There has been a veritable cottage industry examining comparative electoral systems in developed democracies. The primary focus of this literature generally evaluates the respective merits of representing the majority vs "as many people as possible" (Lijphart). Surprisingly, however, very little attention has been directed toward the effects on democratic performance in those countries where parties have been excluded from representation and/or governance.

This paper takes a new look at an issue - representation - that has animated the work of Lijphart and many of his followers and detractors for the past two decades. Specifically, it evaluates whether or not one of these two democratic forms is superior in providing representation - but also in avoiding exclusion - at both the electoral and governmental levels. The data suggests that numerous democratic political parties have been excluded from both the parliaments and cabinets in fifteen developed democracies - both majoritarian and consensus - when electoral results and cabinet compositions from 1945 to 1995 would seem to have suggested different outcomes. The paper asks whether or not this exclusion from legislatures and cabinets has implications for the practice of democracy as an ideal regime type. Extrapolation from the data follows by challenging a dominant approach to representational constraints and then offers insights on the quality of representative democracy.
 
 

Introduction

The last quarter of the 20th century has seen the rapid proliferation of democracies around the world. Eastern Europe, Latin America and East Asia have all witnessed the demise of autocratic and dictatorial regimes and their replacement with democratic forms of government. (Diamond and Plattner, 1996). As newly democratic regimes arise, their framers face several options. The two most common are consensus democracy, often associated with proportional representation electoral systems in parliamentary governments, on the one hand, and majoritarian democracy often associated with either presidential systems or Westminster style parliamentary regimes with single member district, or first-past-the-post electoral systems, on the other.

The differences between these institutional systems have been well documented in the comparative politics literature (Crepaz, 1996a, 1996b, Birchfield and Crepaz, 1998). In the most general terms, the model of consensus democracy, most often associated with Arend Lijphart (1984, 1999), claims to produce superior democracy by representing as many people as possible. The majoritarian model (Dixon, 1968; Finer, 1975) on the other hand, promises more decisive and effective government policy by, not surprisingly, representing the majority.

This paper takes a new look at an issue – representation – that has animated the work of Lijphart and many of his followers and detractors for the past two decades. Specifically, we evaluate whether or not one of these two democratic forms is more effective in providing representation – but also in avoiding exclusion – at both the electoral and governmental levels. Our data suggests that numerous democratic political parties have been excluded from both the parliaments and cabinets in fifteen developed democracies – both majoritarian and consensus – when electoral results and cabinet compositions from 1945 to 1995 would seem to have suggested different outcomes. We ask whether or not this exclusion from legislatures and cabinets has implications for the practice of democracy as an ideal regime type. We then attempt to extrapolate from our data by challenging a dominant approach to representational constraints and then offer insights on the quality of representative democracy.

We present the argument in four sections. First, we address the issue of disporportionality by going beyond the usual seats/votes debate and suggesting that this issue has a cabinet government dimension as well. Second, we articulate our concrete research question and hypotheses. In the third section we present our empirical findings and create a comprehensive index of ‘democraticness’ based on both legislative and executive disproportionality. Finally, we then place our research question in the context of the literature on representation, proportionality and democratic theory.
 
 

(Dis)proportionality in Majoritarian and Consensus Democracies

Both majoritarian and consensus systems have their respective advantages and disadvantages. Most often, the balance of these characteristics is determined as a zero-sum game. More of one quality necessarily means less of another. For example, it is often argued that the effective stability of majoritarian governments comes at a price of the more democratic but less predictable representativeness that is found in consensus systems.

Despite the seemingly apparent logic of these zero-sum arguments, there are certain characteristics present in both majoritarian and consensus systems that have implications for representative democracy. The most glaring of these is that, since the end of WWII, both institutional manifestations of democracies have systematically excluded political parties either from representation in parliament or representation in government. For example, some systems have electoral rules that prevent parties from winning legislative seats, even though the party’s percentage of the overall vote is relatively high. This observation is not new; indeed there have been many attempts to create an acceptable disproportionality index that captures the relative difference between the percentage of votes to the percentage of seats for each party in each country. Creating such an index, however, has not been an easy task. Since some amount of disproportionality between the percentage of votes and seats is inevitable, each country’s vote/seat formula reflects how that country wishes to handle the discrepancies (Gallagher, 1991). As a result, any cumulative index will be inherently flawed, because there is little uniformity in how each country manages this problem.

