9 - DOUBLE-CHECKING THE EVIDENCE (Trường Sơn)

DOUBLE-CHECKING

THE EVIDENCE

Arend Lijphart

Arend Lijphart is research professor emeritus of political science at the University of California, San Diego. He is the author of Patterns of Democracy: Government Forms and Performance in Thirty-Six Coun-tries (1999) and many other studies of democratic institutions, the governance of deeply divided societies, and electoral systems. This es-say originally appeared in the Summer 1991 issue of the Journal of Democracy.

In my article “Constitutional Choices for New Democracies,” I pre-sented systematic empirical evidence concerning the relative performance of various types of democratic systems in an effort to tran-scend the usual vague and untestable claims and counterclaims that surround this topic. I compared four parliamentary-plurality democra-cies (the United Kingdom, Canada, Australia, and New Zealand) with nine parliamentary-proportional representation (PR) democracies (Ger-many, Italy, Austria, the Netherlands, Belgium, and four Nordic countries—Sweden, Denmark, Norway, Finland) with regard to their performance records on minority representation and protection, demo-cratic quality, the maintenance of public order and peace, and the management of the economy.

I found that, where differences between the two groups of democra-cies appeared, the parliamentary-PR systems showed the better performance. There were sizeable differences with regard to minority representation (as measured by the representation of women in national parliaments), the protection of minority interests (measured by innova-tive family policy), democratic quality (measured by voter turnout), and control of unemployment; smaller differences on income inequal-ity and control of inflation; and little or no difference with regard to the maintenance of public order (as measured by riots and deaths from po-litical violence) and economic growth. Since, according to the conventional—but also rather old-fashioned—wisdom, PR may be su-

Arend Lijphart

99

perior to plurality as far as minority representation is concerned but leads to less effective decision making, even my finding of minor or no differences on some of the performance indicators must be counted in favor of the parliamentary-PR type.

Guy Lardeyret and Quentin L. Quade, both eloquent exponents of this conventional wisdom, raise a series of objections to my analysis and conclusions—very welcome challenges because they present an opportunity to double-check the validity of my evidence. Lardeyret and Quade argue that 1) the differences in governmental performance may be explained by other factors than the type of democracy, and hence that they do not prove any parliamentary-PR superiority; 2) that, when other important effects of the different types of democracy are considered, plurality systems are superior; 3) that some of my findings are the result of incorrect measurement; and 4) that my findings are biased by my choice and classification of the countries included in the analysis. I shall demonstrate, however, that whenever their objections can be tested against the facts, they turn out to be invalid.

Alternative Explanations

I agree with Lardeyret’s and Quade’s argument that economic suc-cess is not solely determined by government policy; I said as much in my original article. There are obviously many external and fortuitous factors that influence a country’s economic performance. Neither do I disagree with Quade’s argument that several special circumstances have had a negative effect on Britain. On the other hand, some of the PR countries suffered similar setbacks: The Netherlands and Belgium also lost sizeable colonial empires, the “seismic social-psychological” shock of decolonization suffered by Britain was no greater than the shock of defeat and division suffered by Germany, and ethnic strife has plagued Belgium as well as the Celtic periphery of the United Kingdom. But my comparison was not just between Britain and one or more PR countries; I compared the four parliamentary-plurality democracies as a group with the group of nine parliamentary-PR countries. I assumed that when the economic performance of groups of democracies is examined over a long period of time, and when all of the countries studied have similar levels of economic development, external and fortuitous influences tend to even out. In the absence of any plausible suggestion that, as a group, the parliamentary-PR countries enjoyed unusual economic advantages from the 1960s through the 1980s—and neither Lardeyret nor Quade offers any such suggestion—my assumption and hence my findings concerning differences in economic performance remain valid.

