# Chapter 8: Constitutions and Electoral Systems **Key papers to cover:** Persson and Tabellini (2003) *The Economic Effects of Constitutions*, Lijphart (1999) *Patterns of Democracy*, Cox (1997) *Making Votes Count*, Duverger (1954) electoral laws, Lizzeri and Persico (2001) provision of public goods, Milesi-Ferretti, Perotti, and Rostagno (2002) electoral systems and fiscal policy, Myerson (1993) comparing electoral systems, Carey and Shugart (1995) incentives for personal vote *** ## Opening Puzzle The United States and Brazil adopted presidential systems; Germany and India chose parliamentary ones. These constitutional choices, often made decades ago under extraordinary circumstances, continue to shape policy outcomes today. Consider two neighbors: Costa Rica and Guatemala. They share a border, colonial heritage, similar geography, and a common language. Both gained independence in 1821, both experienced authoritarian episodes, and both are small Central American economies dependent on agriculture and trade. Yet their political outcomes diverged dramatically. Costa Rica abolished its army in 1948 and has been continuously democratic since, with regular peaceful transfers of power. It became the most stable democracy in Latin America, with strong public services, low inequality by regional standards, and per capita income three times Guatemala's. Guatemala, by contrast, experienced 36 years of civil war, military coups, genocide against indigenous populations, and persistent instability. Democracy arrived only in the 1990s and remains fragile. What explains the difference? The easy answer—culture, geography, initial conditions—fails because neighboring countries with similar backgrounds diverged. Political economists instead point to *institutions*: the rules governing how decisions are made, how power is allocated, and how conflicts are resolved. Costa Rica's 1949 constitution created strong checks on executive power, an independent electoral tribunal, and rules that dispersed political authority. Guatemala's constitutional history is one of concentration, manipulation, and breakdown. This chapter asks: Does it matter whether a country has a presidential or parliamentary system? Whether elections use plurality or proportional representation? Whether power is centralized or federal? The answer, developed through theory and evidence, is that institutions matter enormously—but in subtle and context-dependent ways. *** ## Stylized Facts: Constitutional and Electoral Variation Across Democracies The Costa Rica–Guatemala contrast in the opening puzzle reflects a broader empirical pattern: comparable countries with similar histories diverge dramatically based on their political institutions. Before developing theory, we document this variation. Electoral systems vary widely. Among the world's democracies, roughly 40% use proportional representation as their primary legislative electoral system, about 35% use plurality or majority systems (first-past-the-post), and the remainder use mixed systems (IDEA Electoral System Design Database). The distribution follows regional patterns: Western Europe is predominantly PR; the Anglophone world (UK, US, Canada, India) uses plurality; Latin America and Eastern Europe are mixed. The choice between presidential and parliamentary systems also clusters geographically. Presidential systems are the norm in the Americas (North and South) and much of Africa. Parliamentary systems dominate Western Europe, South and Southeast Asia, and most of the Commonwealth. Semi-presidential systems are common in post-communist Europe and Francophone Africa. In many cases, the choice reflects colonial inheritance as much as deliberate design. The effective number of parties varies predictably with electoral system. Countries using single-member plurality elections average around 2.0–2.5 effective parliamentary parties; those using PR with multi-member districts average 4.0–5.0 (Lijphart 1999). This regularity—Duverger's Law—is one of political science's most robust empirical generalizations. Constitutions also vary enormously in how difficult they are to amend. The U.S. Constitution has been amended only 27 times since 1789. France has been governed by five Republics since 1789. Among current democracies, some constitutions (Germany's Basic Law, India's Constitution) have been amended hundreds of times; others rarely. Rigidity predicts which rights and rules are durable versus subject to revision by temporary majorities. Finally, about 25–30% of the world's population lives in formally federal states (United States, Germany, India, Brazil, Canada, Australia, Nigeria, Mexico). Federal states account for a disproportionate share of global GDP and territory. Within federal systems, the degree of fiscal decentralization—what share of public spending is controlled at subnational levels—varies from under 10% (some nominally federal states) to over 50% (Switzerland, United States). These patterns motivate the theoretical frameworks developed in this chapter. *** ## 8.1 What Is a Constitution? ### 8.1.1 Functions of Constitutions A constitution is the fundamental law governing political authority: who holds power, how they obtain it, what they can do with it, and how they can be removed. Constitutions serve several interlocking functions. First, they act as coordination devices, establishing common knowledge about political rules. When everyone knows the rules and knows that everyone else knows, coordination becomes possible—a president who loses an election steps down because everyone expects compliance with constitutional rules. Second, constitutions serve as commitment devices, binding future governments and protecting rights and policies from reversal; because constitutional provisions are harder to change than ordinary laws, they create durable commitments. Third, constitutions function as incomplete contracts: no document can specify responses to every contingency, so constitutions must be interpreted, and institutions for interpretation (courts, customs, political