006.Understanding Interaction Models: Improving EmpiricalAnalyses. Political Analysis 14 (1): 63–82.

Bliss, Harry, and Bruce Russett. 1998. ‘‘Democratic TradingPartners: The Liberal Connection, 1962–1989.’’ Journal ofPolitics 60 (4): 1126–47.

Bueno de Mesquita, Bruce, James Morrow, Randolph Siverson,and Alasdair Smith. 2003. The Logic of Political Survival.Cambridge: MIT Press.

Clague, Christopher, Philip Keefer, Stephen Knack, and MancurOlson. 1999. Contract-Intensive Money: Contract Enforce-ment, Property Rights, and Economic Performance. Journal ofEconomic Growth 4: 185–211.

Frankel, Jeffrey, Ernesto Stein, and Shang-jin Wei. 1995. Tradingblocs and the Americas: The natural, the unnatural, and thesuper-natural. Journal of Development Economics 47 (1): 61–95.

Gibler, Douglas, and Meredith Sarkees. 2004. ‘‘Measuring Alli-ances: the Correlates of War Formal Interstate Alliance Data-set, 1816–2000.’’ Journal of Peace Research 41 (2): 211–22.

Glick, Reuven, and Andrew K. Rose. 2002. Does a CurrencyUnion Affect Trade? The Time-Series Evidence. EuropeanEconomic Review 46 (6): 1125–51.

Gowa, Joanne. 1989. ‘‘Bipolarity, Multipolarity, and Free Trade.’’American Political Science Review 83 (4) 1245–56.

Gowa, Joanne, and Edward D. Mansfield. 1993. ‘‘Power Politicsand International Trade.’’ American Political Science Review 87(2): 408–20.

Gowa, Joanne, and Edward D. Mansfield. 2004. ‘‘Alliances,Imperfect Markets, and Major-Power Trade. InternationalOrganization 58(Fall): 775–805.

Helpman, Elhanan, and Paul R. Krugman. 1985. Market Structureand Foreign Trade: Increasing Returns, Imperfect Competition,and the International Economy. Cambridge: MIT press.

Helpman, Elhanan. 1999. ‘‘The Structure of Foreign Trade.’’ TheJournal of Economic Perspectives 13(2): 121–44.

Keshk, Omar, Brian Pollins, and Rafael Reuveny. 2004. ‘‘TradeStill Follows the Flag: The Primacy of Politics in a Simulta-neous Model of Interdependence and Armed Conflict.’’Journal of Politics 66 (4): 1155–79.

Knack, Stephen, and Philip Keefer. 1995. ‘‘Institutions andEconomic Performance: Cross Country Tests Using Alterna-tive Institutional Measures.’’ Economics and Politics 7 (3):207–27.

Krugman, Paul. 1980. ‘‘Scale Economies, Product Differentiation,and the Pattern of Trade.’’ American Economic Review 70 (5):950–59.

Leeds, Brett Ashley. 1999. ‘‘Domestic Political Institutions,Credible Commitments, and International Cooperation.’’American Journal of Political Science 43 (4): 979–1002.

Lektzian, David, and Mark Souva. 2001. ‘‘Institutions andInternational Cooperation: A Duration Analysis of the Effectsof Sanctions.’’ Journal of Conflict Resolution 45 (1): 61–79.

Li, Quan, and Adam Resnick. 2003. Reversal of Fortunes:Democratic Institutions and Foreign Direct InvestmentInflows to Developing Countries. International Organization57 (Winter): 175–211.

Li, Quan, and David Sacko. 2002. ‘‘The (IR)Relevance ofInterstate Militarized Disputes to International Trade.’’ Inter-national Studies Quarterly 46 (1): 11–44.

Mansfield, Edward D., and Rachel Bronson. 1997. ‘‘Alliances,Preferential Trading Arrangements, and International Trade.’’American Political Science Review 91 (1): 94–107.

Mansfield, Edward D, Helen V. Milner, and B. Peter Rosendorff2000. Free to Trade: Democracies, Autocracies, and InternationalTrade. American Political Science Review 94 (2): 305–21.

Marshall, Monty, and Keith Jaggers. 2002. Polity IV Project:Dataset Users Manual. http://www.cidcm.umd.edu/inscr/polity/index.htm

Morrow, James D., Randolph Siverson, and Tressa Tabares. 1998.‘‘The Political Determinants of International Trade: TheMajor Powers, 1907–90.’’ American Political Science Review92 (3): 649–61.

Morrow, James D., Randolph Siverson, and Tressa Tabares. 1999.‘‘Correction to: The Political Determinants of InternationalTrade.’’ American Political Science Review 93 (4): 931–33.

North, Douglass C. 1990. Institutions, Institutional Change andEconomic Performance. New York: Cambridge UniversityPress.

Polachek, Solomon W. 1980. ‘‘Conflict and Trade.’’ Journal ofConflict Resolution 24 (1): 55–78.

Pollins, Brian M. 1989a. ‘‘Does Trade Still Follow the Flag?’’American Political Science Review 83(2): 465–80.

Pollins, Brian M. 1989b. ‘‘Conflict, Cooperation, and Commerce:The Effects of International Political Interactions on BilateralTrade Flows.’’ American Journal of Political Science 33 (3):737–61.

Rose, Andrew K. 2004. ‘‘Do We Really Know that the WTOIncreases Trade?’’ American Economic Review 94 (1): 98–115.

Sobel, Andrew. 1999. State Institutions, Global Incentives, PrivateCapital. Ann Arbor: University of Michigan Press.

Souva, Mark. 2004. ‘‘Institutional Similarity and InterstateConflict.’’ International Interactions 30 (3): 263–80.

Tsebelis, George. 2002. Veto Players: How Political InstitutionsWork. Princeton: Princeton University Press.

Mark Souva is assistant professor of politicalscience, Florida State University, Tallahassee, FL32306. Dale L. Smith is associate professor of politicalscience, Florida State University, Tallahassee, FL32306. Shawn Rowan is a Ph.D. candidate in politicalscience, Florida State University, Tallahassee, FL32306.

392 mark souva, dale l. smith, and shawn rowan