WikiSummary, the Social Science Summary Database

Lijphart: Patterns of democracy

From WikiSummary, the Free Social Science Summary Database

 

Lijphart. 1999. Patterns of democracy. New Haven: Yale University Press.

General Overview and Main Conclusions

Chapters 1-4 call our attention to two competing types of democracy. The first, majoritarian or Westminster democracy, is what most people immediate think of when they think of democracy: A legislature elected by a simple majority of the voters governs, and voters throw the ruling party out if it governs poorly. Great Britian presents the best example of this type of democracy, hence the term "Westminster." The second type of democracy, consensus democracy, involves far greater compromise and significant minority rights. Westminster and consensus democracies differ along two dimensions, each of which has five elements; read on for details. Lijphart argues that consensus democracy is better (see below).

Place in Literature

Lijphart is the primary advocate of the consensus/consociational approach. See his other works to appreciate the development of this theme. In particular, consider his 1977 book, Democracy in Plural Societies. In this earlier book, he emphasized "consociational" democracy as a solution for states where traditional majoritarian democracy might not work due to deep ethnic, linguistic, or religious cleavages. But in the present book, Patterns of Democracy, Lijphart advocates "consensus democracy" (a modified version of consociational demoracy) as the ideal governance type for any state, not just deeply divided states.

As Lijphart has developed his arguments over the past few decades, he has attracted numerous critics. See Andeweg (2000) for an excellent review of consociationalism and its critics, as well as a discussion of the differences between consociationalism and consensus democracy. Note that the present book studies consensus democracy, not consociationalism.

Lijphart's Main Argument: Consensus Democracy is Better

General Advantages

Chapters 14-17 present statistical evidence to determine whether Westminster democracies produce consistently different policy outcomes than consensus democracies. Lijphart's main conclusion in this book: Contrary to the popular wisdom that decisive, effective, majoritarian leadership leads to better policy outcomes, Westminster democracies do not outperform consensus democracies. On some indicators (e.g. inflation), consensus democracies actually do significantly better than Westminster democracies; on most others, they do insignificantly better--which at the very least means that consensus democracies do no worse. Consensus democracies also have "kinder, gentler" traits: lower incarceration rates, less use of the death penalty, better care for the environment, more foreign aid work, and more welfare spending.

Special Advantages in Divided Societies

Consensus democracy has particular advantages for deeply divided societies. Majoritarian democracy might be criticized for excluding almost half the population from the governmental process, since it can leave 49.9% of the population out of the policy process. In the literature, we read that this criticism is void under two conditions. First, if today's minority has a realistic chance of becoming tomorrow's majority, then exclusion probably isn't a major problem, since each half of the country takes its turn being in charge (which will tend to moderate abuse of the minority by the majority). Second, if society is sufficiently homogeneous, then exclusion might not be a major problem, since the excluded minority's interests don't differ much from the majority's.

Lijphart contests these two arguments by pointing out that in many societies, especially in societies with deep ethnic, linguistic, religious, or ideological cleavages, neither condition holds. These deep divisions can prevent crossover ("swing") voting, preventing today's minority from ever having a realistic chance of being tomorrow's majority. Moreover, there is unlikely to be much overlap between the minority's and the majority's interests in such a society. Thus, the minority's permanent exclusion might lead to unrest or violence. Consensus democracy is Lijphart's institutional solution to this problem, allowing democracy to function by incorporating minority rights and allowing minority groups to influence policies. Though there might be less turnover in the legislature (see p 7), governments will represent a broader swath of interests (see pgs 31-33).

The Two Dimensions of Democracy

Consensus and Westminster democracies differ along two dimensions, each of which has five criteria. (Lijphart's earlier book, 'Democracies,' used factor analysis to show that these ten variables do actually load onto two distinct dimensions.) In the list below, I write the majoritarian characteristic first, followed by the consensus characteristic. Each bullet is discussed in greater detail below.

Executive-parties dimension

In effect: how easy is it for a single party to take complete control of the government?