Lardeyret and Quade do mention a few things that might provide a basis for alternative explanations: the special characteristics of the Nor-dic countries, the advantage of having a constitutional monarchy, the

100 Double-Checking the Evidence

difference between moderate and extreme PR, and the advantage of U.S. military protection. All of these can be tested empirically. Lardeyret claims that unemployment in the Nordic countries is underestimated because of “highly protected jobs” and that income inequality is rela-tively modest because of unusual handicaps that conservative parties must contend with in these countries. Whether these factors change my findings can be checked easily by excluding the Nordic countries and comparing the non-Nordic parliamentary-PR countries with the parlia-mentary-plurality countries. Average unemployment in the Nordic countries was indeed lower than in the non-Nordic countries—2.7 per-cent compared with 5.7 percent—but the latter percentage is still slightly better than the 6.1 percent for the parliamentary-plurality countries. As far as income inequality is concerned, there is virtually no difference between the Nordic and non-Nordic parliamentary-PR countries—39.0 and 38.9 percent, respectively—both of which score lower than the 42.9 percent in the parliamentary-plurality democracies.

When we compare monarchies with republics, the first point to be made is that, if a constitutional monarchy is an advantage, all of the parliamentary-plurality countries enjoy this advantage, whereas only about half of the parliamentary-PR democracies do. Second, when we compare the monarchical countries (Belgium, the Netherlands, Sweden, Norway, and Denmark) with the republican PR countries (Germany, Italy, Austria, and Finland), their growth rates are virtually identical and their inflation rates exactly the same. Only their unemployment rates differ somewhat: the monarchies have a 4.0 percent average unemployment rate compared with 4.9 percent in the nonmonarchical countries; again, the latter percentage is still better than the 6.1 percent average of the parliamentary-plurality countries. On all of the indicators of minority representation and protection and of democratic quality, there are slight differences between the monarchical and non-monarchical groups, but both still clearly outperform the parliamentary-plurality countries.

Is PR’s Achilles’ heel revealed when we focus on the countries that have extreme PR (Italy, the Netherlands, Denmark, and Finland) and contrast these with the more moderate PR systems (Germany, Sweden, Norway, Belgium, and Austria)? The empirical evidence disproves this. The inflation and unemployment rates in the extreme PR group are indeed higher (7.4 and 5.5 versus 5.4 and 3.6 percent) but still at least a bit lower than the 7.5 and 6.1 percent in the parliamentary-plurality systems; their growth rates are virtually identical. On the four indica-tors of representation and democratic quality, the differences are slight, and both groups of PR countries remain way ahead of the parliamen-tary-plurality countries. My own firm preference remains for moderate PR, but the dangers of extreme PR must not be exaggerated.

As Quade correctly states, the parliamentary-PR countries have had the advantage of living under “the umbrella of American military pro-

Arend Lijphart

101

tection”—but so have all four of the parliamentary-plurality countries. In fact, the only slight exceptions are in the PR group: Sweden’s neutral but strongly armed posture entailed heavy military expenditures, and Finland lived in precarious dependence on Soviet restraint. On the whole, however, American military protection benefited all 13 parliamentary democracies more or less equally, and therefore cannot explain any differences in their performance records.

Alternative Standards and Classifications

Partly in addition to and partly instead of the measures that I used to evaluate the performance of different types of democracy, Lardeyret and Quade state that democracies should be judged in terms of factors like accountability, government stability, decision-making capacity, and the ability to avoid “repeated elections.” There are several prob-lems with these suggestions. First of all, while accountability is certainly an important aspect of democratic government, it cannot be measured objectively. Second, it is not at all clear that coalition governments are less responsible and accountable than one-party governments. Quade’s description of coalition cabinets as governments “cobbled together out of postelection splinters by a secretive process of interparty bartering” may apply to a few exceptional cases like Israel (which combines ex-treme PR with an evenly split and polarized electorate), but for most PR countries it is a grossly overdrawn caricature. In fact, once they are formed, coalition cabinets tend to be a good deal less secretive and more open than one-party cabinets.

