Prepared for the volume “Explaining the Development Gap Between the United States and Latin America,” edited by Francis Fukuyama (forthcoming, 2007).

One of the most obvious ways that development in the United States has differed from that of Latin America concerns political institutions. The United States is, of course, the world’s oldest continuously existing democracy, the country that in many respects invented modern democracy. While Latin America is on the whole more democratic than other parts of the world including fast-developing Asia (see Fukuyama and Marwah 2000), no Latin American country has ever had an uninterrupted history of democratic rule, and the deviations in the region, in terms of authoritarian government, suppression of human rights, civil conflict and violence, have frequently been severe. Democracy and rule of law are ends in themselves (see Sen 2001), and they are also obviously related to a society’s ability to achieve other objectives like economic growth, social equity, and political inclusion. Understanding how and why political institutions differ between the US and Latin America is, then, key to understanding the larger puzzle of the latter region’s lagging development.

What is it about the political institutions of the United States that differs from those found in Latin America? Is there some clause or provision in the US Constitution that does not exist in other parts of the Western hemisphere, that accounts for the difference in the quality of government? And if so, could we fix the problem of democratic governance by fixing institutions at this level?

The answer to this question is, of course, very unlikely to be yes. Many Latin American countries deliberately modeled their political orders on that of the United States; while they differ in details there is enough variance throughout the region that if the problem lay only in the design of formal institutions, countries would have fixed whatever didn’t work over time, and evolved towards more efficient ones.

This suggests that there is another, deeper problem in Latin America that explains the political development gap. Formal political institutions in the United States were deliberately designed to inhibit strong political action and to limit the power of government, and yet they have not prevented the United States from acting decisively at times of national crisis or need. Within America’s relatively decentralized institutional structure, the society has been able to generate consensus at key junctures for reasons that have to do more with a political culture that has prized consensus and compromise even as it promoted competition. This consensus broke down and could not be mediated by the political system on one important occasion, the American Civil War, when the issue of slavery had to be settled by violence. But at all other times, social conflicts have been mediated within the constitutional framework originally laid out by the American founding fathers.

The same has not been the case in Latin America. The region’s social conflicts have been more severe, and the ability of social groups to use formal political institutions to resolve, mediate, or mitigate them has been much less effective. While reform of formal institutions can ameliorate certain types of political dysfunctions (e.g., the problem of fiscal federalism in Argentina or Brazil), the root causes of political instability and weak governance are likely to exist at the deeper levels of social structure or political culture.

Indeed, I will argue in this paper that institutions matter much less than many people think, if by institutions one means formal, visible macropolitical rules defined by constitutions and law – presidencies, electoral systems, federalism, and the like. While we can identify certain clearly bad or defective institutional arrangements, there is no such thing as an optimal political institution. Vastly different types of formal institutions can be made to “work,” while the best institutions will fail without the proper supporting structures in the surrounding society. Closing the institutional gap between the United States and Latin America will involve the much more challenging task of using politics to ameliorate some of the underlying social-structural problems, and moving to a different kind of political culture that takes consensus and rule of law more seriously.

Institutions Matter — or Don’t They?

In recent years, a great deal of attention has been paid to the importance of institutions for economic development, leading to a broad consensus that “institutions matter” (see World Bank 1997, 2002; Fukuyama 2004; Stein et. al. 2005). Institutions in this sense include things like property rights, credible enforcement of contracts, rule of law, mechanisms for conflict resolution, and the like. As a starting point for this paper, I take it for granted that this consensus on institutions is correct. There is by now a huge empirical literature supporting this case (see inter alia Knack and Keefer 1995; Acemoglu and Robinson 2000, 2002; Rodrik and Subramanian 2003; and Kaufmann and Kraay 2005; Chhibber et. al. 2006).

In recent years, the only systematic counterargument to the institutionalist one has been made by Jared Diamond, Jeffrey Sachs, and their followers (see Diamond 1997, Sachs and Warner 1995, and Sachs and McArthur 2001) who have pointed to the importance of material factors such as resource endownments, disease burdens, climate, and geographical location as determinants of economic growth. I believe that the institutionalists have won this argument hands down: Easterly and Levine (2002) and Acemoglu and Robinson (2002) have shown, for example, that resource endowments are important only as mediated through institutions, e.g., by providing more or less favorable conditions for the emergence or survival of certain types of institutions. The institutions themselves remain the proximate causes of growth, and in many cases can be shown to be exogenous to the material conditions under which a given society develops.

If one goes further back in history, however, one finds a different kind of anti-institutional argument that was much more popular in the years immediately following World War II, but which has not been articulated so clearly in recent years. American political science in the pre-war period focused heavily on legal studies and formal political institutions. But the collapse of democracy in the 1930s in the face of the twin totalitarian challenges of Communism and fascism convinced many observers that the exact specifications of formal institutions mattered much less than underlying structural conditions unconducive to democracy.

This type of anti-institutionalism was of course the longstanding position of most Marxists, who maintained that the state in a capitalist democracy was simply the “executive committee of the bourgeoisie,” reflective of underlying social forces and not an autonomous agent in the shaping of outcomes. In a curious way this belief in state as superstructure jibed with a certain American anti-statist penchant (see Nettl 1968) that saw the state simply as a mechanical processor of societal demands, without interests or a logic of its own. Much of the post-World War II research agenda focused not on the design of formal institutions, but on subjects like political culture or value systems (McClellan 1953; Almond 1963); in modernization theory, sociology and anthropology played as important a role as political science in explaining development outcomes.