It is largely accepted that majoritarian systems that use single member districts and first-past-the-post electoral rules have fewer and larger parties. The winner-take-all aspect of systems naturally weeds out smaller contenders. However, where small parties do exist in majoritarian systems, the value for that country on most disproportionality indices is automatically distorted. Thus we know, a priori, that majortiarian systems will usually cluster with higher scores of disproportionality. These are often compared with the lower scores for consensus systems, which predominately use multimember districts and proportional representation. Moreover, among the consensus style governments, how votes are tallied to determine the number of seats a party will receive (in other words, which electoral formula they use) will determine where they find themselves on the various disproportionality indices. Countries that utilize the d’Hondt formula, for example, will find themselves with higher disproportionality scores than countries that use modified Sainte Lague or quota methods. To sum the overall tendency between consensus and majoritarian systems, Lijphart notes, "PR systems do tend to be considerably less disproportional than plurality and majority systems" (Lijphart 1999).

Lijphart’s excellent 1999 book clearly sides with the consensus system of government as a superior model for democracy. He asserts that it is "kinder and gentler" (p. v.) and allows as many people as possible to involved in the government process. Thus, inherent in his argument is the suggestion that the quality of democracy is higher in consensus systems than in majoritarian ones. This is based on a factor analysis that reveals two ‘dimensions’ of democracy. The first, the executives-parties dimension, contains a clustering of variables around 1. the effective number of parliamentary parties, 2. minimal winning one-party cabinets, 3. executive dominance, 4. electoral disproportionality, and 5. interest group pluralism. The second, the federal-unitary dimension, contains a clustering of variables around 1. federalism-decentralization, 2. bicameralism, 3. constitutional rigidity, 4. judicial review, and 5. central bank independence (Lijphart 1999, 243-257).

This discussion of electoral disproportionality (first dimension, fourth variable) only centers on the electoral (or legislative) side of democracy. When we think of disproportionality, we automatically think of the ratio between percentage of votes to percentage of legislative seats and how far that amount deviates from what we would expect, given ideal conditions. Thus, most often, values of disproportionality are determined by how votes are translated into legislative seats. We wish to suggest is that there is another, very important, part of this picture because the emphasis that disproportionality indices place on the legislative side to government leaves out a crucial dimension of the political process. We want to argue that for any discussion of disproportionality to be complete, executive disproportionality must be considered as well. In the spirit of Lijphart’s work, we consider this our "second dimension" of proportionality.

In parliamentary democracies, governments, or cabinets, are formed after legislative elections. Parties, if they have received enough of the seats to command a minority or majority government, rule alone, otherwise they are forced to work together to form governing coalitions. In some countries there is significant alternation between these parties in governments over time. However, there is little or even no alternation in the cabinets of other democracies. Therefore, what we wish to suggest is that there is another kind of disproportionality at work within the political process itself, one that occurs after the votes have been tallied and seats assigned. While previous indices of ‘disproportionality’ measure the relative ratio of percentage votes to percentage seats in democracies (what we call legislative disproportionality- our first dimension), a new type of disproportionality index can be envisioned, one that measures the extent of each party’s participation at the cabinet level of government. When this particular executive disproportionality index is added to the more common index that measures legislative disproportionality, a clearer picture concerning the level of democraticness that is present in each system is revealed.

On face value we may assume that the higher the score of legislative disproportionality, the less democratic the country. Most often these systems are majoritarian ones: the single member districts and the first-past-the-post electoral rules naturally distort the disproportionality index. As a result, majoritarian systems often have higher scores and may bias discussion against them. However, when we also consider how much government alternation is present in each system, majoritarian systems may actually be more democratic in practice than consensus systems that score very low on the legislative disproportionality indices (Anderson, 2000).

How democratic can some regime types (consensus or majoritarian) actually be, if democratic political parties – playing by the rules of the game – are denied a chance at governance despite an electoral performance that should otherwise provide such representation? We are not necessarily suggesting an intentionally malicious exclusion by large dominant parties of their smaller brethren, since some of this exclusion can simply be attributable to clearly established electoral rules and/or thresholds. However, if persistent episodes of exclusion of sizable political parties continue to take place among advanced industrialized democracies -- whether intentional or merely accidental -- then we need to assess how this phenomenon affects the quality of democracy for any given country. To assess this problem, we concentrate our analysis of representational exclusion among fifteen developed parliamentary democracies from 1945 to 1995.

For a fully effective and accountable democracy to be present, we suggest that all viable parties (meaning those that pass the electoral requirements for legislative and government participation) have to be equally coalitionable. Thus, those countries that regularly have non-participating parties time after time may be less democratic than others. We therefore suggest that large, dominant parties in all democracies must allow all parties, even the smaller ones, fair and equitable opportunities to participate at some time in order for a strong democracy to be consolidated effectively. Simply calling a regime democratic doesn't necessarily mean that it's really an optimal democracy. (Dahl, 1998)

These observations are important because forms of political party exclusion are present in many developed political systems, which include both consensus and majoritarian parliamentary systems and this phenomenon might have implications for the quality of democracy we find in these systems. For example, the Italian Communist Party (PCI) regularly received between 25% and 30% of the vote between 1945 and 1989 in national elections, were awarded a significant number of legislative seats, but never obtained cabinet seats. Similarly, the centrist Liberals in Britain have often won more than 20% of the votes in national elections but because of electoral rules, scarcely have ever received more than 5% of the seats in the House of Commons. As a result, the Liberals, like the PCI in Italy, were excluded from the government process. Thus, the puzzle that leads us to this research is that some political systems exclude various democratic political parties from participation either in parliament or in government, while others do not. Why is this the case? And does perpetual party exclusion from government have implications for the type, style and quality of democracy that we find in parliamentary regimes, both consensus and majoritarian?
 