Third, government stability can be measured in terms of average cabinet duration. On the basis of previously collected figures, my cal-culation shows that the average cabinet life in the parliamentary-plural-ity countries is about twice that in the parliamentary-PR systems.1 Longer cabinet duration, Lardeyret assumes, means greater decision-making strength because of greater continuity in government personnel. But when coalition cabinets change they usually do not change as much as the radically alternating cabinets in the parliamentary-plurality coun-tries. Lardeyret admits this when he complains about the “long tenures in office for fixed groups of key politicians” in the PR countries. Fourth, if Lardeyret is right about the superior decision-making capacity of parliamentary-plurality governments, the only convincing proof is that their decisions result in more effective policies. This brings us back to the evaluation of government performance in terms of successful mac-roeconomic policy making and the successful maintenance of public order. As we have already seen, this hard evidence does not show any parliamentary-plurality superiority.

Lardeyret’s complaint about unnecessarily frequent elections in the parliamentary-PR systems suggests an additional useful measure of demo-

102 Double-Checking the Evidence

cratic performance—and one that, happily, can be measured and tested easily. In the 29-year period from 1960 to 1988—the same period for which two of the three OECD economic indicators were collected—the parlia-mentary-plurality countries conducted an average of 10.0 national legisla-tive elections, compared with an average of 8.8 in the parliamentary-PR countries.2 The frequency of elections is actually smaller in the PR sys-tems, contrary to Lardeyret’s assertion, although the difference is slight. However, Lardeyret’s hypothesis is clearly disproved by this simple test.

Lardeyret and Quade have only a few disagreements with my mea-surements. One question that Lardeyret does raise is the measurement of voter turnout: The U.S. voter-turnout figure would be considerably higher if counted as a proportion of registered voters. He is quite right on this point, but all of my turnout figures are percentages of eligible voters— which means that all countries are treated equally. Moreover, if turnout figures are used as a measure of democratic quality, the low figure for the United States accurately reflects not only an unusually high degree of political apathy but also the fact that voting is deliberately discour-aged by the government by means of onerous registration procedures.

Quade questions my equation of “the number of women in legisla-tures with representation of women’s interests.” But I did not equate the two at all: I used a separate measure (the innovativeness and expansive-ness of family policy, which is of special concern to women) to test whether women’s interests were actually better taken care of in the PR countries—and I found that this was indeed the case.

Finally, Lardeyret questions my use of Robert Dahl’s ratings of demo-cratic quality because of their alleged pro-PR bias. I already admitted a slight bias of this kind in my original article, but I decided to use the Dahl ratings anyway since they are the most careful overall ratings that are available. However, since they are obviously less objective than my other indicators, I shall not insist on their being used as evidence.

Quade criticizes my favorable judgment of the parliamentary-PR com-bination by pointing out some examples in which PR did not work well, especially the two cases that are often regarded as spectacular failures of democracy: the Weimar Republic and the French Fourth Republic. Nobody can disagree with the assessment that the Weimar Republic was a failure, but it is less clear that PR was the decisive factor or that plurality would have been able to save Weimar democracy. Moreover, Weimar was a semi-presidential rather than a parliamentary system. In France, the Fourth Republic indeed did not work well, but a reasonable argument can be made that relatively small reforms within the parlia-mentary-PR framework might have cured the problems and that the radical shift to semi-presidentialism and away from PR was not abso-lutely necessary. And examples of PR failures can be matched by examples of the failure of plurality systems, such as the failed democra-cies of West Africa. Sir Arthur Lewis, who served as an economic advisor

Arend Lijphart

103

to these governments, became convinced that “the surest way to kill the idea of democracy” in these divided societies “is to adopt the Anglo-American electoral system of first-past-the-post [plurality].”3

Lardeyret does not question my focus on stable contemporary de-mocracies, but argues instead that some of these countries should have been classified differently. Although France is neither fully presiden-tial nor fully plurality, I accept his suggestion that it is close enough on both counts to be classified alongside the United States. I agree that Spain and Portugal belong in the parliamentary-PR category, but com-parable data are lacking since the two countries were not yet democratic during the full period covered by the empirical evidence. I disagree that Germany lacks PR and should be classified as a plurality system; it is almost entirely PR in terms of how Bundestag seats are allocated to the parties, though its 5 percent threshold makes it a moderate PR system.