practice) fill the gaps. Fourth, constitutions operate as power maps, allocating authority among branches, levels of government, and between state and citizen. This allocation shapes policy by determining who holds veto power and agenda control. ### 8.1.2 Constitutional Rigidity Constitutions vary in how difficult they are to amend. At one extreme, rigid constitutions require supermajorities, multiple votes, ratification by subnational units, or other hurdles. The U.S. Constitution is among the most rigid: amendments require two-thirds of both houses of Congress and ratification by three-fourths of state legislatures, and only 27 amendments have passed since 1789. At the other extreme, flexible constitutions can be amended by ordinary legislative majorities. The UK has no single constitutional document; Parliament can change fundamental rules by simple majority. New Zealand similarly has a flexible constitutional framework. Between these poles lie intermediate cases such as Germany's Basic Law, which permits amendment by two-thirds majority but makes some provisions (human dignity, federalism, democracy) unamendable. Rigidity creates a tradeoff. Rigid constitutions provide stronger commitment—rights and rules are protected from transient majorities. But rigidity also prevents adaptation to changing circumstances and may generate pressure for extra-constitutional change. ### 8.1.3 Constitutional Origins Why do constitutions take the forms they do? Several theories offer competing explanations. The efficient bargaining view holds that constitutional designers choose rules maximizing joint surplus, subject to ensuring all parties prefer the constitution to alternatives. North and Weingast (1989) interpret England's Glorious Revolution as establishing constitutional constraints that enabled credible commitment to property rights. A redistributive conflict perspective instead emphasizes that constitutions reflect the balance of power at the moment of drafting—groups write rules that favor their interests. Acemoglu and Robinson (2006) model constitutions as emerging from conflict between elites and citizens, with outcomes depending on each group's ability to threaten the other. A third view stresses historical inheritance: many constitutional features reflect colonial origins, imposed templates, or path-dependent evolution rather than deliberate design. Former British colonies tend toward Westminster parliamentarism; former French colonies toward presidentialism. Finally, imitation and diffusion play a role, as constitution-makers often copy successful models. The wave of democratic constitutions after 1989 drew heavily on Western European templates, particularly Germany's Basic Law. Constitutional design is intimately linked to the mode of democratic transition analyzed in Chapter 5. Pacted transitions—negotiated agreements between regime and opposition—tend to produce constitutions that protect outgoing elites through amnesty provisions, property rights guarantees, and reserved seats, as in Chile and South Africa. Constitutions written after regime collapse or revolution reflect the balance of power among opposition factions rather than between old and new regimes. The founding moment thus casts a long shadow: constitutional provisions adopted under the pressures of transition prove remarkably persistent, shaping politics for decades after the original circumstances have changed. *** ## 8.2 Electoral Systems ### 8.2.1 The Basic Choice Electoral systems translate votes into seats. The fundamental distinction runs among three families. Under plurality or majority systems, the candidate with the most votes wins. In single-member districts with plurality rule ("first past the post"), one seat goes to whoever gets the most votes, even without a majority; the UK, US, Canada, and India use this system. Under proportional representation (PR), seats are allocated in proportion to vote shares—if a party wins 30% of votes, it receives approximately 30% of seats. Most European democracies use PR. Mixed systems combine both approaches, allocating some seats by plurality and others by PR. Germany's mixed-member proportional system combines single-member districts with proportional allocation to achieve overall proportionality; Japan and New Zealand use variants. ### 8.2.2 Duverger's Law and Hypothesis Maurice Duverger (1954) proposed two regularities, which have implications not just for party systems but for electoral competition more broadly (see Chapter 12 for formal models of how electoral rules shape candidate positioning and voter behavior): {% hint style="info" %} **Duverger's Law.** Plurality rule in single-member districts tends to produce two-party systems through (1) the mechanical effect (third parties win fewer seats than votes) and (2) the strategic effect (voters and donors abandon non-viable candidates). Proportional representation permits multiparty systems. {% endhint %} Three mechanisms drive these regularities. The mechanical effect is straightforward: plurality rule systematically underrepresents small parties. A party with 15% support spread evenly across districts may win zero seats if it never leads in any district, whereas PR translates 15% of votes into approximately 15% of seats. The strategic effect compounds the mechanical one. Voters anticipate the translation of votes into seats, and supporters of small parties under plurality rule face incentives to vote for their second choice among viable candidates ("strategic voting" or "lesser evil" voting), further concentrating votes on major parties. Finally, entry deterrence matters: political entrepreneurs do not form parties that cannot win, and under plurality rule, barriers to third-party entry are high. **Formal statement (Cox 1997):** In a single-member district with rational voters, the number of viable candidates is at most $$M + 1$$, where $$M$$ is the number of winners (i.e., 2 candidates for single-member plurality). More precisely, if voters have strict preferences and vote strategically, at most two candidates