Institutions and the autonomous state were re-inserted into the research agenda only in the 1970s and 80s, which came partly in response to recognition of the importance of state-directed development in regions like East Asia which could not simply be accounted for by structural or cultural models (see Nettl 1968, Mann 1984; Skocpol, Evans, et. al. 1985). In addition, the rise of Douglass North’s new institutional economics gave a new legitimacy to the study of institutions in economics, as well as a powerful set of new conceptual tools. These converging streams led to a bursting of the dykes and the emergence of a large literature not just on the general question of state autonomy, but also a rich and contextualized literature on institutional design that continues up to the present. The debate on presidentialism started by Linz (1990), discussed in much greater depth below, was one important example of the new focus on institutions and institutional design in this tradition.

The new institutional economics muddied the waters in one important way, however, which has led to confusion in subsequent thinking about institutions. North (1990) defines institutions as “the humanly devised constraints that shape human interaction,” whether formal or informal. The eliding of formal and informal institutions made North’s definition conceptually robust and useful as a means of attacking the institution-less premises of earlier neoclassical economics. But the older anti-institutionalist position had been built around the distinction between formal and informal institutions: it argued that formal legal structures at best depended on and at worst were undermined by factors like political culture, discordant value systems, or social structures that gave rise to political preferences at odds with the institutional structure. There was also a very important practical difference between formal and informal institutions. Formal institutions can be established, abolished, or changed literally with the stroke of a pen. They were thus the typical objects of public policy to be manipulated at will. Informal institutions, by contrast, reflected embedded social practices that were often hard to perceive and measure, and even harder to manipulate through the usual levers provided by public policy. It is one thing, for example, to mandate certain terms and appointment rules for supreme court justices, but if politicians routinely fail to follow them because packing the supreme court has become a normative behavior for an entire political elite, then one has left the realm of conventional public policy. The new institutionalism of course recognized the importance of informal institutions, but oftentimes failed to separate them out clearly as a distinct conceptual category subject to a very different evolutionary dynamics.

Today, there are relatively few scholars who continue to insist on the importance of the distinction between formal and informal institutions, and the priority of the latter over the former. ¹ One is Samuel Huntington, who in recent years has been making a larger argument in favor of the centrality of culture in shaping political outcomes. His book Who Are We? (2004) points to the importance of culture in the shaping of American national identity and in the success of American democracy. One could not have a clearer statement of the older anti-institutionalist position than the following (p. 59):

Would America be the America it is today if in the seventeenth and eighteenth centuries it had been settled not by British Protestants but by French, Spanish, or Portuguese Catholics? The answer is no. It would not be America; it would be Quebec, Mexico, or Brazil.

While Huntington has been severely criticized for his policy prescriptions regarding Mexican immigration. But his broader point that American national identity has not simply been a political one based defined by institutions like the Constitution and system of laws but also rooted in certain religious and cultural traditions – what he labels Anglo-Protestantism – would seem to be incontrovertible has an historical fact. It is an interpretation of American society that was shared by observers from Tocqueville to Bryce to Lipset, and constitutes one long-standing answer to the question of why Latin America, which on independence from Spain or Portugal modeled many political institutions explicitly on those of the United States, failed to achieve North American levels of either growth or political stability.

I do not believe that political culture determines political or economic outcomes. Culture changes over time, and is shaped by formal institutions even as it shapes them. Formal institutions matter; they change incentives, mould preferences, and solve (or fail to resolve) collective action problems. On the other hand, the informal matrix of norms, beliefs, values, traditions, and habits that constitute a society are critical for the proper functioning of formal institutions, and a political science that pays attention only to the design of formal institutions and fails to understand normative and cultural factors will fail. A large cultural variable like Catholicism may not be very helpful in explaining or predicting political behavior or institutional development, but a norm that assumes that bureaucratic appointments ought to favor friends and relatives over people with formal credentials might. One must therefore look both to formal and informal institutions in explaining the difference in development outcomes between different societies, taking each side of the equation seriously, as well as the importance of their interaction.

The Perils of Presidentialism and After

I will illustrate the complexity of the problem of specifying the nature of good formal political institutions by recapping the history of the evolution of the debate over institutional design that has taken place over the last two decades. This debate was initiated by Juan Linz in his article “The Perils of Presidentialism” (1990), in which he argued that political instability in Latin America was due to the fact that many of the democracies there were presidential rather than parliamentary, based on a North American model that did not work well in other parts of the hemisphere.²

Linz pointed to four basic problems with presidential systems. First, presidential systems were inherently majoritarian, which led to the possibility that a president would be elected by a slim plurality of the population and therefore lack legitimacy. Second, presidential systems had rigid terms and did not provide easy mechanisms for removing a president who had lost legitimacy after being elected. Parliamentary systems dealt with this problem through no-confidence votes; impeachment was the messy alternative in presidential ones. Moreover, term-limited presidents often spent a great deal of time and political capital figuring out how to add terms to their tenure. The third problem had to do with dual legitimacy. In a presidential system, both the executive and the legislature are directly elected and thus have separate sources of legitimacy; since they survive separately, there is always the possibility of gridlock and political paralysis when the two branches are controlled by different parties. And finally, presidential systems tended to personalize politics, emphasizing the character and foibles of the president rather than the broad program of a political party.

Linz’s critique of presidentialism was based on one prominent case, the election in 1971 of the socialist Salvador Allende as president of Chile. Allende received a mere 37 percent of the popular vote, but nonetheless proceeded to take this as a mandate to initiate a series of radical economic policies like nationalization of the banking system, provoking an economic crisis and then a coup by General Augusto Pinochet. Linz pointed out that in a parliamentary system, Allende would have been forced into a coalition with the Christian Democrats, which would have constrained his ability to change economic policy as he did.