  1. "concentration of executive power in single-party majority [MWC] cabinets versus executive power-sharing in broad [not MWC] multiparty coalitions." (MWC means "minimal winning coalition"; is the cabinet just barely large enough to control a parliamentary majority (51%), or does the cabinet include more parties than it has to?)
  2. the executive (president or cabinet/prime minister) is dominant over the legislature vs. a legislative-executive balance of power
  3. two-party vs. multiparty system. (This is partly a function of electoral rules; see Cox 1997.)
  4. pluralistic first-past-the-post electoral rules (which lead to disproportionate results) vs. proportional representation (PR)
  5. pluralist (i.e. atomistic) interest groups vs. "'corporatist' interest group systems aimed at compromise and concertation"

Federal-unitary dimension

In effect: once your party controls the government, how much can you change policy? Are there mechanisms to preserve the minority's voice/rights?

  1. unitary vs. federal/decentralized structure
  2. unicameral vs. bicameral legislature (with two "equally strong but differently constituted houses")
  3. flexible, easily amended (or non-existent) constitutions vs. rigid, supermajority-amended constitutions
  4. legislatures determine constitutionality of own legislation vs. judicial review of constitutionality by an independent court
  5. executive control of central bank vs. central bank independence

Examples of Each Type

Lijphart presents Britain as the archetypical majoritarian democracy. Despite some caveats, it was almost purely majoritarian on most of the ten indicators (from 1945-1996). New Zealand was also almost purely majoritarian, at least until the recent adoption of PR. (See chapter 2.)

Lijphart illustrates the concept of consensus democracy by presenting Switzerland, Belgium, and even the EU as cases. (See chapter 3.)

The Executive-Parties Dimension: Details

Chapter 5: Party Systems (Two-party vs Multiparty)

Discusses measurement of political parties. Uses the "effective number of parties." Also discusses measurement of issue dimensions. Shows a strong correlation between the number of parties and the number of issue dimensions. Comment: I suspect that both are caused by something prior: per Cox (1997), institutional constraints impose an upper bound on the viable number of candidates. Since you can't have many more salient issue dimensions than there are parties, the strong correlation may be an artifact not only of social cleavages, but also of institutions. In other words, I think Lijphart has a correlation here, but not causation.

Chapter 6: Cabinets and Grand Coalitions

There are two key factors to examine when classifying governing coalitions: The number of parties in the cabinet, and the breadth of the cabinet's support base in parliament (i.e. how many seats the cabinet controls). The most majoritarian variant is a single party cabinet controlling a bare majority of seats (often called a "minimal winning coalition," or MWC). The consensus variant is a multiparty cabinet controlling more seats than it has to (an "oversized" cabinet, or a "grand coalition").

Lijphart gives an excellent summary of the six main theories of cabinet selection (p 92-96). Unfortunately, the literature's six theories have a problem (per Lijphart): They tend to predict some kind of MWC. Why is this a problem? If we ignore countries where a single party is large enough to control the cabinet without forming a coalition, only 39% of the remaining cabinets are actually MWCs; the rest are either oversized coalitions or minority cabinets. Why? Lijphart lists several possible factors.

Why form an oversized cabinet?

  • Information effect: you aren't sure that one of your coalition partners won't defect, so you include extra partners. Since no one partner is necessary to maintaining a majority, you have insurance against any particular partner that acts unexpectedly.
  • Size principle: You want your party to contain the median coalition member so that you can have bargaining strength. If your party forms a coalition with only one other party, then the median coalition member might be in the other party or in the fringe of your own. By including another party that is on the other side (ideologically) of you (relative to the first partner), you ensure that the median coalition member will be near the median of your own party.
  • External threats, like war, encourage all the pro-system parties to form a single coalition. Grand coalitions are more common during times of crisis.
  • Constitutional requirements: Belgium (e.g.) is required to have a linguistically balanced cabinet.
  • Need for a supermajority: if your agenda includes some constitutional amendments, and those require more supermajority approval, you want an oversized cabinet.

Why form a minority cabinet?

  • No investiture requirement: you can take power without specific parliamentary approval
  • Constructive vote of no confidence can keep a minority govt in power (since it only loses power if the opposition can pass a no confidence measure).
  • Right to make proposals matters of confidence can keep minority govt in power: proposals become law unless an absolute majority votes to remove the cabinet
  • Strong parliamentary committees: excluded parties still have lots of power if they aren't in cabinet, so they are less concerned about being excluded.