receive positive votes in equilibrium. ### 8.2.3 District Magnitude **District magnitude** $$M$$ is the number of representatives elected per district. Single-member districts have $$M = 1$$. PR systems typically have multi-member districts with $$M$$ ranging from 3-5 (small) to the entire country as one district (Netherlands, Israel). Higher district magnitude increases proportionality, lowers the threshold for representation, and tends to increase the effective number of parties. The **effective number of parties** (Laakso and Taagepera 1979) is: $$ N = \frac{1}{\sum\_{i} s\_i^2} $$ where $$s\_i$$ is party $$i$$'s seat share. If two parties split seats equally, $$N = 2$$. If one party has all seats, $$N = 1$$. Empirically, the relationship between district magnitude and party fragmentation is approximately: $$ N \approx M^{0.25} $$ This suggests diminishing returns: increasing $$M$$ from 1 to 5 increases fragmentation substantially; increasing from 5 to 100 adds relatively less.
Figure 8.1: Electoral Systems and Party Fragmentation

Figure 8.1: Average district magnitude vs. effective number of parties across democracies. Source: Bormann and Golder (2013), updated.

### 8.2.4 Thresholds and Formulas PR systems use various formulas to convert votes to seats. The D'Hondt method divides each party's votes by 1, 2, 3, ... and allocates seats to the highest quotients, slightly favoring large parties. The Sainte-Laguë method divides by 1, 3, 5, ... and produces more proportional outcomes. Largest remainder methods allocate seats based on quotas, with remaining seats going to parties with the largest remainders. Many PR systems also impose explicit thresholds—minimum vote shares required for representation. Germany's 5% threshold prevents fragmentation but excludes small parties. Turkey's 10% threshold (among the world's highest) has produced highly disproportionate outcomes, as measured by the Gallagher index of seat-vote disproportionality (Gallagher 1991; Lijphart 1999). *** ## 8.3 Presidential vs. Parliamentary Systems ### 8.3.1 Defining Features Presidential systems have separately elected executives and legislatures, each with independent constitutional authority; the president is not dependent on legislative confidence. The U.S., most of Latin America, and some African countries use presidential systems. Parliamentary systems, by contrast, have executives (prime ministers) who emerge from and depend on legislative confidence. The government can be removed by a vote of no confidence. Most European democracies, plus Canada, Australia, India, and Japan, use parliamentary systems. Semi-presidential systems combine elements of both: a directly elected president with significant powers coexists with a prime minister dependent on legislative confidence. France (Fifth Republic), Russia, and several post-communist states use this hybrid. ### 8.3.2 Separation of Purpose vs. Survival Shugart and Carey (1992) clarify the distinction along two dimensions. Separation of survival means that in presidential systems, the executive and legislature have independent mandates—neither can remove the other except through extraordinary procedures like impeachment. Separation of purpose means that when the executive and legislature are elected separately, they may represent different constituencies, time horizons, and policy preferences; a nationally elected president may differ sharply from a Congress dominated by local interests. Parliamentary systems fuse survival (the government depends on parliament) but may separate purpose if coalition partners have different preferences. ### 8.3.3 Linz's Critique of Presidentialism Juan Linz (1990) argued that presidential systems are inherently less stable than parliamentary systems, identifying four sources of fragility. The problem of dual legitimacy arises because both president and legislature claim democratic mandates; when they conflict, no constitutional mechanism resolves the dispute, inviting constitutional crisis. Rigidity compounds the problem: fixed terms mean failing presidents cannot be removed absent impeachment, whereas parliamentary systems can replace governments through votes of no confidence. The winner-take-all character of presidential elections further raises the stakes, since losers are excluded from executive power entirely—increasing the cost of losing and potentially encouraging extra-constitutional actions. Finally, presidential systems enable outsider candidates without party experience or legislative support to win power. Such presidents may lack the coalitional skills needed to govern effectively. ### 8.3.4 Cheibub's Response José Antonio Cheibub (2007) challenged the empirical claim that presidentialism causes instability. He pointed first to confounding: presidential systems are concentrated in Latin America, a region with many independent sources of instability, and controlling for region, income, and other factors weakens the presidential effect. He also emphasized the military legacy of many presidential systems—where transitions from military rule produce presidential constitutions, the military retains influence and may intervene regardless of constitutional form, so the instability attributed to presidentialism may actually reflect military politics. Cheibub further argued that, contrary to Linz, presidential systems can and do form legislative coalitions; Brazilian presidents routinely govern through multiparty coalitions assembled with cabinet positions and pork. The debate remains open. Recent evidence tilts toward parliamentary systems being more stable, but the effect is modest and context-dependent. *** ## 8.4 Federalism and Decentralization ### 8.4.1 What Is Federalism? Federalism divides authority between central and subnational governments, with each level having constitutionally guaranteed powers. True federations (U.S., Germany, Switzerland, Brazil, India) differ from unitary systems, in which the central government holds ultimate authority and can reorganize or abolish subnational units (France, UK, Japan), and