According to an old joke, a patron goes into a library and asks for a copy of the French Constitution, only to be told that the library does not stock periodicals. The joke feeds the Anglo-American habit of needling France, in this case suggesting a country with suspect democratic credentials, more concerned with fashion and form than substance. Yet France is more typical of national constitutional practice than the United States with its venerable 218-year-old constitution. By our estimate, national constitutions have lasted an average of only seventeen years since 1789 . This is an unsettling estimate of life expectancy for a document whose basic function is to express guiding national principles, establish basic rules, and limit the power of government—all of which presuppose constitutional longevity.
On balance, constitutions that endure should be more likely to promote effective, equitable, and stable democracy. How durable are constitutions and what factors lead to their demise? Our concern is whether aspects of the design of constitutions have any significant effect on constitutional durability, net of other risk factors .
This question is not merely of academic interest. Recent constitutional drafting exercises in Afghanistan (2003) and Iraq (2004 and 2005) have been central milestones of American foreign policy. Each of these efforts sought to solve particular institutional problems, with different levels of success. It is, of course, too early to say whether either of these constitutions will survive to adulthood, but circumstances do not appear propitious in either country. In a far less volatile context in 2006, the 1997 constitution of Thailand—considered by many a model of institutional design adopted with extensive citizen participation—died a peaceful death in a bloodless coup at the age of nine. Indeed, given our estimated mortality rates, it is likely that constitutional replacement will be underway at any given moment somewhere in the world. Understanding what leads to such instances, and in particular whether design choices matter, has the potential to inform a science of constitutional design.
Any such epidemiological analysis requires an accurate historical census, a resource heretofore unavailable. As part of a large-scale research project, we have identified every major constitutional change—whether replacement, amendment, or suspension—in every independent state since 1789. We have also acquired the text for nearly every “new” constitution, as well as that for a large majority of amendments, and have recorded aspects of their design that our theory would predict to be relevant to constitutional longevity. Our analysis suggests a revision of the conventional wisdom regarding constitutional (and, more generally, institutional) change. The common intuition—rarely tested systematically—is one of sticky institutions that are unstuck only by cataclysmic world events, such as wars and economic crises. We find that constitutions are, as suspected, vulnerable to such crises. However, the inclusion of important design and process elements can add tens of years to constitutional lives, perhaps allowing the charters to survive even these intense shocks.
CONSTITUTIONS AS COORDINATION DEVICES
We hold the view that a successful constitution serves as a coordination device that renders its underlying political bargains self-enforcing, meaning that it must create a state of equilibrium from which no party has an incentive to deviate. Self-enforcement is important because, unlike normal contracts, there is in most circumstances no external guarantor who will enforce the constitutional agreement independent of the parties. Even though constitutional bargains may have relative winners and relative losers, they will endure to the extent that the losers either (1) believe they are better off within the current constitutional bargain than in taking a chance on negotiating a new one or (2) are unable to overthrow the existing order. Stability of the bargain depends on the winners upholding the commitments and limitations embodied in the constitution so that they do not provoke losers to resort to extra-constitutional action.
In democracies, enforcement of these constitutional limitations ultimately relies on citizens. If they can coordinate, citizens can prevent the government from imposing costs on them and violating the political bargain. If they cannot coordinate, democracy may not be stable, as the government will continuously adjust the bargain in its favor with political acquiescence.
Written constitutions can assist citizens in overcoming the coordination problem by providing a definition of what constitutes a violation by government, thus providing a focal point for coordination and enforcement activity. Resolving the coordination problem among citizens attempting to enforce limits on government behavior is extremely difficult, however, and the mere presence of a written constitution is no guarantee that coordination will in fact occur.
THE PROBLEM OF INCOMPLETE INFORMATION
A central problem for constitutional endurance is that, although the constitutional bargain may be an optimal, self-enforcing arrangement for the parties at the time it is drafted, it may not remain so.
At the start of any peacefully created constitutional arrangement, each party comes to the negotiating table to bargain. The bargaining process is costly, because it requires negotiation and the expenditure of political resources. The parties will conclude a bargain or not based on an expected stream of benefits to particular groups, net the transaction costs of negotiation.