Chapter 7: Executive-Legislative Balance (Presidentialism vs Parliamentarism)

Lijphart presents three dichotomous variables by which to differentiate parliamentary and presidential systems:

  1. In parl. system, the head of government requires the legislature's continued confidence to stay in power; in presidential systems, the head of government stays in power until his term is up.
  2. Popular selection vs. selection by legislature. Presidents are popularly elected; parliamentary cabinets are selected by the legislature.
  3. Collective/collegial executive vs. one-person executive. Presidents tend to control the executive as an individual (albeit with assistance from appointed cabinet members), while parliamentary cabinets tend to share executive power, with individual ministers exerting considerable control over their portfolio.

These three variables lead to a typology of 8 governmental types: pure presidential, pure parliamentary, and six hybrids. The typology does not account for "semipresidential" (or "premier-presidential") regimes, as in Shugart and Carey (1992): in these unusual cases, Lijphart examines only the executive (prime minister or president) that he considers dominant. Using exceptionally bad data, he shows that, as the proportion of a state's cabinets that are elected by minimal winning coalitions (MWCs) goes up, the dominance of the executive over the legislature also goes up.

Chapter 8: Electoral Systems (Pluarlistic vs Proportional Representation)

Lijphart presents a typology of electoral systems with some (uncritical) discussion of Duverger's law. A focus on seven attributes of electoral systems: electoral formula, district magnitude, electoral threshold, size of body to be elected, influence of presidential elections on legislative elections (i.e. coattails), malapportionment (i.e. mismatch between a district's population and vote share), and interparty electoral links. Uses Gallagher's (1991) index of disproportionality, and finds a strong association between disproportionality and electoral system (Table 8.2). He then finds a strong association between frequency of "manufactured" majorities and electoral system.

Comment: This is bad analysis--the relationship is an artifact of having fewer parties in plurality democracies, which will naturally lead to more manufactured majorities, even without having greater disproportionality. He claims that Figure 8.2 shows a relationship between electoral disproportionality and the effective number of political parties, but there is no relationship whatsoever if you control for PR/other electoral system. (Note that the PR countries are all on the right, below 8 or so, and the plurality systems are all above 8 (on the x-axis).

The Federal-Unitary Dimension: Details

Chapter 10: Federalism

The literature has given us two ways to think about federalism:

  1. A guaranteed division of power. Each level of govt has final say in some areas (from Riker)
  2. "Distribution [decentralization] of power among multiple centers" (e.g. Elazar)

Lijphart favors a different method: Congruent vs. incongruent federalism.

  • Congruent federalism: Federal units all have similar demographics as the nation as a whole. Example: USA and Canada. Federal lines are (for the most part) not drawn with ethnic groups in mind.
  • Incongruent federalism: Federals units have differing demographics; within each federal unit, there is usually demographic (i.e. ethnic/religious/linguistic) homogeneity. Examples: India, Belgium, Switzerland.

Lijphart also discusses "secondary characteristics" of federalism; these are institutional features that ensure that federalism will persist (i.e. that the national majority will not be able to move power away from the federal units and back to the central government):

  1. Bicameralism with a strong federal (territorial) chamber, like the US Senate
  2. A written, rigid constitution
  3. Judicial review to ensure constitutionality of legislation

Measuring how "federal" a country is can be difficult. Lijphart measures federalism using a five-point scale (p 189) that correlates well with other measures of federalism used in the literature (such as the percent of tax revenues that go to the central government).

We often hear that federalism promotes institutional/policy experimentation. This is possible, but rarely done; instead, most countries have the same regime type at national and regional levels. Example: USA is presidential both in the states and in Washington. But Lijphart promotes federalism not because it allows experimentaion, but because it ensures that minority groups have total control over some aspects of policy relevant to them.

Chapter 11: Bicameralism

The concentration/division of legislative power fits into Lijphart's federal-unitary dimension of consociationalism. He classifies legislatures along three variables:

  1. bicameralism/unicameralism,
  2. symmetrical/asymmetrical bicameralism (is one house dominant, or are they essentially equal?),
  3. and congruent/incongruent bicameralism (do the two houses represent essentially the same population, as in the American states, or does the second chamber reflect a different basis of representation, like the territorial representation in the US Senate?).