from mere decentralization, in which the central government delegates authority but retains the power to reclaim it. Federal systems themselves vary in how powers are allocated. Under dual federalism, each level has exclusive jurisdiction over certain domains, as in the original U.S. design. Under cooperative federalism, powers are shared and overlapping, with levels collaborating on most policies, as in Germany and the contemporary United States. ### 8.4.2 Why Federalism? Several arguments favor federalism. Where regions have different preferences, decentralization allows policies tailored to local conditions; Tiebout (1956) argues citizens can "vote with their feet," choosing jurisdictions matching their preferences. Local governments may also have better information about local needs and conditions. Subnational units can serve as "laboratories of democracy," testing policies before national adoption. Multiple levels of government create veto points that constrain central authority. And large, heterogeneous countries may require federalism to accommodate regional identities and prevent secession (India, Nigeria, Ethiopia). ### 8.4.3 Costs of Federalism Federalism also carries costs. Decentralized systems may under-provide policies with spillovers across jurisdictions, such as environmental regulation or infrastructure networks. Jurisdictions competing for mobile capital may reduce taxes and regulation below efficient levels—the "race to the bottom." Decentralization may permit or exacerbate regional inequalities in public services. Local elites may more easily capture small governments than national ones. And multiple levels of government create opportunities for blame-shifting, confusion about responsibility, and administrative inefficiency. ### 8.4.4 Fiscal Federalism The study of fiscal federalism analyzes how taxing and spending powers should be allocated across levels of government. {% hint style="info" %} **Oates's Decentralization Theorem.** In a federal system without spillovers or economies of scale, welfare is at least as high with decentralized provision as with centralized uniform provision. {% endhint %} Formally, let regions $$i = 1, ..., n$$ have different preferences over a public good $$g$$. If the central government must provide the same level $$\bar{g}$$ everywhere, welfare is: $$ W^C = \sum\_i u\_i(\bar{g}) $$ If regions choose independently: $$ W^D = \sum\_i u\_i(g\_i^\*) $$ where $$g\_i^*$$ is region $$i$$'s optimum. Since $$g\_i^* = \arg\max u\_i(g)$$, we have $$u\_i(g\_i^\*) \geq u\_i(\bar{g})$$, so $$W^D \geq W^C$$. The theorem fails if there are spillovers (externalities across regions) or economies of scale, in which case centralization may dominate. *** ## 8.5 Theoretical Frameworks ### 8.5.1 Persson-Tabellini Model of Electoral Systems Persson and Tabellini (2000, 2003) develop models comparing policy outcomes under different electoral systems. **Setup:** Three regions $$j \in {1, 2, 3}$$ with populations normalized to 1. Each region has an equal number of voters. The government provides a public good $$g$$ benefiting all citizens equally and region-specific transfers $$f\_j$$. Citizens in region $$j$$ have utility: $$ u\_j = y - \tau + H(g) + f\_j $$ where $$y$$ is pre-tax income, $$\tau$$ is tax, and $$H(\cdot)$$ is concave. The budget constraint is: $$ 3\tau = g + \sum\_j f\_j $$ **Majoritarian (single-member districts):** Each region elects one representative. The party winning two or more seats controls government. Politicians maximize the probability of winning their district. To win a majority, parties target swing voters in "pivotal" districts—the two most competitive. **Result:** Under majoritarian systems, transfers concentrate in pivotal districts. The two targeted regions receive $$f\_j > 0$$; the third receives $$f\_j = 0$$. Public goods provision is low because resources are diverted to targeted transfers. (The median voter logic underlying these predictions is developed formally in Chapter 12, and the redistributive implications are examined in Chapter 16.) **Proportional representation:** All voters in the country elect representatives proportionally. The party with majority support wins. Since all voters contribute equally to electoral success, parties provide broad benefits. There is no geographic targeting. **Result:** Under PR, transfers are spread broadly (all $$f\_j$$ equal), and public goods provision is higher because they benefit all voters equally. **Formal derivation.** Using the setup above, the social optimum maximizes total welfare: $$ \max\_{g, f\_1, f\_2, f\_3} ; 3(y - \tau) + 3H(g) + \sum\_j f\_j \quad \text{s.t.} \quad 3\tau = g + \sum\_j f\_j $$ Substituting the budget constraint and taking the FOC with respect to $$g$$: $$ 3H'(g^\*) = 1 $$ The optimal public good level equates the aggregate marginal benefit of the public good ($$3H'(g)$$, since all three voters benefit) to its marginal cost. All remaining revenue goes to transfers, distributed equally: $$f\_j^\* = \tau - g^\*/3$$ for all $$j$$. **Under proportional representation:** Parties maximize vote share across all regions. Since all voters contribute equally to electoral success, the party's objective is equivalent to maximizing aggregate welfare. The equilibrium replicates the social optimum: $$g^{PR} = g^*$$ and $$f\_j^{PR} = f^*$$ for all $$j$$. **Under majoritarian elections:** Each region elects one representative. A party wins by securing two of three seats. The party targets the two most competitive ("swing") districts, offering them higher transfers at the expense of the third. The objective becomes: $$ \max\_{g, f\_1, f\_2} ; 2(y - \tau + H(g) + f\_j) \quad \text{s.t.} \quad 3\tau = g + f\_1 + f\_2, ; f\_3 = 0 $$ Taking the FOC with respect to $$g$$: $$ 2H'(g^{MAJ}) = 1 \implies H'(g^{MAJ}) = \frac{1}{2} $$ Since $$H$$ is concave, $$H'(g^{MAJ}) = \frac{1}{2} > H'(g^{PR}) = \frac{1}{3}$$ implies $$g^{MAJ} < g^{PR}$$. Majoritarian systems produce less public goods and more targeted transfers than PR systems. {% hint style="info" %} **Persson-Tabellini Result.