Should they conclude a bargain, it will of necessity be incomplete, in that the parties will be unable to specify every future contingency. One reason it will be incomplete is the familiar one of the transaction costs of negotiating terms of a deal: parties that seek to specify every contingency will never conclude an agreement. Beyond the costs of negotiation, we focus on two types of obstacles to specifying a complete constitutional contract.
First, there is uncertainty about future payoffs, which may vary with exogenous factors. Exogenous change means that even endogenously stable constitutions may come under pressure for renegotiation. Our examination of constitutional histories confirms that constitutions frequently appear to die because of exogenous shocks, such as wars, regime change, and shifts in the boundaries of the state.
Another source of incompleteness is a lack of awareness of one’s negotiating partner’s position. This is the problem of hidden information. A party to constitutional negotiation may misrepresent its own endowments and intentions for strategic reasons. Hidden information can lead to a miscalculation of relative costs and benefits. One can imagine that if the miscalculation is severe enough, the disadvantaged party will seek to renegotiate the deal. Even if it does not, the difference between expected and actual costs and benefits may eventually increase the advantaged party’s power to the point where it is in a position to demand a better overall deal. Thus hidden information at the time of drafting can exacerbate pressures on the constitution later on.
The problem of hidden information is particularly severe in the first period of constitutional performance, and we have many examples of constitutions that die in their first year of operation, particularly in the context of failed peace agreements. Future contingencies, however, grow more difficult to predict with time, as more and more exogenous factors arise and interact with each other in complex ways.
WHY STANDARD SOLUTIONS TO INCOMPLETE INFORMATION DO NOT WORK
One standard answer to the problem of incomplete information is to write loosely defined contracts that allow flexible adjustment over time as new information is revealed. The parties specify performance within general parameters that can accommodate changing circumstances. There is, however, a well-known risk of moral hazard from such loosely specified contracts. If performance is not precisely specified, one might claim that circumstances have changed in order to take a greater share of the constitutional surplus. Indeed, knowing that this is a possibility down the road, a party might seek to conceal its intentions and endowments from its constitutional partners during negotiation. Trying to address the problem of incomplete information through drafting flexible “framework” constitutions, then, may exacerbate strategic problems of hidden information.
The reverse is also true. A standard response to the problem of hidden information is to write a more complete agreement specifying contingencies. By forcing the other party to reveal information during negotiation, one can minimize strategically generated surprises down the road. But this solution to the problem of hidden information, in turn, exacerbates the risk of rigidity in the face of exogenous change.
A third standard solution to problems of hidden and incomplete information is to rely on third parties. Analogizing to contract law, one might imagine a theory of constitutional review in which the courts seek to correct bargaining problems down the road. In such a case, the role of the court would be to provide default rules that reflect its understanding of the position the parties would have bargained to, should they have had all the information at the time. In contract theory, courts playing this role can provide a disincentive for negotiating parties to hide information from the other party.
There are significant problems, however, with expecting courts to serve this function for constitutions. First, there are capacity issues in which the courts may be unable to determine what the appropriate rule is. Second, in the constitutional context, no matter what decision the court makes, the relevant parties still face the second-order decision as to whether or not to comply with the court decision. That is, there is no guarantee that the decision will be followed and there is no external enforcer of the court decision. One must thus return to the incentives of the parties to understand constitutional endurance. Finally, the assumption that constitutional courts are able to correct bargaining problems of hidden information is problematic because courts are not automatically granted the power of judicial review. Indeed, the existence of a constitutional court is itself a product of the constitutional negotiation, a term over which parties will bargain.
To summarize, two sources of uncertainty, the first caused by variance in exogenous parameters and the second caused by strategic incentives to hide information, mean that parties will never be able to produce a complete constitutional contract. For each of them, information revealed later in time may affect the parties’ perceptions of the arrangement, putting pressure on bargains that may have been self-enforcing at the time they were written.