 

Research Question and Hypotheses

We are trying to assess the consequences of party exclusion, at both the legislative and executive levels of government, on the quality of democracy in 15 advanced democracies. For the moment we are less interested in why individual parties have been excluded than we are in highlighting this neglected aspect of democratic exclusion and its implications for the quality of democracy. Thus, our approach is more consistent with the historical institutionalist school than with more rational choice institutionalist frameworks that emphasize why parties are excluded. Both are valid approaches in studying representation and democracy, however we believe that the former allows us to investigate more effectively the quality of democracy issue.

What is the most likely explanation for the diminished quality of democracy in

some countries other than the majoritarian/consensus distinction?

(H1) The quality of democracy is diminished in some countries becausethere is a systematic exclusion – by either accident or institutional design in both consensus and majoritarian democracies -- of otherwise democratic political parties from either gaining seats in parliaments or participating in government coalitions. We argue that this phenomenon diminishes democracy by both denying representation in parliament and/or participation in government.

What is the most likely alternate explanation?

(H2) The quality of democracy is not diminished in some countries because these apparent exclusions of political parties from participation in either parliament or in government could be nothing more than the idiosyncratic vagaries of democratic politics. In this case, democracy is hardly diminished, rather these machinations are the essence of democracy.

What kind of evidence would confirm these two hypotheses?

(H1) To suggest that the quality of democracy has been diminished in some countries, we would need to show that either the electoral system and/or the pattern of coalition formation over the 1945-1995 period does not provide the full range of representation and participation in government that one could reasonably expect given the existing configuration of political parties in these countries. Confirmation of this hypothesis requires the development of two sets of indices to measure the level of both legislative and executive disproportionality in 15 democracies from 1945 to 1995. These will be combined to present a comprehensive picture for the level of ‘democraticness’ for each country in our dataset. The first index will measure the difference between the percentage of votes and percentage of allocated seats for each party in each country to provide a measure of expected party inclusion for these countries’ legislatures. The second index will measure the number of coalitions versus the number of expected coalitions over this 50 year time period and will thus provide a measure of whether there were more coalitions then there "should have been". Indirectly, the second measure will indicate that there was something faulty about the coalition formation process, suggesting that parties that might have made potential coalition partners were not considered cabinet after cabinet. When the number of actual coalitions greatly exceeds the number of expected coalitions, we might suspect that these systems experience systematic party exclusion from the executive level of government.

What would disconfirm this hypothesis? Our data would have to find that there was no exclusion either at the parliamentary or coalition level among these major advanced industrialized democracies. In this case we argue that democracy is functioning and healthy at both the legislative and executive levels of government.

(H2) Evidence or data to support this alternate hypothesis would have to find any variation explained by such factors as idiosyncratic electoral rules, vagaries of political culture in any or all of these countries, and/or conscious political decisions made by reasonable politicians under plausible circumstances, etc. (Strom et al., 1994).

What would disconfirm this alternate hypothesis? We would find a consistent pattern that cuts across all of these countries both spatially and temporarily and suggest there were patterns that went far beyond any of the above mentioned idiosyncrasies. In this case we would suggest that the quality of democracy in certain countries has been compromised because of these patterns of systematic party exclusion.

A Comprehensive Index of "Democraticness"

The purpose of this paper is to evaluate 15 advanced parliamentary democracies on two indicators of "democraticness". These indicators are combined to form a comprehensive index that includes both electoral and governmental representation. The electoral indicator measures the level of proportionality between the percentage of votes and seats in the electoral system. This measure will parallel similar measures that have been investigated in the literature. Our principle contribution, therefore, is our second indicator, which indirectly measures the "democraticness" of the government. This indicator will measure the difference between the number of elections and the number of governments that formed in each country from roughly 1945 to 1995. The rationale behind this indicator may not be intuitively clear. Many countries experience several governments after a single election. Our argument is that countries in which the number of governments greatly exceeds the number of scheduled elections may be forming counterproductive cabinets.