Although he did not address this issue explicitly, Linz was concerned with two separate design goals that were in some sense at cross purposes. On the one hand, he was concerned with effectiveness of democratic decision-making, and was thus concerned that the dual legitimacy of presidential systems would lead to executive-legislative deadlock. On the other hand, he was also concerned with legitimacy, and the possibility that an executive would receive support from a relatively small minority of the population. One can see immediately that the mutual checking of the two branches is actually an advantage with respect to legitimacy, while the plurality election of a president may be an advantage with regard to effectiveness. What was difficult was to optimize both effectiveness and legitimacy simultaneously.

Linz’s critique of presidentialism was immediately attacked by Lijphart (1991), Horowitz (1990), Shugart and Carey (1997), and a variety of other authors. These critics pointed out that parliamentary systems could be as weak and illegitimate as presidential ones; indeed, some like the interwar Weimar Republic or the French Fourth Republic became illegitimate because of their weakness. Parliamentary systems required strong political parties; while party discipline could to some extent be engineered, party fragmentation was also based on the religious, ethnic, class, and geographical structure of the underlying society. Presidentialism by contrast had certain advantages: voters knew exactly whom they were electing, and that official remained directly accountable to voters, in contrast to parliamentary systems where parties or coalitions of parties could remove chief executives without any change in the popular mandate. The inherent majoritarianism of presidential systems could, moreover, be tempered by requirements for second-round runoffs or, as in Nigeria or Sri Lanka, requirements that the president receive pluralities in multiple electoral districts.

Lijphart (1991) early on argued that a single design axis like the nature of executive power could not be understood in isolation from other aspects of the political system; electoral systems, in particular, were critical in determining the overall effectiveness of a political system. He suggested the following matrix for characterizing combinations of executive and electoral systems:

Lijphart argued that presidential systems coupled with single-member plurality systems like that of the United States tended to produce two relatively strong and cohesive parties (Duverger’s Law; Duverger 1950). While the possibility of different parties gaining control of the executive and legislative branches has in fact been a reality for much of recent American history, this has not led necessarily to gridlock because politics was still organized around two relatively coherent competing ideological points of view. From his standpoint, the worst combination was presidentialism together with proportional representation in the legislature, which he argued characterized many political systems in Latin America. This led not just to gridlock, but to presidents having to bargain with disorganized and fragmented parties – the worst features of both parliamentary and presidential systems.
A more general system for categorizing political systems

It soon became clear that presidentialism interacted not just with the electoral system, but with virtually all other aspects of the political system. Cox and McCubbins (in Haggard and McCubbins 2001) introduced the general concept of veto gates: that is, actors within the political system that have the power to stop or modify legislation or policy. All political systems can be arrayed on a continuum from a perfect authoritarianism that had only one veto gate (the dictator’s will), to a perfect consensual democracy in which all citizens had to agree to a policy. The concept of veto gates in a sense reprises the conceptual framework laid out by Buchanan and Tullock (1962) to explain the principle of majority voting, where they posited a clear tradeoff between legitimacy and effectiveness (see Figure 2).

The more members of a society that participate in a decision, the higher are the expected decision costs; for large societies the costs rise exponentially as one approaches consensus. Buchanan and Tullock argued that the principle of majority voting had no inherent normative logic; one could choose any point on the curve in Figure 2 as an appropriate tradeoff between effectiveness and legitimacy and in the case of constitutional law supermajorities were indeed often required. In the case of monetary policy, by contrast, many democracies delegated decision rights to an independent central bank with a very small number of decision-makers.

The concept of veto gates as used by Cox and McCubbins does not refer to individual voters, but to organized institutions within the political system that, through delegation, have veto rights in political decision-making. It becomes clear that legislatures and the rules under which they are elected are only one of several possible veto gates. These include:

• The electoral system. Proportional representation usually increases the number of veto gates over plurality systems because it leads to a more fragmented party system. Small district size (especially when combined with PR) tends to increase fragmentation, as do electoral cycles not synchronized with presidential ones. PR systems with thresholds reduce fragmentation.
• Party discipline. Strong parties (e.g., those operating under closed-list systems) are better able to make decisions than ones with weak discipline.
• Bicameralism. An upper house adds another veto gate, often based on territorial criteria and/or electoral rules different from those in the lower house.
• Federalism and decentralization. Federalism delegates important decision rights to subunits like states, which can be further delegated to even smaller subunits like municipalities and districts.
• Independent judiciaries. If courts are truly independent and have powers of constitutional review, as in the United States, they can constitute a major check on the powers of the other two branches. Courts can intervene in different ways as well, from merely interpreting legislative intent to initiating policies on their own.

From the above it should be clear that political systems need to be categorized not by Lijphart’s 2×2 matrix, but by an n-dimensional matrix that would array all of these design axes against one another. It is possible to come up with a very large number of combinations of design features that will add or subtract veto gates from the political system, thus shifting the balance between effectiveness and legitimacy.

Since it is not practical to graphically portray an n-dimensional matrix, we will have to substitute a continuum like that on the x-axis of Figure 2 instead as a means of ranking different political systems. Political systems with different types of veto gates are hard to compare in the abstract: Is a system with weak party discipline but no federalism and an independent constitutional court stronger or weaker than one with federalism, cohesive parties, and a somewhat politicized court? Is the premier-presidential system of the French Fifth Republic, whose president does not appoint the cabinet but which has clear reserved powers in foreign and defense policy and does not have to devolve powers to federal subunits, stronger or weaker than the American presidential system whose president appoints the cabinet and shares powers in foreign affairs?

Taking these complexities into account, it is nonetheless possible to do a rough rank-ordering of different political systems in terms of their aggregate number of veto players.