** PR systems produce higher public goods spending and more universal transfers. Majoritarian systems produce lower public goods and geographically targeted spending. The mechanism is electoral incentives: under PR, all voters matter equally; under majoritarian rules, swing districts receive disproportionate resources. {% endhint %} ### 8.5.2 Lizzeri and Persico (2001) Lizzeri and Persico model the choice between redistribution and public goods under different electoral rules. **Setup:** Candidates commit to policies before elections. Policy involves allocating budget $$B$$ between a public good $$g$$ with per-capita benefit $$v(g)$$ and transfers $$t\_i$$ targeted to individual voters. Under **proportional representation**, candidates need broad support. They weight all voters equally, choosing policies maximizing average utility. If public goods are efficient ($$v'(0) > 1$$), candidates provide public goods. Under **winner-take-all** plurality, candidates need only a minimal winning coalition (50% + 1). They can win by promising transfers to slightly more than half the voters. If they can target precisely, they ignore public goods and offer targeted redistribution. **Key prediction:** Plurality rule leads to more redistribution and less public good provision than PR, holding preferences constant. ### 8.5.3 Baron-Ferejohn Model The Baron-Ferejohn (1989) model analyzes legislative bargaining, applicable to understanding how electoral systems affect policy through legislative composition. **Setup:** $$n$$ legislators divide a budget of size 1. A randomly selected proposer offers an allocation. If a majority accepts, the allocation is implemented. Otherwise, the process repeats with a new proposer. **Equilibrium:** Let $$\delta < 1$$ be the discount factor. In the stationary subgame perfect equilibrium: The proposer keeps $$\frac{1}{1 + (n-1)\delta/2}$$ for herself, offers $$\frac{\delta}{1 + (n-1)\delta/2}$$ to each of $$(n-1)/2$$ other legislators (a minimal winning coalition), and the remaining legislators receive nothing. The model yields three main implications. The proposer extracts disproportionate benefits from agenda control. Coalitions exclude unnecessary members to maximize per-member shares—the minimum winning coalition result. And which legislators receive transfers depends on who is selected as proposer and as coalition partners, making the distribution of benefits essentially random. ### 8.5.4 Myerson's Comparison of Electoral Systems Myerson (1993) uses mechanism design to compare electoral systems based on their resistance to corruption. **Setup:** Candidates can be "corrupt" (extracting rents if elected) or "honest." Voters prefer honest candidates but may be uncertain about candidate types. An electoral system is a mapping from votes to winning probabilities. Systems differ in how effectively they allow voters to coordinate on punishing corrupt candidates. **Results:** Systems that allow voters to transfer support to "approved" alternatives better discipline corrupt politicians. Approval voting (which lets voters approve multiple candidates) and the Borda count (which assigns points based on ranked preferences) both perform well. Plurality rule (a single vote for one candidate) is more vulnerable to corruption because voters cannot coordinate against a corrupt frontrunner without first coordinating on a specific alternative. ### 8.5.5 Personal Vote and Electoral Rules Carey and Shugart (1995) analyze how electoral rules affect politicians' incentives to cultivate a "personal vote" based on individual characteristics and constituency service, versus a "party vote" based on party label. Several features of electoral rules increase personal vote incentives. Ballot structure matters: when voters can choose among candidates of the same party (open lists, primaries), personal reputation becomes valuable. Vote pooling also matters: when votes for a candidate help only that candidate (as in single-member districts) rather than the party as a whole (as in closed-list PR), personal effort is rewarded. Larger district magnitude increases intra-party competition in open systems. And weak party control over nominations further increases personal vote incentives. These incentives have consequences. Where personal vote incentives are strong, legislators invest in constituency service and pork-barrel politics, run candidate-centered rather than programmatic campaigns, and exhibit weaker party discipline in the legislature. Where personal vote incentives are weak (as under closed-list PR), parties and programmatic politics are strengthened. ### 8.5.6 Tsebelis: Veto Players Tsebelis (2002) provides a unified spatial framework for analyzing how constitutional structures affect policy change. {% hint style="info" %} **Veto Player.** An actor whose agreement is required to change the status quo. Veto players may be *institutional* (required by constitution: president, upper house, constitutional court) or *partisan* (required by political practice: coalition partners in government). {% endhint %} The framework yields three key results. Policy stability increases with the number of veto players, since more players means more potential blockers of change. Policy stability also increases with the ideological distance between veto players—even with few veto players, wide disagreement makes change difficult. And policy stability increases with the internal cohesion of each veto player, since a unified actor is harder to circumvent than a divided one. The framework's unifying power is considerable: it brings presidentialism versus parliamentarism, bicameralism, federalism, and coalition government under a single logic. A presidential system with divided government has multiple veto players (president, House, Senate); a unified parliamentary majority has one. The implications for policy follow directly. Gridlock is more likely with many veto players, and policy is more resistant to shocks. Whether this stability is desirable depends on context—entrenching good policy is