The two indices for the comprehensive index have been given equal weight and added together. The index for legislative, or electoral, disproportionality is determined by the Sainte Lague index, the formula for which is taken from Gallagher (1991) and is similar to the one utilized by Lijphart (1999). It has been calculated using data from Mackie and Rose (1982) and from Mackie’s annual contributions in the European Journal of Political Research. Every election has been used from 1945-1995 in 15 parliamentary democracies: Australia, Austria, Belgium, Canada, Denmark, Finland, Germany, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Sweden and the United Kingdom. The index is measured per election by squaring the differences between the percentage of votes and percentage of seats received per party and weighting each difference by dividing by the size of the party (the percentage of votes received). To create an overall summary score for each country between 1945 to 1995, the values were summed across elections and then divided by the number of elections. This produced an average disproportionality score for each country.

The Sainte Lague index has particular virtues because it uses squared values, unlike many other disproportionality indices. The squared values serve to exaggerate larger differences between votes and seats, while not ignoring the fact that several small parties may have won some of the vote but none of the seats. The Sainte Lague index registers the vote share from smaller parties, but if support for a small party is minimal and if the party receives no seats after the election, the value for the index will not be significantly altered because the squared values will minimize the smaller values (especially when they are close to 1). The value for the index, however, will increase significantly as the ratio between the percentage of votes to percentage of seats for several parties grows large. In addition, the Sainte Lague index is not prone to numerical paradoxes that are theoretically found in other indices.

The values for the legislative disporpotionality index are calculated for each of the 15 parliamentary democracies in our dataset using the Sainte Lague index formula. The raw values are presented in Table 1:

Table 1: Legislative Index Based on Sainte Lague formula for 15 parliamentary democracies, 1945-1995. N=240
 

Country	Score	# of Elections
SWEDEN	1.47	16
THE NETHERLANDS	2.68	15
AUSTRIA	2.74	16
DENMARK	2.96	21
IRELAND	3.46	15
BELGIUM	3.70	17
FINLAND	3.74	15
GERMANY	4.29	13
NORWAY	5.40	13
JAPAN	7.10	19
ITALY	7.70	13
CANADA	13.07	16
UK	13.16	14
AUSTRALIA	13.79	20
NEW ZEALAND	16.80	17

The first striking feature is that the Westminster systems have much higher values than systems that use PR. More parties in these countries win greater amounts of the popular vote, but are denied legislative representation in parliament. This should not be a surprise, since these countries (Australia, Canada, New Zealand and the United Kingdom) utilize first past the post electoral systems with single member districts. In one sense, it may not be fair to include these countries in the index; indeed the index itself is based on a PR formula. Proportionality, for majoritarian systems, is not a primary concern. Looking at the electoral performance and the consequent seat allocation of the several ‘third parties’ in majoritarian countries offers several examples: the varying levels of over and under representation of the Labour Party, National/Country Party, and Liberal Party in Australia over time, the consistent under representation of the CCF/New Democratic Party in Canada, the Social Credit Party from 1954 to 1984 in New Zealand and the Liberal Democratic Party in the UK are prime examples. The disproportionality between the earned vote and consequent seat allocation in these countries is not necessarily a failure of democracy or the concept of representation, but a discrepancy which results in majoritarian systems that utilize FPTP systems.

Nonetheless, this first index, which calls attention to blatant differences majoritarian and consensus systems, might have implications for the level of democraticness in certain countries. Many authors have called attention to the fact that high legislative disproportionality may be a symptom of a democratic deficit, since the popular mandate, as reflected in the percentage of the vote received by certain parties, is not reflected in the allocation of legislative seats.

The purpose of this exercise, however, is also to include the second dimension, that of executive disproportionality, to this picture. This index is calculated much more easily than the legislative disproportionality index, but its interpretation is not as straightforward. The index is measured as the ratio between the number of governments to the number of elections from 1945-1995. The inspiration for the index comes from the intuitive appeal of a one-to-one ratio for governments to elections in an ideal world. A perfect one-to-one ratio would indicate that a government has either survived all votes of no confidence, has not be subject to any such vote, or has not resigned on its own accord or for any other reason. A value slightly higher than one indicates that a country has experienced more governments than elections. There can be many reasons for this, some of them completely expected. However, what concerns us are the values that are

exceptionally high, values that indicate that a country has experienced several governments for each election. While some of this government turbulence could be nothing more than the natural result of parliamentary dispute between parties and the individuals within them, an unusually high score might suggest something else. We want to suggest that governments that fall on a regular basis are not healthy for democratic practice. Furthermore, governments that fall on a regular basis which are replaced with the same, or close to the same, configuration of political parties might suggest rational action on the part of the individual members within the coalition parties and may represent deliberate party exclusion in these systems. MPs that defect cabinet after cabinet might be confident enough to know that their actions will not result in the expulsion of their own party in subsequent government formations. One reason for this confidence might be the systematic, a priori, exclusion of certain political parties from the negotiation process. When this happens, the number of future possibilities is dramatically reduced. Over time, the systematic exclusion comes to resemble an institution, or a perpetuated norm that constrains the coalition process.