1. Classical Westminster (New Zealand before 1994)
2. Parliamentary/PR with cohesive parties, no federalism (Austria, Belgium, Netherlands, Thailand)
3. Premier-presidential, no federalism (French 5th Republic, Finland)
4. Presidential with plurality voting, with federalism (US, Philippines)
5. Parliamentary with fragmented parties (French 4th Republic/Italian republic pre-1994)
6. Presidential with PR and fragmented parties (Colombia, Brazil)

Of all democratic systems, the classical Westminster system has by far the fewest veto gates and is capable of the most decisive action. Such a system in its pure form is parliamentary with a plurality voting system and party discipline, leading to exaggerated majorities in parliament; there is no federalism or decentralization; no written constitution and therefore no requirements for supermajority voting; and no judicial review.³ A simple majority in parliament (which, given the electoral system, can represent less than a majority of the popular vote) is sufficient to change any law in the land, and leads to what some have described as a democratic dictatorship. In the 2001 British general election, for example, the Labour Party received only 42 percent of the popular vote, and yet received 62.5 percent of the seats in Parliament, while the Liberal Democrats received almost 19 percent of the popular vote and got only 8 percent of the seats.

The American system, by contrast, is deliberately designed to place many more veto gates – what Americans call checks and balances – in front of executive decision-making, by adding separated powers, bicameralism, federalism, weak party discipline, and judicial review. The only important feature of the US political system that increases rather than decreases decisiveness is its single-member plurality voting system. A British Prime Minister’s budget is approved within days of its being submitted to parliament; an American budget, by contrast, takes the better part of a year to pass, and never survives in the form proposed by the president.

While there are clear differences between the Westminster system and the US system, and between either of them and those that prevail in Argentina or Peru, the actual behavior of political systems may not correspond to a simple quantitative tabulation of veto gates. As will be seen below, even the notional ranking of systems above can be very misleading. Within the broad categories of veto gates given above, there are countless other rules that affect the ability of political systems to generate decisions or enforce policies. Sometimes these rules are formal, but are so detailed that they are invisible to most outside observers (e.g., World Bank country directors or North American academic specialists). At other times they are informal, and intrude into the realm of political culture that will be explored further in the next section. Let me give an example concerning legislative coherence.

Legislative coherence is the ability of legislatures to pass legislation, hopefully, legislation that is public-regarding rather than patronage-based and/or clientelistic. Legislative coherence is the product of the interplay of various institutional design features, such as the electoral system (usually held to be the prime determinant), party discipline, rules concerning executive-legislative interaction (i.e., presidentialism and which branch controls the legislative agenda), and the party system which reflects the underlying structural conditions of the society.

Political scientists associate legislative incoherence with the following factors: PR systems, particularly those with open list voting and no minimum thresholds; weak party discipline; and party systems that are not firmly anchored in important social groups or cleavages. By this account, Colombia and Brazil have traditionally been put forth as examples of weak legislative systems. The former has coherent parties but very weak party discipline, in which the parties were unable to control the use of their own party labels (Archer and Shugart in Mainwaring and Shugart 1997); Brazil has open-list PR, a traditionally weak party system, and supposedly weak party discipline (Mainwaring 1991; Mainwaring in Mainwaring and Shugart 1997). Argentina, by contrast, should have much greater legislative coherence, since it has a closed-list PR system and relatively coherent parties (Corrales 2002).

The actuality is rather different. Colombian presidents have in fact had to work with incoherent legislatures that have demanded particularistic payoffs in return for votes, as the theory predicts. Major reforms have required Colombian presidents to resort either to emergency powers, or to maneuvers of questionable legality.⁴ The Brazilian congress, on the other hand, has actually been able to pass a large volume of legislation since ratification of the 1988 constitution; while presidents have never had legislative majorities of their own party, they have nonetheless been able to put together coalitions of parties with relatively strong party discipline in support of far-reaching reforms like the Fiscal Responsibility Law of 2000 which restricted the ability of Brazilian states to run budget deficits (Figueiredo and Limongi 2000). Argentina, on the other hand, has suffered from legislative incoherence despite the fact that presidents from both the Partido Justicialista (the PJ, or Peronist party) and the Allianza have had strong legislative majorities/pluralities in Congress.

The reason for this has to do with the fact that legislative coherence is the product of the interplay of many more rules than the usual ones of open- or closed-list PR. Brazilian presidents, for example, through their control over fiscal policy and bureaucratic appointments have been able to discipline legislators and enforce party-line voting. That presidents Cardoso and Lula have used this power not to build clientelistic bases but to enact public-regarding reforms of fiscal federalism or the social security system may reflect changing political culture rather that formal institutions. Nonetheless, the result is contrary to the expectations of many political scientists and at odds with simple models of how formal rules correlate with policy outcomes.

In Argentina, by contrast, legislative coherence was undermined by the way that the electoral system interacted with federalism. While the national electoral system was closed-list PR, voting was done by province and it was not the national party but the provincial party chiefs that determined the voting lists. The national party was in fact not a cohesive bloc but an alliance of provincial fiefdoms (Tommasi 2004). In the years leading up to the economic crisis of 2001, the Argentine government could not maintain fiscal discipline because the PJ’s leader, president Carlos Menem, got into a spending duel with Eduardo Duhalde, governor of Buenos Aires province and Menem’s leading rival with the Peronist party. The entrenched power of provincial party bosses is also the reason why this system will be extremely difficult to reform. The coherence of the Argentine legislature, then, was more apparent than real, and goes far in explaining why Menem continued to rule by decree even when his party possessed a majority in Congress.

Why There Is No Optimal Political System

The groundwork has already been laid for the beginning of an explanation of why there can be no such thing as optimality in the design of political systems. Political systems seek conflicting social goods between which there is often a continuous tradeoff. The balance of goods that the system seeks to achieve will depend on a host of contextual factors like the society’s historical traditions and political culture, the external environment, economic conditions, and the like.