valuable, but blocking needed reforms is not. *** ## 8.6 Empirical Evidence ### 8.6.1 Electoral Systems and Government Spending Persson and Tabellini (2003) test their theoretical predictions using cross-country data. **Data:** 85 democracies, 1960-1998. Their main findings support the theoretical predictions. Majoritarian systems are associated with lower government spending (about 5% of GDP less) than PR systems, controlling for income, population, trade openness, and other factors. The composition of spending also differs: majoritarian systems spend less on broad social transfers (welfare, social security), while PR systems spend relatively more on transfers and less on geographically targeted programs. PR systems are also associated with larger deficits, possibly because coalition governments face common-pool problems. These results come with important caveats. Cross-country regressions face severe identification challenges. Electoral systems are not randomly assigned; they reflect political history and preferences. Unobserved factors correlated with electoral system choice may drive spending patterns. ### 8.6.2 Natural Experiments and Reform Episodes France switched from plurality to PR for one election (1986), then back. Blais and Carty (1991) document increased party fragmentation under PR, consistent with theory, though policy consequences are harder to isolate given the brief episode. New Zealand switched from plurality to mixed-member PR in 1996; post-reform governments are more likely to be coalitions, policy has shifted toward broader social spending, and minor parties now achieve representation. Italy moved from pure PR to a mixed system emphasizing plurality in 1993, then back toward PR; the reforms aimed to reduce party fragmentation and corruption, and evidence suggests the plurality phase did consolidate parties, though corruption proved resistant. Japan replaced single non-transferable vote (SNTV) with a mixed system in 1994. SNTV had encouraged intra-party competition and personal vote seeking (associated with factionalism and corruption), and post-reform party cohesion increased. ### 8.6.3 Presidentialism and Instability Przeworski et al. (2000), analyzing regime survival across 1950-1990, found that presidential democracies were more likely to collapse than parliamentary ones, controlling for income. The effect was concentrated in Latin America and weakened with additional controls. Cheibub (2007) reanalyzed the data and argued that the presidential effect operates through military inheritance: where military regimes transition to democracy, they tend to choose presidential constitutions, and the observed instability reflects military influence rather than presidentialism per se. More recent evidence from Svolik (2015) suggests that executive constraints (whether institutional or informal) predict democratic stability better than the presidential/parliamentary distinction. What matters is whether checks on executive power are effective, not the formal institutional label. ### 8.6.4 Federalism and Economic Performance Cross-country evidence shows that federalism is associated with higher income, but the relationship is confounded by the fact that large, economically successful countries tend to be federal (U.S., Germany, Switzerland). Within-country evidence is more informative but mixed: studies of fiscal decentralization in China and Indonesia's "big bang" decentralization in 2001 find improved public service delivery in some cases and increased corruption or inequality in others. Weingast (1995) offers a theoretical lens, arguing that "market-preserving federalism" promotes growth when subnational governments face hard budget constraints and compete for mobile capital. China's fiscal decentralization in the 1980s and 1990s may fit this model, but soft budget constraints (bailouts of subnational governments) undermine the mechanism. *** ## 8.7 Constitutional Design and Commitment ### 8.7.1 The Commitment Problem Constitutional rules are valuable only if they constrain future behavior. But what prevents future majorities from ignoring or changing inconvenient rules? One answer is self-enforcement: a constitution is self-enforcing if actors find it in their interest to obey, given their expectations about others' behavior. Weingast (1997) models constitutions as coordination equilibria—if citizens coordinate to punish violations, officials obey; if citizens are uncoordinated, officials may transgress. A second mechanism is judicial review: courts can enforce constitutional limits by striking down unconstitutional actions. But courts have no army, and their power depends on other actors accepting their authority. Judicial independence is itself a constitutional commitment that must be self-enforcing. A third mechanism is political culture: long-standing democracies develop informal norms that complement formal rules, and actors obey rules even when violation would be advantageous because they value the rules themselves or fear reputational costs. ### 8.7.2 Hadfield and Weingast: Coordination on Constitutions Hadfield and Weingast (2012) model constitutional order as a coordination problem. **Setup:** A ruler can choose to comply with constitutional limits or transgress. Citizens can coordinate to oppose transgressions (costly but effective) or fail to coordinate. If citizens are certain to coordinate against transgressions, the ruler complies. If citizens fail to coordinate, the ruler transgresses. Constitutions work by providing "focal points" for coordination—clear, public rules that citizens can monitor and rally around. Ambiguous rules are harder to enforce because citizens disagree about whether violations occurred. The implications are far-reaching. Clear rules are easier to enforce than vague standards, and public knowledge of those rules is essential for coordination. Incremental violations may escape punishment when each individual transgression seems minor, suggesting that "bright-line" rules (clear prohibitions) may be more robust than "balancing tests" that