RENEGOTIATION AND BREAKDOWN
In considering whether to renegotiate, each party will consider its position in the current bargain, comparing it with expected outcomes of a constitutional renegotiation. Change in the constitution, however, is not costless. Amendment processes vary widely in their difficulty and complexity, and this will be a factor that affects a decision to seek to change the terms. Even more costly than amendment is total replacement, because there are more issues to bargain over and putting all the issues on the table renders the bargaining results less predictable ex ante.
If the expected outcome of constitutional renegotiation (conceived of as the set of all possible alternatives multiplied by their probabilities of obtaining, less negotiation and switching costs) exceeds the current stream of benefits, parties will opt for renegotiation. Once a party seeks renegotiation, it must consider what means to use. Here the constitution itself comes into play, for it will typically have provisions for amending its terms. There are two primary mechanisms by which constitutional change occurs: formal amendments to the text and informal amendments that result from interpretive changes. We expect that flexibility in these mechanisms of amendment will facilitate renegotiation.
Sometimes, however, infra-constitutional means of adjustment may be unavailable. For example, one party may control crucial institutions such as the legislature or courts that are necessary for formal or informal amendment. In such circumstances, the party may simply transgress the constitution and see if other parties acquiesce. Here the logic of coordination takes over. If the citizens or opponents can coordinate among themselves, they can refuse the proposed change by enforcing the terms of the constitution. In these instances, the original constitution survives, enforced. If coordination does not occur, the constitution may be replaced.
This model of constitutional transgression leaves us with three general propositions regarding the lifespan of constitutions:
- Strong enforcement mechanisms will decrease the probability of transgression and extra-constitutional replacement;
- External shocks and crises will increase the probability of transgression, as will characteristics of the state that lead to such crises; and
- Conditional on transgression, easily adaptable constitutions will decrease the probability of replacement.
INVESTIGATING MORTALITY: SHOCKS AND STRUCTURE
Analytically, it is useful to organize risk factors into three categories. Constitutional lifespan will depend on (1) the occurrence of shocks and crises (precipitating events), (2) structural attributes of the constitution, and (3) structural attributes of the state.
THE OCCURRENCE OF SHOCKS AND CRISES
We have rather strong intuitions about what sort of events would destabilize constitutional systems—those that significantly alter the balance of power within either the regime or the state. It is not hard to assemble a list of such events, as they constitute the milestones of a state’s political history. They include military subjugation, state merger or secession, diffusion (the tendency of one constitutional replacement to spur other replacements, especially in geographically or culturally proximate countries), regime change, leadership transition, and institutional crisis (an internal crisis irrespective of any ideological, leadership, or regime change).
STRUCTURAL ATTRIBUTES OF THE CONSTITUTION
The shocks that we describe above threaten the existing political order and undoubtedly have some effect on the lifespan of constitutions. Our crucial question, however, concerns the degree to which underlying structural factors play a role in mortality. In particular, do aspects of constitutional design play a decisive role? Our theory suggests that constitutions need to resolve problems of hidden information, provide incentives for enforcement (particularly by citizens), and also need to provide for flexibility in the face of exogenous pressures. We expect that the specificity of the document, the inclusiveness of the constitution’s origins, and the constitution’s ability to adapt to changing conditions will be important predictors of longevity. We also expect that a set of structural conditions associated with the state render the constitutional system more or less stable.
Specificity. One strategy that can help constitutions survive is to anticipate relevant sources of pressure and deal with them in the constitutional text, more fully specifying the constitutional bargain. It is particularly helpful in solving problems of hidden information among the bargainers. By forcing counter-parties to consider various possible future shocks and scenarios, the drafters can minimize problems of strategic behavior and holdup once the constitution comes into effect. We thus predict that specificity will be associated with constitutional survival. We use the term specificity in a general manner, as involving not only detail in the particular terms but also in the scope of the types of events the constitution covers. A broader scope of the constitution also indicates a certain amount of investment by the parties in negotiation, which may raise the prospective cost of renegotiation.