Cox and McCubbins describe this tradeoff as one between decisiveness and resoluteness. Systems with fewer veto gates produce more decisive political decisions. On the other hand, decisions that have to be vetted by more actors within the political system generally produce more lasting results, because there are fewer players interested in overturning the initial decision: hence greater resoluteness. It should be clear that the Cox-McCubbins tradeoff between decisiveness and resoluteness largely corresponds to the tradeoff between effectiveness and legitimacy described earlier. That is, the more members of a society who participate in a decision, the more legitimate it is (with perfect legitimacy being perfect consensus); the reason that legitimate decisions are resolute is because there are fewer interest groups or sectors of society opposed to the decision.

To complicate things a bit further, excessive resoluteness/legitimacy can sometimes undermine itself, while decisiveness on occasion becomes self-legitimating. That is, a democratic political system with excessive decision costs often times fails to produce policies of any sort, leading to voter disillusionment not just with the current administration, but with democracy as a whole. Societies may actually express a preference for the decisiveness of authoritarian governments that can cut through the miasma of ordinary politics and get things done.

What is the optimal political system for a developing country that seeks rapid economic growth? There was for a long time a strong bias on the part of a certain part of the academic and policy communities in favor of decisive over resolute/legitimate systems. One version of this was the “authoritarian transition,” advocated four decades ago by Samuel Huntington (1968) and re-articulated more recently by Huntington’s student Fareed Zakaria (2003). They argued not simply that authoritarian modernizers were more decisive, but that in many developing countries liberal autocracy alone was capable of supplying basic public goods like physical security and pubic order that were preconditions of development of any sort.

But even among those committed to development under democratic conditions, there was a pronounced bias in favor of decisiveness (see Sachs in Williamson 1994). The argument went something as follows. A typical developing country needed liberalizing economic reform, usually in the form of tariff reductions, deregulation, privatization, and reductions in consumer subsidies (i.e., the famous Washington Consensus). Often these policy reforms needed to be undertaken in the context of economic crisis such as a current account deficit leading to currency devaluation and high interest rates. While these reforms were expected to bring stability and long-term economic growth, they also produced a great deal of transitional pain as workers were laid off or consumers lost subsidies. Technocratic experts could see the long-term logic of these policies, but ordinary voters and politicians might not; therefore, a developing country in this position needed a decisive political system that would shield technocratic experts from populist demands and push through long-term public-regarding policies. A decisive system that implemented successful reforms would then become self-legitimating as it produced long-term stability and economic growth. This was supposedly the path followed by Chile under General Pinochet.

Political systems with excessive checks and balances, by contrast, slowed down decision-making and imposed many other decision costs as interest groups were paid off. What Geddes (1996) refers to as the “politician’s dilemma” is the phenomenon that reformist governments often need to pay such a high price to get reform that they end up undermining the goals of the reform that they are seeking to achieve. This price would presumably be lower in a more decisive political system.

Mexico under the presidency of Vicente Fox illustrates this problem. The Mexican political system is roughly similar to the American one, being presidential, federal, though with a mixed PR/plurality voting system. During the decades of dominance by the Partido Revolucionario Institucional (PRI), most observers thought of the Mexican system as a strong presidential one, since Mexican presidents often acted like authoritarian rulers. In retrospect, however, it is clear that this was simply the byproduct of PRI dominance of both the executive and legislative (and, indeed, the judicial) branches. Fox’s election in 2000 produced a president of a different party, the PAN, which did not command a majority in the Congress and which could not put together a coalition in favor of major reforms like modernization of the electricity or reform of the judicial system. The Mexican system, in other words, began behaving like a typical presidential system with checks and balance, biased towards resoluteness rather than decisiveness, and the result was political gridlock.

In the end, it is not clear that decisive political systems are preferable to resolute/legitimate ones from the standpoint of long-term development. Constitutional rules that amplify executive power by reducing veto gates can produce policies that come to be regarded as illegitimate; without a broader underlying social consensus, reforms are likely to be undermined over time. There have been a number of notable cases of this. The liberalizing reforms undertaken by Venezuelan president Carlos Andres Perez in 1989-90 provoked opposition not just within the broader Venezuelan society, but within Perez’s own Accion Democratica (Adeco) party (Corrales 2002). Perez used his office, supported by a small group of technocrats, to change policy, without making a broad-based effort to convince Venezuelan society of their necessity and logic. Even though the reforms produced economic growth, they were immediately undermined not just within the political system, but outside it as well (in the form of the military coups launched by Hugo Chavez).

The preference for decisive political systems, moreover, reflects a moment in history stretching from the mid-1980s through the late 1990s in which the development problem was seen as excessive state scope, and in which the posited solution was in almost all cases understood to be liberalizing economic reforms. In the early 21st century, the agenda has already begun to shift: left-leaning or outright populist presidents have come to power in Brazil, Ecuador, Argentina, Uruguay, Venezuela, and Bolivia. In many cases their agenda is the reassertion of state power, re-regulation of the economy and re-nationalization of certain economic sectors, and control over the media and civil society. Decisive political systems will only enhance the ability of populist presidents to enact bad economic policies and return their countries to state control and closed markets.

Institutions are only enabling devices; those that facilitate or encourage strong and decisive political decision-making are only as good as the policies being pursued. What inhibits the ambitions of a liberalizing reformer also checks the power of a would-be populist dictator. If decisive government were always preferable, then we should always want Westminster-type systems with their largely unchecked executive powers. But Americans have since the founding of their republic expressed a strong preference for a system of checks and balances that limits government power. This relatively non-decisive political system reflects the preferences of American political culture (Lipset 1995), which has always been distrustful of state power. Checks and balances make large-scale reform much more difficult, but in the long-run it also reduces the risks of the government being captured by politicians advocating policies that would not receive the support of the broader society. The preference for resolute/legitimate political systems over decisive ones can thus been seen as a preference for lower long-term political risk.