require judgment about degree. ### 8.7.3 Democratic Backsliding and Constitutional Erosion Recent experience with democratic backsliding highlights the fragility of constitutional commitment: Several recent cases illustrate these dynamics. In Hungary (2010 onward), Fidesz won a two-thirds supermajority and used it to rewrite the constitution, pack the courts, control media, and manipulate electoral rules—all through formally legal means. In Poland (2015 onward), PiS undermined judicial independence through legislation that superficially complied with constitutional forms while gutting substantive independence. In Turkey during the 2010s, Erdogan used emergency powers, constitutional referenda, and judicial appointments to concentrate power. These cases share a common pattern: gradual erosion rather than sudden coup, legal formalism that maintains the appearance of constitutional compliance, constitutional amendments that remove constraints, and attacks on independent institutions such as courts, media, and electoral commissions. They suggest that constitutional commitment depends on broader political conditions—opposition strength, civil society, international pressure—not just constitutional text. *** ## 8.8 Formal Model: Electoral Systems and Policy We develop a simplified model illustrating how electoral systems affect policy outcomes. ### 8.8.1 Setup A continuum of voters with mass 1 is distributed across 3 regions of equal size. Voters care about public goods $$g$$ and regional transfers $$f\_j$$: $$ u\_{ij} = \alpha \ln(g) + f\_j $$ where $$i$$ indexes individual voters and $$j \in {1, 2, 3}$$ indexes regions. Tax revenue is $$T$$, and the budget constraint is: $$ g + \sum\_{j=1}^{3} f\_j = T $$ Two parties compete by committing to policies $$(g, f\_1, f\_2, f\_3)$$. ### 8.8.2 Proportional Representation Under PR, the party winning the most votes nationwide forms government. Each voter votes for the party offering higher utility. A party maximizes votes by maximizing average utility: $$ \max\_{g, f\_j} \quad \frac{1}{3}\sum\_{j=1}^{3}\left\[\alpha \ln(g) + f\_j\right] $$ $$ \text{s.t.} \quad g + \sum\_j f\_j = T $$ The Lagrangian is: $$ \mathcal{L} = \frac{1}{3}\sum\_{j=1}^{3}\left\[\alpha \ln(g) + f\_j\right] + \lambda\left(T - g - \sum\_j f\_j\right) $$ Taking FOCs with respect to $$g$$ and $$f\_j$$: $$ \frac{\partial \mathcal{L}}{\partial g}: \quad \frac{\alpha}{g} = \lambda $$ $$ \frac{\partial \mathcal{L}}{\partial f\_j}: \quad \frac{1}{3} = \lambda $$ From the second condition, $$\lambda = \frac{1}{3}$$. Substituting into the first: $$g^{PR} = 3\alpha$$. Transfers: $$f\_j^{PR} = \frac{T - 3\alpha}{3}$$ for all $$j$$ (uniform distribution). ### 8.8.3 Majoritarian System Under majority rule with single-member districts, the party winning 2 of 3 districts forms government. Assume one district is "safe" for each party (no competition). The election is decided in the one "swing" district. Parties maximize utility of the swing district's voters: $$ \max\_{g, f\_j} \quad \alpha \ln(g) + f\_{swing} $$ $$ \text{s.t.} \quad g + \sum\_j f\_j = T $$ The party sets $$f\_j = 0$$ for non-swing districts and allocates all transfer resources to the swing district: $$ g + f\_{swing} = T $$ In a pure public-good-vs-transfers setup with identical preferences, this yields the same $$g$$. The key distinction arises when we introduce *targeting efficiency*. Suppose the policy choice is between a broad public good with efficiency $$\eta > 1$$ (where spending one dollar produces $$\eta$$ dollars of benefit distributed equally to all $$n$$ citizens) and targeted transfers (where spending one dollar produces one dollar of benefit to the recipient only). Under PR, all voters matter equally, so parties internalize the full social benefit of public goods. A party maximizes: $$ \max\_{g} \quad n \cdot \frac{\eta g}{n} + (T - g) = \eta g + T - g $$ Since $$\eta > 1$$, the party sets $$g^{PR} = T$$: all revenue goes to public goods. Under majoritarian rules, the party needs only a majority in the swing district. If the swing district contains a fraction $$1/J$$ of the population and the party needs 50% of that district, it can win by targeting transfers to a fraction $$1/(2J)$$ of the population. The party compares the electoral benefit of spending one dollar on public goods ($$\eta / n$$ per voter, spread across all voters) against targeting that dollar as a transfer to a swing-district voter (worth $$1$$ to the recipient). The party prefers public goods only if: $$ \frac{\eta}{n} > 1 \quad \Rightarrow \quad \eta > n $$ Since $$n$$ is large but $$\eta$$ is typically modest, majoritarian systems favor targeted transfers: $$g^{Maj} < g^{PR}$$. **Result:** Majoritarian systems are more likely to choose targeted transfers over efficient public goods, while PR systems tilt toward broad public goods provision. This is the core Lizzeri-Persico (2001) and Persson-Tabellini (2000) insight. ### 8.8.4 Extensions If districts differ in the proportion of "swing" versus "partisan" voters, a swing voter extension predicts that parties target districts with many swing voters—the most electorally responsive—leading to geographic concentration of spending in competitive areas. The model can also be extended to coalition government under PR: coalition partners may each demand transfers to their constituencies, creating a common pool problem that generates higher total spending and larger deficits. *** ## 8.9 Research Frontier ### 8.9.1 Endogenous Institutions Most research treats institutions as exogenous. But institutions are chosen, often by actors who anticipate their effects. Recent work endogenizes institutional choice: Acemoglu and Robinson (2006, 2012) model institutions as emerging from conflict between groups with different interests; institutional change occurs when relative power shifts through economic change, external shocks, or collective action. Aghion, Alesina, and Trebbi (2004) model constitutional choice as a tradeoff between "insulation" (protecting policy from future majorities) and "responsiveness" (allowing policy to reflect changing preferences), showing that groups uncertain about their future power prefer more rigid institutions. The empirical challenges remain steep: testing theories of institutional origins requires identifying exogenous variation in the determinants of institutional choice. ### 8.9.2 Electoral System Reform Several democracies have reformed electoral systems, providing quasi-experimental evidence. New Zealand switched from plurality to MMP in 1996. Italy has undergone various reforms since 1993. Japan moved from SNTV to a mixed system in 1994. The UK held a referendum on the alternative vote in 2011, which voters rejected. These reforms allow before/after comparisons, though the political circumstances driving reform may confound the estimated effects. ### 8.9.3 Representation and Descriptive Representation Does it matter whether legislators "look like" their constituents? Research on descriptive representation examines quotas for women and minorities, the effects of descriptive representation on policy, and voter responses to candidate identity. Recent work uses regression discontinuity designs in close elections to identify causal effects of electing minority or female candidates. ### 8.9.4 Digital Democracy and Institutional Innovation Technological change creates possibilities for institutional innovation. Researchers are investigating whether electronic voting changes participation, whether digital forums can enhance democratic deliberation, whether "liquid democracy"—allowing voters to delegate votes to others on specific issues—is feasible, and whether algorithmic redistricting can reduce gerrymandering. These remain frontier areas with limited evidence but active experimentation. *** ## Further Reading ### Foundational * Lijphart, Arend. 1999. *Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries*. Yale University Press. * Duverger, Maurice. 1954. *Political Parties: Their Organization and Activity in the Modern State*. Wiley. * Cox, Gary W. 1997. *Making Votes Count: Strategic Coordination in the World's Electoral Systems*. Cambridge University Press. ### Intermediate * Persson, Torsten, and Guido Tabellini. 2003. *The Economic Effects of Constitutions*. MIT Press. * Cheibub, José Antonio. 2007. *Presidentialism, Parliamentarism, and Democracy*. Cambridge University Press. * Carey, John M., and Matthew S. Shugart. 1995. "Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas." *Electoral Studies* 14(4): 417-439. ### Frontier * Aghion, Philippe, Alberto Alesina, and Francesco Trebbi. 2004. "Endogenous Political Institutions." *Quarterly Journal of Economics* 119(2): 565-611. * Hadfield, Gillian K., and Barry R. Weingast. 2012. "What Is Law? A Coordination Model of the Characteristics of Legal Order." *Journal of Legal Analysis* 4(2): 471-514. * Fujiwara, Thomas. 2015. "Voting Technology, Political Responsiveness, and Infant Health: Evidence from Brazil." *Econometrica* 83(2): 423-464. * Svolik, Milan W. 2019. "Polarization versus Democracy." *Journal of Democracy* 30(3): 20-32. * Levitsky, Steven, and Daniel Ziblatt. 2018. *How Democracies Die*. Crown. * Ginsburg, Tom, and Aziz Z. Huq. 2018. *How to Save a Constitutional Democracy*. University of Chicago Press. * Acemoglu, Daron, and James A. Robinson. 2019. *The Narrow Corridor: States, Societies, and the Fate of Liberty*. Penguin Press. *** ## Exercises 1. **Duverger's law in practice:** Choose a country using plurality elections. Examine the effective number of parties at national and district levels. Is the pattern consistent with Duverger's law? Are there exceptions, and if so, why? 2. **Electoral reform analysis:** New Zealand switched from plurality to MMP in 1996. Research the reform: Why was it adopted? How did government composition change? Is there evidence of policy changes predicted by theory? 3. **Federalism tradeoffs:** Consider either the U.S. or Germany. Identify one policy area where federal structure creates coordination problems (e.g., environmental regulation, education). What institutional mechanisms address these problems? How well do they work? 4. **Presidential instability:** Linz argues presidential systems are prone to crisis. Choose a case of presidential system crisis (e.g., Brazil's impeachments, Peru's constitutional conflicts). Does the case support Linz's mechanisms (dual legitimacy, rigidity, winner-take-all)? Or does Cheibub's critique apply? 5. **Model extension:** Extend the model in Section 8.8 to include a fourth region that is "safe" for one party. How does this change the allocation of transfers under majoritarian rule? What does this predict about spending patterns in countries with dominant regional parties? 6. **Constitutional commitment:** Hungary and Poland have experienced democratic backsliding despite EU membership and constitutional courts. What features of their constitutions or political situations enabled backsliding? What institutional reforms might have prevented it? 7. **Comparative application:** Compare the electoral systems of the UK and Germany. Using the theoretical frameworks from Section 8.5, predict how the systems should differ in: (a) number of parties, (b) government composition, (c) spending levels, (d) spending composition. Then examine the evidence. Where does theory succeed or fail? 8. **Formal model exercise:** In the Persson-Tabellini framework (Section 8.5.1), suppose voters differ in their valuation of public goods: high types have $$H(g) = 2\sqrt{g}$$, low types have $$H(g) = \sqrt{g}$$. If high types are concentrated in one region under majoritarian elections, how does this affect equilibrium policy? Compare with PR where types are distributed uniformly across regions. *** *Constitutions and electoral systems set the rules of the game; the next question is how those rules play out within legislatures themselves. Chapter 9 turns to legislative bargaining, agenda control, and the internal organization of legislatures—the institutions through which constitutional rules are translated into actual policy.*