There has been disagreement on how to treat excluded political parties and whether or not this phenomenon represents a formal institution blockage that constrains the coalition bargaining process. On the one hand, Laver and Schofield (1990) warn that incorporating the ostracism of pariah parties as constraints may be treating as law a phenomenon which is better explained as purposeful action by the actors themselves. They explain that "one of the strongest behavioural regularities that we observe in the politics of coalition in Europe is that certain parties are designated by others as ‘non-coalitionable’. A general account of coalitional behavior should be able to tell us why this is the case. Simply to assume that such-and-such a party can never go into government in Denmark, for example, because this is one of the ‘rules of the game’, does not get us very far. It is assuming what we should be setting out to explain" (p.201). However, on the other hand, Strom et al. (1994) do incorporate the exclusion of certain parties as a party rule constraint on cabinet formation. They find that "the systematic exclusion of certain parties from coalition bargaining is the most striking party constraint found with any regularity" (p317). While much of this is country-specific, the systematic exclusion of certain political parties is being revealed as a general trend with comparative implications: "The examples of the Gaulists or Communists in the French Fourth Republic, Sinn Fein in Ireland, or the Italian Communists in the 1950s show that certain parties, as a consequence of their strong ‘anti-system’ stance, can effectively be discounted as members of any potential government" (p.317). Strom and his colleagues further suggest, however, that some of this ostracism may be self-imposed. However, even when parties begin to change their positions on the acceptability of government participation, they may find themselves unable to enter the bargaining process, as "most other parties still treated them as anti-system outcasts and refused to bargain" (p.317). As a result, the systematic exclusion of certain political parties does, over time, come to resemble a type of constraining institution.

When these excluded parties are of considerable size, yet still too small to govern alone, the number of possibilities for coalitions is greatly reduced. It is this observation that drives the theory of this paper to suggest that for democracy to function properly, all parties must be equally coalitionable. When they are not, the conditions for (individually) rational behavior are much more likely because the members of the leading parties know that their specific party or parties will be present in the next coalition, regardless of their behavior in the short-term. Thus, the exclusion of the pariah parties may be the key explanatory factor explaining why some countries experience many more governments than elections than other countries where all parties have a chance at entering the government bargaining realm. Since the members of the largest parties in the coalitions that systematically exclude certain parties are confident enough to assume that the next coalition government will be greatly similar to the previous one, they do not have as much to lose in the short term by letting the government fall. And when this trend is perpetuated over time, as it appears to be in countries such as Italy and Japan, voters may become disillusioned with prospects for change and help perpetuate the situation by not changing their vote, or perhaps not voting at all (Cox, 1997). In other words, there could be various kinds of "wasted votes" each of which could diminish democracy.

Table 2 reports the raw values for the ratio of governments to elections for the 15 parliamentary democracies: Data taken from Woldendorp et al (1998), Mackie and Rose (1993) and Mackie’s annual contributions to the European Journal of Political Research.
 
 

Country	Governments	Elections	Ratio
CANADA	20	16	1.25
AUSTRIA	21	16	1.31
IRELAND	20	15	1.33
AUSTRALIA	27	20	1.35
UK	19	14	1.36
DENMARK	29	21	1.38
NETHERLANDS	21	15	1.40
NEW ZEALAND	24	17	1.41
SWEDEN	23	16	1.44
NORWAY	24	13	1.85
GERMANY	26	13	2.00
JAPAN	40	19	2.11
BELGIUM	37	17	2.18
FINLAND	46	15	3.07
ITALY	55	13	4.23

Most countries, including all of the majoritarian systems for this index, have had between 1.25 and 1.85 governments per country between 1945 and 1995. However five of the countries have had at least 2 or more governments per each election: Germany, Japan, Belgium, Finland and Italy. This suggests that these countries may have formed counterproductive coalitions.

To combine the legislative and executive indices, the scores have been standardized with regard to each other so that the values for each individual index are bounded between 0 and 1 and can be compared to each of on a relative, rather than absolute, scale. For the legislative index, each raw score has been divided by 16.8, the high score for New Zealand. Now scores closer to 1 reflect higher legislative disproportionality. This transformation produces a low of .09 for Sweden and a high of 1.0 for New Zealand. Similarly, the raw scores for the executive index have been divided by Italy’s high of 4.23, producing a low of .31 for Austria and a high of 1 for Italy. To create a comprehensive index with a possible low of 0 and a possible high of 2, the scores have been added together and can now be compared to each other.
 