Inclusion. It is very clear that constitutions throughout the world are treated with varying amounts of respect by citizens and elites alike. For some countries (e.g., the United States), the document is an important symbol of sovereignty and statehood; for others (e.g., many Latin American constitutions of the 1800s), the constitution is of considerably lesser stature. In part, the connection between legitimacy and survival is reciprocal: framers and citizens will be more attached to a legitimate document, and documents that survive will in turn engender norms of attachment. Our theory suggests a further reason: constitutions whose provisions are known and accepted will more likely be self-enforcing, for common knowledge is essential to resolving coordination problems. This suggests that inclusion in the process of producing the constitution will help ensure enforcement by the public or other relevant actors. Constitutional durability should increase with the level of public inclusion during both the drafting stage and the approval stage.
Adaptability. As described above, exogenous change puts pressure on constitutional bargains. The ability of a constitutional system to adapt to changes in its environment will determine whether it remains in equilibrium. There are two primary mechanisms by which constitutional change occurs: formal amendments to the text and informal amendment that results from interpretive changes. To a certain extent, these mechanisms are substitutes. If the methods of securing formal amendment are difficult (as in the United States, with its requirements of ratification by three-quarters of state legislatures), there may be pressures to adapt the constitution through judicial interpretation. If, on the other hand, formal amendment is relatively simple, there may be less need for judicial or other institutional reinterpretation of the constitution.
Optimal adaptation thus results from the interaction of amendment rigidity and the possibility of reinterpretation of the constitution. The optimal level of flexibility is not universal, but determined in any particular constitutional situation by both exogenous factors (such as the rate of technological or environmental change) and endogenous factors (such as the level of responsiveness of political institutions under the constitution and the level of inclusion at the outset of the constitution scheme). A rigid constitution that fits its society well at the outset may be suitable if the rate of technological or environmental change is low, but the same constitution may perform poorly if change is rapid.
Constitutions that lack either flexible formal amendment processes or effective mechanisms of informal reinterpretation may not adapt to changing environmental conditions. We predict that such constitutions will force actors to take extra-constitutional action to secure changes and will thus die young.
STRUCTURAL ATTRIBUTES OF THE STATE
We expect that some state environments will be more conducive to constitutional survival than others. One set of such factors, of course, includes those that promote stability by mitigating internal conflict among groups. Such factors will sometimes be manifest in the crises that we specify above, but they likely affect constitutional lifespans directly as well. One is the age of the state, with the expectation that older states have a stronger sense of national unity and have achieved some degree of accommodation among conflicting groups (whether they be culturally or politically based). One can conceive of older states as having a more established set of unwritten constitutional conventions that will allow for adjustment over time and provide some insulation when shocks put pressure on the written bargain. Another factor may be the level of development. A basic empirical finding is that development tends to stall political change, in whatever direction. One can think of development as indicating that current constitutional arrangements are successful at providing a stream of benefits to various players, such that their absolute status is more secure under the current bargain that it would be under alternative arrangements. Transition costs are also likely to be higher in richer environments, as the opportunity costs of negotiating basic principles are greater. Finally, ethnic heterogeneity is likely to promote instability, inasmuch as political competition often falls along ethnic lines.
RESULTS OF THE DATA ANALYSIS
Constitutions, in general, do not last very long. The mean lifespan across the world since 1789 is 17 years. Interpreted as the probability of survival at a certain age, the estimates show that one-half of constitutions are likely to be dead by age 18, and by age 50 only 19 percent will remain. Infant mortality is quite high—a large percentage, approximately 7 percent, do not even make it to their second birthday. Also, we see noticeable variation across generations and across regions. For example, Latin American and African countries fit the joke of the French-constitution-as-periodical much better than does France itself. Our current analysis suggests that the mean lifespan in Latin America (source of almost a third of all constitutions) and Africa is 12.4 and 10.2 years, respectively, with 15 percent of constitutions from these regions perishing in their first year of existence. Constitutions in Western Europe and Asia, on the other hand, typically endure 32 and 19 years, respectively, and their lifespans are the least skewed. Organisation for Economic Co-operation and Development (OECD) countries have constitutions lasting 32 years on average, suggesting a development effect analogous to its well-known relationship with democracy. Finally, unlike the trend of improving human health, the life expectancy of constitutions does not seem to be increasing over the last 200 years. Through World War I, the average lifespan of a constitution was 21 years, versus only 12 years since.