There is no optimal level of long-term political risk, and thus no optimal balance between decisiveness and resoluteness/legitimacy. The United States and Britain are among the world’s oldest and best-established democracies, and yet they have completely different political systems arising out of very different historical experiences. The largely unchecked Westminster system is a high-risk institutional arrangement that has worked reasonably well in the English-speaking world where it has been implemented. Margaret Thatcher’s reforms of the late 1970s-early 1980s could not have been carried out but for the exaggerated parliamentary majorities held by the Conservative Party. Similarly, Roger Douglas’ liberalizing reforms in the mid-1980s benefited from New Zealand’s even purer Westminster system.⁵ Both stand in sharp contrast to the situation faced by Chancellor Angela Merkel under Germany’s far less decisive institutional rules, which following the election of 2005 have forced her into a coalition and will sharply limit the kinds of liberalizing policies she will be able to put into place.

But a Westminster system would likely produce disastrous results if transported to a country with a different social structure and political culture (e.g., an ethnically fragmented society with a dominant ethnic group). Where it has been copied, it has been heavily modified to meet local conditions, as in the case of India the success of whose democracy would scarcely be conceivable in the absence of thoroughgoing federalism. The American presidential system, which is much less decisive than the Westminster system, has nonetheless been capable of achieving decisive action at certain critical junctures in American history. When American presidentialism was transplanted to Latin American countries, it worked only indifferently well. How each formal set of institutional rules plays out in practice is thus highly dependent on local social context, tradition, history, and the like.

The fact that we cannot specify an optimal set of formal institutions does not mean that we have no knowledge of the likely impact of changes to formal institutional rules. A number of institutional reforms, like central bank or judicial independence,⁶ have a clear logic and are broadly accepted as being desirable. Changes in electoral rules have broadly predictable effects, and there are a number of recent cases where electoral reform has produced desired results:

• Chile, which always had a coherent party system, has been operating since 1988 under an electoral system designed to force the country’s 4-5 large parties into two broad left-right coalitions, which has in fact happened.
• Japan changed its single non-transferable vote system to a mixture of single-member constituencies and PR in 1994. The SNV system forced parties to run multiple candidates in the same electoral district, which was blamed for the factionalism within the ruling Liberal Democratic Party. While it took over ten years to produce the desired effect, Prime Minister Joichiro Koizumi’s electoral victory in 2005 marked the demise of the faction system.
• Italy modified its low-threshold PR system that produced notoriously weak coalition governments to a mixed single-member/PR system in 1994, which had the desired effect of forcing parties into broad left-right coalitions. Italian politicians gamed the system, however, to ensure the survival of the smaller parties, and the system will revert back to a modified form of PR in 2006.
• New Zealand, which had a classical Westminster-style single-member plurality voting system in its parliament, changed over to a mixed-member proportional system in 1996. The result has been broader representation of smaller parties, together with relatively weak coalition governments in place of the two-and-a-half party system that prevailed earlier.
• In Thailand’s 1997 constitution, the electoral system was changed from a PR system that had produced weak coalitions, to a mixed system of 400 single-member constituency and 100 PR seats. The reform has given prime minister Thakskin Shinawatra an absolute legislative majority.

In addition, there is accumulating knowledge about the design of federal systems. Federalism has posed a problem for many large states in Latin America like Argentina, Brazil, and Mexico, because the subunits were delegated too much budgetary discretion and could run fiscal deficits. This differed from the situation in the United States, where most states are constitutionally prohibited from running budget deficits and face hard budget constraints based on their own ability to raise revenues. In Argentina and Brazil, by contrast, states could run deficits that would have to be covered by the federal government, a form of fiscal federalism that undermined overall budget discipline. In Argentina, the rules were particularly problematic because they were constantly being renegotiated; governors would spend a great deal of time politicking in Buenos Aires rather than raising their own tax revenues. The solution to this problem – putting states and other subunits under hard budget constraints – is relatively straightforward conceptually; what is hard is implementation, since such a reform means a de facto shift in power from the states to the federal government. Brazil, despite its supposedly weak political parties and strong federalism, moved in this direction with passage of the Fiscal Responsibility Law, while Argentina has failed to deal with this problem due to the entrenched power of state-level politicians. (See Stein, Tommasi et. al. 2006).

The fact that we can connect certain changes in institutional designs with certain behavioral outcomes does not mean that institutional change is easy to bring about; institutions are in fact very “sticky” or path dependent. The transaction costs of institutional change are often far greater than the transaction costs of weak or suboptimal institutions. Societies need to generate political will to bring about reform, and to prevent new institutions from being undermined by losers in the initial struggle.

Political Culture

As noted above, most conventional analyses of the formal structure of political institutions would have come to the conclusion that Brazil should produce weaker government than Argentina, given its open-list PR system, weak political parties, and entrenched federalism. And yet, Brazil weathered the period from 1990-2005 better than Argentina, avoiding the latter’s severe economic crisis in 2001-2002 and moving ahead with a series of structural reforms and public policies. And both countries have done much less well than neighboring Chile, not just over the past 15 years but for the preceding 15 as well. What accounts for these differences?

One factor is clearly leadership. Economists generally do not like to talk about independent variables like leadership because it amounts to throwing a massive random-number generator into their models. They prefer modeling institutions and hierarchy as endogenously, as the result of strategic interactions of individual agents who cannot achieve collective action without it. But leadership is more often than not exogenous. It was simply Argentina’s bad luck that Carlos Menem chose to throw away the positive legacy of his first term as president by seeking not just a second but a third term as well, leading him into a spending competition with Eduardo Duhalde. President Cardoso, by contrast, chose not to waste his political capital seeking ways to remain in office, but used it rather to try to solve some longstanding public policy problems. Many good development outcomes are thus attributable not to the structure of formal institutions, but rather to the emergence of the right leader at the right time.