Country	Legislative Index	Executive Index	Cumulative Index
SWEDEN	0.09	0.34	0.43
AUSTRIA	0.16	0.31	0.47
THE NETHERLANDS	0.16	0.33	0.49
DENMARK	0.18	0.33	0.50
IRELAND	0.21	0.32	0.52
GERMANY	0.26	0.47	0.73
BELGIUM	0.22	0.51	0.73
NORWAY	0.32	0.44	0.76
JAPAN	0.42	0.50	0.92
FINLAND	0.22	0.72	0.95
CANADA	0.78	0.30	1.07
UK	0.78	0.32	1.10
AUSTRALIA	0.82	0.32	1.14
NEW ZEALAND	1.00	0.33	1.33
ITALY	0.46	1.00	1.46

Before we argued that the majoritarian systems score particularly poorly on the legislative disproportionality indices. However, in the cumulative index of disproportionality, we see that the consensus systems of Japan, Finland and Italy are clustered more closely to the scores of Canada, the UK, Australia and New Zealand than the other consensus systems. The reason for this, of course, is that the ratio of governments to elections for these three consensus countries is much greater than one. Again we suggest that this phenomenon might be related to an inherent executive disproportionality. While the majoritarian systems exclude parties at the legislative level, these three consensus systems might be excluding parties at the executive level, resulting in higher cumulative scores for disproportionality.

The following table reports the percentages of time between 1945 and 1995 (counted in days) that a single party was included as the first (the first party is usually the largest) of the coalition. The most blatant observation is that the DC was the first coalition partner in Italy 99% of the time from 1945-1995; in 1994 it was replaced by Forza Italia, which was replaced nine months later by a government of political independents. Also noteworthy is that the Swedish Social Democrats, the Belgian Christian Democratic Party, the Japanese Liberal Democratic Party, the Finnish Centre Party and the Norwegian Labor Party and were in power as the primary coalition party over 75% of the time.
 

Country	%				Party
ITALY		99%		DC		Democratic Christian Party
SWEDEN		84%		SDA		Social Democrats
BELGIUM		82%		CVP		Christian Social Party
JAPAN		79%		LDP		Liberal Democratic Party
FINLAND		78%		KESK		Finnish Centre Party
NORWAY		75%		AP		Norwegian Labour Party
GERMANY		72%		CDU		Christian Democratic Union
CANADA		70%		LIB		Liberal Party
UK		67%		CON		Conservative Party
IRELAND		64%		FF		Fianna Fail
NETHERLANDS		64%		KVP		Catholic People's Party
NEW ZEALAND		64%		NP		National Party
AUSTRALIA		61%		LIB		Australian Labor Party
DENMARK		60%		SD		Social Democrats
AUSTRIA		52%		SPO		Socialist Party of Austria

Data taken from Woldendorp et al (1998).

When we compare these values to the raw values on the governments/election ratio that we used for our executive disproportionality index, we find a fairly strong positive correlation, r=.79. This suggests that the greater the percentage that one party has been in power in a country, the more likely the country will be to experience several governments per election. This adds to the validity of our executive disproportionality index. The longer one party dominates the cabinet in a given country, the more likely that country is to witness fallen governments, which we believe jeopardize a healthy democracy. Thus it may very well be the case that when parties deliberately are kept out of the cabinet, other parties find themselves in opportune situation (Mershon, 2000). In these cases, the coalition parties are not threatened by a fallen government, since they are secure in knowing that an alternative possibility will probably not replace it.
 

Representation and Democracy in Comparative Politics

What exactly is democracy? Is it simply a process through which voters express their preferences? Or is there also a second dimension within the definition of democracy that accounts for what actually happens once the votes are tallied and representative bodies take office? This paper takes the position that both sides need to be evaluated. One is the level of electoral representation. How closely do votes match seats in the main legislative body? Are citizens’ preferences adequately reflected in the composition of the legislature? Our data shows that there are clear differences between SMD/winner take all and MMD/PR in this respect, with the latter representing "as many people as possible" as opposed to the former representing "the majority". This is no new secret and this is also the basket in which Lijphart places most of his theoretical eggs in measuring disproportionality (majoritarian) vs. proportionality (consensus).

What we call our second dimension of democracy has received a little less attention in the scholarly literature on representative democracy. After popular votes are tallied, depending on the numbers, parties must come together and form governments in the executive. Realistically speaking, the executive is much more important than the legislature, as this is the body that is responsible for the charting of future policy. Once the governing party or coalition takes office, it can expect its policy proposals to be accepted in the legislature. This procedure continues until support within the government or coalition begins to unravel, resulting in defection of original members, and eventually to government defeat.

Lijphart has stressed the differences between majoritarian and consensus democracies. He systematically shows how consensus democracies perform better on many indicators than majoritarian ones. On the whole, we do not disagree. We do, however, want add to this debate by highlighting an aspect of parliamentary democracies that we feel has been neglected. Though most attention concerning democracy is focused on the electoral end (legislative disproportionality), the process through which governments are formed in each country is equally as crucial to distinguishing how some countries might score higher on levels of "democraticness". While some countries may score very well on indicators of electoral representativeness (having relatively balanced proportions of seats to votes), these same countries do not always fare as well on indicators of governmental representation.