Using these variables, Lijphart gives bicameralism scores in Table 11.2 to each country in the sample. This operationalization isn't as easy as it should be: Iceland and Norway fall somewhere in the middle.

A true consensus democracy would have symmetrical, incongruent bicameralism; a true majoritarian democracy would have unicameralism (or something close to it, like Britain's starkly asymmetrical bicameralism).

Chapter 12: Constitutions

The things that preserve federalism are constitutional rigidity (hard to amend) and judicial review by an independent court. It is largely irrelevant whether the constitution is written down, as long as it is rigid and subject to judicial review.

Lijphart notes that you cannot have constitutional rigidity without constitutional review. Without constitutional review, the legislature can bypass constitutional rules and effectively change the contitution.

Chapter 13: Central Bank Independence

Independent central banks are characterized by longer tenures for their governors and a focused mission (e.g. to only fight inflation, not to bail out failing banks). Central bank independence is highly correlated with the federal-unitary dimension. Comment: This isn't surprising, since central bank independence is part of the federal-unitary index that Lijphart is correlating it against.

Additional Comments and Criticisms

Chapter 16: Consociationalism and "kinder, gentler democracy"

Chapters 14-17 have some methodological shortcomings. Consider chapter 16, which argues that consensus democracies are "kinder" and "gentler" because they are more likely to (1) be welfare states, (2) take care of the environment, (3) imprison fewer people, (4) eschew the death penalty, and (5) provide generous foreign aid. Also, they (1) elect more women, (2) reduce economic disparity, (3) have higher electoral turnout, (4) produce citizens who tell pollsters that they are satisfied with their democracy, (5) select leaders with opinions that correspond more closely with citizens', (6) enhance accountability and minimize corruption, (7) ensure that the government is always backed by a majority, not just a plurality (like Bush in 2000 and several UK cabinets), and so forth. If you value these things, you should value consensus democracy.

Consider several methodological critiques of chapter 16, though:

  • Problem: x (consociationalism) is measured once, in 1971-1996. This is only a cross-sectional correlation, not a causal (time-series) test. (See page 248 for a list of how each country rates on his scale. Upper-right is majoritarian, lower-left is consociational.)
  • Problem: He assumes that economic disparity can be used as a proxy for political equality. Not true. This only works if you assume that everybody wants redistribution to themselves.
    • Counterexample: Why are there rich Democrats and poor Republicans? At least in the USA, politics is not solely about redistribution. Also, recall that Howard Dean (in 2004) bemoaned the fact the the Democratic Party wasn't reaching out to poor whites in the South; these poor whites were voting based on something other than economic redistribution, and Dean wanted the Democratic party to change that. But if Lijphart's proxy were valid, then Dean's concern never would have arisen in the first place. You cannot use economic disparity as a proxy for political disparity.
  • Problem: he argues against the idea that it is easier to hold parties accountable in a majoritarian system than in a consensus democracy. But although he gives evidence that it isn't always easy to exercise accountability in a majoritarian system, he gives insufficient evidence that consensus democracies are any better. In fact, one of the biggest criticisms of Lijphart's approach is that you can't throw a bum Grand Coalition out. Austrians had to vote for someone that many of them disliked--the rather extreme Joerg Haider--to break up persuade their wayward Grand Coalition to change.
  • Big problem: Lijphart tests lots of hypotheses, but only has 15-36 countries in each regression. He used up all of his degrees of freedom rapidly. (That he didn't test all the hypotheses in a single regression is irrelevant--he used up his d.f. by running several regressions on the same data set.) Also, a data set so small is easily influence by large outliers. Many of his tests could be driven by a couple of very strong outliers, the US and the UK especially. Further, he's got a major multicolliearity problem; most of the majoritarian systems are former British colonies. Are there cultural effects here? Also, most of Western Europe is in the "consensus" category. Again, is that driving anything? He has too few cases, too many hypotheses, too many variables that he should be controlling for (but does not), and too few degrees of freedom to resolve these problems. Is there a better way to test these hypotheses?