Does the hazard rate (the probability a constitution will die at a certain age conditional upon its survival to that point) increase, decrease, or stay the same throughout the lifespan? In trying to answer this question, we restricted our analysis to the first 50 years of a constitution’s life, after which only 25 percent of constitutions remain and our confidence intervals are quite large. Our results suggest that constitutions are most likely to be replaced around age 10. However, the risk of replacement is relatively high during most of this period, and it appears constitutions do not begin to crystallize until almost age 50. Small samples do not allow us to describe the relative risks to those over 50, except to emphasize that even these hardy seniors are not immortal. Sweden’s constitution lasted 165 years, only to be replaced in 1974.
The statistics of the overall model suggest that both shocks and structural factors are important predictors of mortality. We find that several internal features of the constitution are strong predictors of durability, in accordance with our theory. In terms of inclusion, public ratification produces more enduring constitutions in democracies, but not in autocracies. This is intuitive: referenda in dictatorships do not genuinely confer legitimacy or facilitate collective enforcement of constitutional terms. We find that constitutions written in democratizing times are more resilient when precipitating events are included in the model. The most influential variables are clearly constitutional review and the ease of the amendment process, both of which decrease mortality. Adaptability, it appears, is crucial for constitutional survival. In the case of amendment ease, for example, an easily amended constitution (one whose probability of amendment is one standard deviation above the mean) has a 70 percent chance of lasting until age 50 versus 13 percent for those whose amendment probability is estimated at one standard deviation below the mean. Consistent with our expectations we find that constitutions that cover more topics are more durable than shorter ones, suggesting that specificity matters, although length of constitution alone does not seem to increase endurance.
Among the structural variables, several findings stand out. Ethnic fractionalization and wealth (as captured through energy consumption) have effects in the predicted directions (increasing and decreasing mortality, respectively). We find no effect for common law, consistent with our own intuition and contra to the well-known results in the law and finance literature. We also note that the trend toward shorter lifespans over the 200 years remains even after we control for a full set of covariates. Constitutions adopted from 1919–1944 are more vulnerable than are those adopted in earlier periods, and those adopted in the post-1945 period are more fragile still.
Our analysis of the constitutional life cycle leads us to think of constitutions as rather fragile organisms. Indeed, the average citizen outside of North America and Western Europe should expect to see her country cycle through six or seven constitutions in her lifetime. That estimate, of course, will depend on general levels of stability in any particular country. Those states that are the setting for crises such as war, internal violence, and coups should experience more frequent change. However, over half of the world’s constitutions survive even these major shocks, prompting our inquiry into the internal characteristics that may support resilience.
Enduring constitutions share three important qualities that date back to the circumstances of constitutional birth. First, durable constitutions tend to emerge under conditions characterized by an open, participatory process—conditions that encourage enforcement of constitutional terms. Second, durable constitutions tend to cover a wide range of topics, inducing the parties to reveal information and to invest in the negotiation process. Third, durable constitutions tend to be flexible ones, in that they provide reasonable mechanisms by which to amend and interpret the text to adjust to changing conditions. These findings have natural implications for constitutional design.
Thomas Ginsburg is a Professor of Law at the University of Chicago Law School. Zachary Elkins is an Assistant Professor in the Department of Political Science at the University of Illinois. James Melton is a PhD candidate in the Department of Political Science at the University of Illinois. This article is an excerpt from their forthcoming book, The Endurance of National Constitutions, to be published later this year by Cambridge University Press. To learn more about constitutional design, visit www.constitutionmaking.org.
 The median lifespan is only eight years, while the mode is a miniscule one year.
 We identified constitutions in the data that follow by a set of three conditions. The first is sufficient to qualify the document as a constitution, while the others are alternative sufficient conditions if the first is not met. Constitutions are those documents that either (1) are identified explicitly as the “Constitution,” “Fundamental Law,” or “Basic Law” of a country; or (2) contain explicit provisions that establish the documents as highest law, either through entrenchment or limits on future law; or (3) change the basic pattern of authority by establishing or suspending an executive or legislative branch of government.