Our alternate hypothesis would challenge our data and suggest that executive exclusions like these are just the way democracy works, or perhaps that the allegedly excluded parties didn't want to participate in government in the first place (Strom, et al, 1994). However, we are making a slightly different argument. We're not directly challenging the motivation of those who form the government or those who are excluded from participating in government or Parliament at an individual decision-making level. Oddly enough, those political parties who have made conscious choices to allow or not allow parties into government often have done so for reasons of expediency, political calculation, or simply a in search for political stability, among many other reasonable motives. We are arguing instead that the understandable action of excluding marginalized or unacceptable political parties from the point of view of individual actors, may adversely affect the collective "democraticness" of these countries.

Our alternate hypothesis (Strom, et al., 1994) is grounded in methodological individualism and concentrates its focus on constraints on individual actors in the system, and not the broader issue of quality of democracy across the entire polity. By "quality" of democracy, we mean the ability of the entire society to be represented as fully as possible at both the legislative and executive levels. Strom’s approach sees institutions as only constraining structures, not as vehicles for aggregating interests and serving as a crucible for resolving contested positions. Clearly, such constraints on individual actors may be real and explain much with respect to the details of electoral outcomes or coalition formation in various countries. However, what it does not do – and apparently is unconcerned with – is the larger question on the quality of representative democracy that is independent of individual actors' choices. Such approaches simply do not measure what we are trying to explain. We believe that our data shows that the inability to represent all segments of a society along both dimensions of representation can be a systemic flaw, not just one of constrained choices. For example, Strom suggests that the French double ballot system of the French 5^th Republic was a clear and understandable attempt by de Gaulle to insure continued center-right governance by permanently dividing the left since the Cold War made the SFIO and PCF ideological rivals. We suggest that this was not only a problem of restricted choice for the individual actors, but also one of diminished democracy for the French polity.

In essence, the difference between Strom's position and ours is the unit of analysis. His looks at the nested games of individual actors within each system and explains sets of serial constraints, while our data analyzed the quality of representation aggregately along two dimensions, legislative and executive, in trying to get at the "democraticness" of different developed democracies. His "neo-institutional" position uses a more narrowly-defined conception of the term that assumes that the rational calculations of individual actors within each polity are the only way to look at this issue. Kathleen Thelen’s (1999) work would challenge his use of the term because it says nothing about the capacity of institutions to enhance and not just constrain the quality of democracy, or "democraticness" that we are stressing.

Strom’s approach is important -- as far as it goes -- but it is limited by a narrowness of vision that is characteristic of much of the methodologist individualist literature. We are using two different measures of democraticness and our central argument holds that both of these measures should allow for all political parties to come as close to representing the entire population as possible. Where we believe our contribution is most significant is to include the executive dimension of representation and then combining it to produce an aggregate index. No country is a "perfect" democracy, but our study offers a novel way to examine 15 countries and determine which of these do the best job of representing their populations along two important dimensions. In addition, we feel that these issues are salient for new and developing democracies making choices between majoritarian and consensus style systems. In reality, it may not be the case that one is necessarily better than the other in providing representation at all levels of government. They may be yet another trade-off in the choice between majoritarian and consensus systems. What is important is to determine where we want to focus the discussion, and we suggest that it may be possible to look at both dimensions at the same time. While consensus systems may perform better on legislative disproportionality indices, our data shows that some of these same systems have performed poorly on the executive disproportionality index. In these countries there has not been significant alternation of political power at the executive level of government, even though they may enjoy the virtue of having more balanced legislative representation. On the other hand, we also find that some countries, which are notorious for excluding political parties at the legislative level of government, have experienced much more executive alternation and that one party has rarely been permitted to dominate the cabinet at all times. This leads us to the conclusion that the systems which have been able to provide adequate legislative and executive representation should fall in the ‘most democratic’ category and we suspect that these are the systems which have avoided the problem of party exclusion at both levels of government.

Finally, while we generally support both the spirit and much of the empirical work of Lijphart and his "cottage industry" of consensus democracy, we also take issue with him on another significant theoretical point. Namely, his emphasis on consensus democracy seems to be much more concerned with the mechanics of consensus and much less with the quality of the democratic process. In that sense, much of his work almost assumes democracy once certain procedural and/or institutional factors are met. The strongest evidence to support this viewpoint is his "second dimension" of democracy that he emphasizes in his Patterns of Democracy (1999). Much of his earlier research concentrated on forms of electoral representation, but in this latter work – an comprehensive extension of his 1984 volume – he stresses the importance of two institutions: corporatism and an independent central bank as essential features of consensus democracy.