The only way in which leadership may become a more tractable variable is when a certain leadership style is not simply the outgrowth of the foibles of a particular individual, but reflective of a broader political culture. Carlos Menem was widely blamed for packing the Argentine Supreme Court with his political cronies, but he was neither the first nor the last Argentine president to do so. Argentine elites have been notorious for avoiding or manipulating rules they find inconvenient, a behavioral tendency that shows up not just in Supreme Court appointments but in rates of tax compliance and ordinary corruption. These phenomena exist in Brazil and Chile as well, but the degree of disregard of law and rules seems simply to be lower, particularly in the latter case. A well-functioning rule of law is not simply a set of visible formal institutions like courts, bar associations, police, and judges. No formal arrangement of incentives will make such a system operate properly unless the participants share a certain normative respect for law and rules.

As noted earlier, there was an earlier critique of institutionalism that argued that formal institutions mattered less than variables like political culture and social structure in explaining political and development outcomes. This critique remains valid, but only if we understand political culture properly. To say that political culture is important is not necessarily to affirm the importance of certain large cultural categories like “Catholicism” or “Anglo-Protestantism.” Chile, Argentina, Ecuador, and Costa Rica are all predominantly Latin Catholic countries and ex-colonies of Spain, and yet they all have distinctive political cultures with respect to rule of law, something that can be seen clearly in their very different rankings on various indices measuring levels of corruption and governance. Political culture varies among groups and regions within societies, and over time; it is shaped not only by large symbolic forces like religion, but by shared historical experience like war or economic crisis. And it is key to understanding why certain formal political institutions do or do not work.

The United States as a Weak Form of Government

The importance of political culture can be seen clearly if we look closely at American political institutions and how they have functioned over the years. In the context of the analytical framework presented above, the United States has a relatively non-decisive set of formal institutions, meaning that the system has a large number of veto gates that bias it towards resoluteness/legitimacy. This choice was a deliberate one: the American republic was born in a revolution against centralized monarchy, and the American founders sought to create a system of checks and balances to prevent the re-emergence of strong, centralized power. The dual legitimacy of its presidential system is something with which many Americans are comfortable; poll data indicate that a majority of Americans are actually happier when the presidency and Congress are controlled by different parties, and want the two branches to serve as checks on each other’s power. And yet, the American political system has been capable of decisive action as well, from mobilizing to fight a series of wars, to building a twentieth century welfare state during the New Deal and Great Society years, to scaling back that same welfare state from the Reagan presidency to the present. The American political system has of course produced its share of political failure, deadlock, and missed opportunities. But over the years it has proved adaptable to changing conditions, at least when compared to the political systems of other developed democracies in Europe and Japan.

In American history, major policy initiatives have not been achieved through the inherent powers of the executive. As noted earlier, an American president is much weaker than a British Prime Minister. The president has no guaranteed majority in either house of Congress, and American presidents have been stymied by Congress even when their own party controlled the legislative branch (e.g., the defeat of Bill Clinton’s health care reforms in the early 1990s). Nor have American executives undertaken major initiatives through grants of special emergency powers by Congress.⁷ Some of the most impressive legislative accomplishments by American presidents, like President Truman’s passage of the Marshall Plan, or President Reagan’s passage of tax cuts during his first term, occurred when the other branch of government was under the control of the opposition political party.

Successful American presidents have achieved their goals not by exploiting the formal powers accorded them, but by using their office as a “bully pulpit” to rally broad public support across party lines (Neustadt 1990). Indeed, the greatest of American presidents like Franklin Roosevelt, Harry Truman, and Ronald Reagan, have understood that their chief function was to communicate broad messages and build coalitions across party lines. President Lyndon Johnson, for all of his failings in foreign policy, was able to enact major legislation regarding civil rights and poverty based on his intimate knowledge of Congress based on his days as Senate majority leader.⁸ Those who took a more technocratic approach to policymaking like Jimmy Carter or Bill Clinton (at least with regard to his health care initiative) were far less successful.

The example of the United States shows that a political system biased towards resoluteness/legitimacy can nonetheless be capable of decisive action. Great Britain by contrast is a case of a political system heavily biased towards decisiveness, which nonetheless has not used its inherent executive powers to trample over the rights of minorities. While Britain entered the modern era as a relatively homogeneous country in ethnic and religious terms, it inherited a highly stratified class structure that could have been the basis for serious class conflict. Yet despite the society’s evident ideological differences, neither fascism nor communism took root there as they did in other European countries. The Westminster system could have easily facilitated sharp oscillations in public policy between Left and Right, something that did not happen arguably until the arrival of Margaret Thatcher. Rather, Britain saw the displacement of the Liberals by the Labour Party, and the steady growth of the welfare state under both Labour and Conservative governments throughout the first seven decades of the twentieth century. Argentina began this same period with a similar degree of ethnic and religious homogeneity, and yet managed to exacerbate existing class differences through a series of violent oscillations between Left and Right. Differences in the design of the formal political system simply cannot explain this divergence in behavioral outcomes. They can, however, be more readily attributed to differences in political culture.

Conclusions: Social Structure and the Limits of Instituional Design

Formal political institutions do in the end matter. As a result of scholarly work done over the past generation, we can relate certain institutional forms to certain outcomes, e.g, how to design an electoral system to increase or decrease the number of parties, how to improve party discipline; how to promote greater fiscal responsibility on the part of states; and how to decrease incentives for patronage and clientelism. We also know that certain types of institutions are almost always dysfunctional and should be avoided, such as an overly politicized judicial system or patronage-based public expenditure.