Clearly Lijphart intends to stress the institutional foundations of consensus democracy, an understandable and logical choice. However, in examining the essence of what representation is in democratic societies, his choice of these two institutions is a strange one. The role of corporatism and independent central banks may be responsible for improved macroeconomic performance in advanced industrialized countries. However, this raises the intriguing puzzle of how two clearly undemocratic institutions – corporatism and independent central banks – can be responsible for the success of consensus democracy. Perhaps one might make the claim that corporatism can be more or less democratic depending on how the actors within corporatism actually function, whether trade unions and works councils have active forms of rank-and-file democracy (such as in Germany) or whether they are much more hierarchical (as in Austria). However, no such internally democratic claim can plausibly be made for independent central banks. In fact, in his introductory chapter in Patterns of Democracy (pp. X-xi) he explicitly states that he is "demoting" partisan conflict and referenda as indicators and replacing them with corporatism and independent central banks.

We are asserting that this formulation may be removing the heart of democracy in favor of an elusive search for a form of government where "as many people as possible" govern, including those with disporportionate economic and political power. Lijphart was pressed to include institutions such as corporatism and independent central banks because they do seem essential for the functioning of consensus systems. However, by including such potentially undemocratic – and in the case of central banks, clearly elitist – institutions as essential for consensus democracy, we wonder how such an argument might view the German Social Democrats’ asking the undemocratic Army to "guarantee" democracy in the early years of the Weimar Republic. What does the quality of democracy mean in such a context?

We would like to suggest ultimately that solid consensus is not something that magically appears or should be striven for by dampening democratic participation, but that it results only after conflict has been successfully contested and then resolved as a part of an ongoing, dialectical process. What our data measured is the process by which this conflict is engaged and then institutionalized either well or poorly depending on the outcome of our measurements. We are not diminishing Lijphart’s emphasis of the importance of institutions, rather we are suggesting that they need to be seen in a different light than the one that he seems to emphasize in his later work. Parliamentary chambers and cabinet conference rooms are not dry, dusty institutions often dismissed as unexciting "nuts and bolts" of democratic governance. Rather, we prefer to see them in a "new institutionalist" framework closer to the approach of Thelen than that of Strom in which these institutions are the crucibles within which a "contentious" democracy must have before we can see if it is also a "consensus" democracy too.
 
 

Bibliography

Liam Anderson, (Athens, GA: Ph.D. Dissertation, University of Georgia, 2000).

Birchfield, Vicki and Crepaz, Markus M.L. "The Impact of Constitutional Structures and Collective and Competitive Veto Points on Income Inequality in Industrialized Democracies." European Journal of Political Research, 34 2: 175-200.

Cox, Gary W. (1997) Making Votes Count: Strategic Coordination in the World’s Electoral Systems. Cambridge: Cambridge University Press.

Crepaz, Markus M.L. (1996a) "Consensus versus Majoritarian Democracy: Political Institutions and their Impace in Macroeconomic Performance and Industrial Disputes."Comparative Political Studies 29:4-26.

______, (1996b) "Constitutional Structures and Regime Performance in 18 Industrialised Countries: A Test of Olsen’s Hypotheses." European Journal of Political Research, 29: 87-104.

Robert Dahl, On Democracy (New Haven: Yale University Press, 1999).

Diamond & Plattner, The Global Resurgence of Democracy (1996) 2^nd ed.

Dixon, Robert G. Jr., Democratic Representation: Reapportionment in Law and Politics (New York: Oxford University Press, 1968).

Finer, S.E. ed., Adversary Politics and Electoral Reform (London: Anthony Wigram, 1975).

Gallagher, M. (1991). "Proportionality, Disproportionality and Electoral Systems." Electoral Studies 10:1. 33-51.

Laver, Michael and Norman Schofield. 1990. Multiparty Government: The Politics of Coalition in Europe. Oxford University Press.

Lijphart, Arend., Democracies (1984).

Lijphart, Arend (1994) Electoral Systems and Party Systems. Oxford University Press.

Lijphart, Arend., Patterns of Democracy (New Haven: Yale University Press, 1999).

Mackie, Thomas T. and Richard Rose. 1991. The international almanac of electoral history. Third Edition. Washington, D.C. : Congressional Quarterly Inc.

Mershon, Carol (2000) The Costs of Coalition. Stanford: Stanford University Press.

Przeworski, Stokes, and Marin, eds., Democracy, Accountability and Representation (Cambridge UP, 1999).

Rueschemeyer, Rueschemeyer, and Wittrock, eds., Participation and Democracy: East and West (M.E. Sharpe, 1998).

Strom, Kare, Budge, Ian, and Laver, Michael J., "Constraints on Cabinet Formation in Parliamentary Democracies." American Journal of Political Science 38 2 (May) 1994: 303-335.

Thelen, Kathleen. "Historical Institutionalism in Comparative Perspective." Annual Review of Political Science 2 (1999): 369-404.

Woldendorp, J. Keman, H and Ian Budge. 1998. "Party government in 20 Democracies: an update (1990-1995)" European Journal Of Political Research. Vol 33: No 1.

pp.125-164.