On the other hand, there is no such thing as an optimal political institution. Institutions are only as good as the policies they promote, and no set of procedural rules for making political decisions will by itself ensure good public policy. These rules often seek to maximize two competing social goods, like decisiveness and resoluteness/legitimacy, and soceities will vary in their preference long-term risk in policymaking. Leadership matters a great deal; with good leadership, apparently dysfunctional institutions can be made to work well, while no set of rules will fully compensate for bad leadership. And while we understand cause-and-effect relationships between certain forms of institutional design and policymaking outcomes, the institutions in question come in highly complex, interdependent packages where a change in one requires complementary changes in several others, or else will produce unanticipated consequences.

All would-be institutional reformers face what Barbara Geddes calls the “politicians dilemma”: they can either spend their political capital achieving some short-term policy change, or they can spend it reforming the underlying instition. But in pursuing the latter, they often have to pay so much to get what they want that they undermine the very policies they hoped to promote.

It is not clear that going for a deeper institutional reform is always the right choice. Take the case of Mexico today. As noted above, Mexico has a presidential system and federalism that is biased toward resolutely/legitimacy and away from decisiveness. From the standpoint of a reformer seeking to liberalize the Mexican economy, President Fox’s inability to get major reforms through the Mexican congress has been intensely frustrating. The dual legitimacy of the presidential system has led to legislative-executive gridlock, while federalism has allowed Fox’s opponents like Manuel Lopez Obrador to prosper and pursue policies at odds with those of the president. One might be tempted under these circumstances to try to change the institutional rules to increase the power of the president, either to bias the electoral system to produce more decisive majorities, to reduce the autonomy of states, or to provide the president with emergency powers.

All of these changes presuppose, however, that it is a liberal reformer who is president, and that he is being blocked by populist or reactionary forces. Were the shoe to be on the other foot and a Hugo Chavez-type populist were to come to power, then these exact same institutions would serve as checks on a president pursuing bad policies.

The American system, as noted above, is not necessarily more decisive than the Mexican one in its formal institutional rules. American presidents who have wanted to promote ambitious reform agendas of course want as large a base for their own party as possible, but oftentimes have sought to cross party lines to build coalitions in favor of particular policies. All of this is a function of leadership, persuasion, campaigning, compromise – in other words, traditional political skills put at the service of public-regarding policies. In the United States, building consensus has always been a slow matter, and it should not be surprising that Mexico’s first president to come from a party other than the PRI should have trouble doing this. But the problem lies not in the formal institutions as much as with the skill with which they are exploited. President Cardoso’s relative success in getting important legislation through a formally weak Brazilian system should be seen as an example of this.

The preceeding argument that formal institutions need to be supplemented by a supportive political culture should not be taken as a form of cultural essentialism, that is, the view that societies are locked into certain outcomes due to unmanipulable cultural characteristics. Political culture reflects the shared experiences of groups of people at particular points in time, and can readily evolve under the influence of leadership, education, new environments and challenges, foreign models, and the like.

Besides political culture, another factor that makes democratic politics more difficult in Latin America is social structure. Democratic institutions are designed to mitigate social conflict, but their ability to do this depends in part on how severe those social conflicts are. Latin America is full of countries with sharp class, ethnic, and racial divisions, many of which have been the fault lines that have triggered coups, insurgencies, extralegal actions, and the like. In other cases (like Argentina), new lines of social cleavage were created de novo where none existed previously.

Democratic politics in Latin America has often been the province of social elites. Venezuela and Colombia, seen for many decades the best examples of stable democracy in the region, are cases in point. The 1958 Pacto de Punta Fijo in Venezuela and the 1957 National Front accord that ended the period of La Violencia in Colombia were pacted democratic transitions negotiated by two dominant elite parties that then shared power for the next four decades. This top-down approach preserved the positions of COPEI and Adeco in Venezuela and the Liberal and Conservative parties in Colombia at the expense of other political actors, leading to the progressive ossification of both political systems. Given large social stratifications and weak states in both cases, it is inevitable that excluded parties would over time try to force their way into the system, through guerilla insurgency in the case of Colombia and by populist military coup in Venezuela.

Countries with serious underlying social conflicts face a chicken-and-egg problem with regard to institutional development. It is hard to see how either Venezuela or Colombia can create democratic institutions that are stable in the long run without being more socially inclusive; but the very process of inclusion weakens or dismantles existing democratic institutions. Something similar is going on with regard to indigenous groups in Bolovia with the election of Evo Morales as president in 2005. Hopefully, existing democratic institutions will be used as the bridge to greater inclusion, but that depends very much on the agenda, ideas, and aims of the groups being included.

The United States began its national existence as

1. One important source for thinking about informal institutions is the large literature on social capital.
2. This article, which was published in the Journal of Democracy in 1990, circulated for several years prior to that as a mimeo.
3. The British system is no longer a pure Westminster system; there is a bicameral legislature with increasing devolution of powers to different regions in Britain. In addition, courts, including European courts, have increasingly been able to limit the discretion of the British parliament.
4. An example of this was President Gaviria’s bypassing of the legislature in the constitutional reform of 1991.
5. New Zealand ceased to have a pure Westminster system when their electoral system was changed from single member first-past-the-post to mixed member proportional in 1994.
6. Though even there, judiciaries that are too independent of public opinion have become highly controversial in the United States and other developed democracies that can take basic judicial independence for granted
7. American presidents have taken on special powers during war time, such as Abraham Lincoln’s suspension of habeas corpus during the Civil War, or President Roosevelt’s internment of Japanese-American citizens during World War II. More recently, President George W. Bush has claimed special powers with regard to treatment of “enemy combatants” and wiretapping as part of the war on terrorism. But these decisions have been very controversial and unusual in American politics.
8. While the Democrats controlled both houses of Congress during his presidency, the Democratic party itself was heavily influenced by southerners opposed to civil right legislation and in control of many key congressional committees.