But let us concede Germany to the plurality category; my analysis still stands. Lardeyret’s counter-hypothesis is that in “the order of rank according to standards of both efficiency and democracy,” the two plu-rality systems (parliamentary and presidential) are ahead of the parliamentary-PR systems. This can be tested by comparing the seven plurality systems (the parliamentary-plurality countries plus the United States, France, and, arguably, Germany) with eight PR systems (all of the parliamentary-PR systems except Germany). Thus reclassified, the PR countries still have the better record with regard to control of unem-ployment (4.6 percent versus 5.5 percent average unemployment) and do not differ much with regard to growth (3.5 versus 3.4 percent) and inflation (6.6 versus 6.5 percent). On the indicators of minority repre-sentation and protection and of democratic quality, the PR countries are still far ahead of the plurality systems: 17.5 versus 4.5 percent women in parliament; a score of 8.0 versus 4.4 on family policy; 84.5 versus 73.5 percent on voter turnout; and 38.9 versus 41.9 percent of total income earned by the top 20 percent of households. The evidence clearly disproves Lardeyret’s counter-hypothesis.

Choices and Changes

The demonstrable advantages of parliamentarism and PR appear to be appreciated by the citizens and politicians of democratic countries. In many, if not most, presidential countries, there is widespread dissat-isfaction with the operation of presidentialism and sizeable support for a shift to a parliamentary form of government; the contrary sentiment can be found in hardly any parliamentary democracy. Similarly, there is great unhappiness about how plurality elections work and strong senti-ment for a shift to PR in most democracies that use plurality, but few calls for plurality in PR countries. One important reason for this pattern is that the divisive, winner-take-all nature of plurality and presidentialism

104 Double-Checking the Evidence

is widely understood. From the turn of the century on, democracies with ethnic or other deep cleavages have repeatedly turned to PR in order to accommodate such differences. Lardeyret’s recommendation of plural-ity elections for South Africa and other deeply divided countries is therefore particularly dangerous.

Another important reason for PR’s popularity is the feeling that dis-proportional election results are inherently unfair and undemocratic. None of postwar Britain’s governing parties was put in power by a majority of the voters; all of these parties gained power in spite of the fact that most of the voters voted against them. Lardeyret’s and Quade’s opinion that electoral disproportionality is unimportant is simply not shared by most democrats. As a recent editorial in the Economist puts it, “since the per-ception of fairness is the acid test for a democracy—the very basis of its legitimacy—the unfairness argument overrules all others.”4

Fundamental constitutional changes are difficult to effect and there-fore rare, but the prevailing pattern of democratic sentiment makes shifts from plurality to PR more likely than the other way around. The reason for this is not, as Lardeyret suggests, that “it is almost impossible to get rid of PR, because doing so requires asking independent parties to co-operate in their own liquidation.” On the contrary, this is the main reason why the big parties that benefit from the plurality rule will try to keep it. In PR systems, the large parties usually have enough votes to shift to a system that would greatly benefit them, especially because, as Lardeyret correctly observes, the electoral system is “curiously omitted in most [written] constitutions.” That they rarely try to do so cannot be explained in terms of narrow partisan self-interest; the feeling that scrap-ping PR is undemocratic and dangerous plays a major role. Both the empirical evidence and the weight of opinion in existing democracies make a strong case for the proposition that PR and parliamentarism are also the wisest options for new democracies.

NOTES

1. Arend Lijphart, Democracies: Patterns of Majoritarian and Consensus Gov-ernment in Twenty-One Countries (New Haven: Yale University Press, 1984), 83. A cabinet is defined as the same cabinet if its party composition does not change; on the basis of this definition and for the 1945–80 period, average cabinet life in the four parliamentary-plurality countries was 88 months and in the parliamentary–PR countries, 44 months.

2. The dates of parliamentary elections for the 13 countries can be found in the respective country chapters of Thomas T. Mackie and Richard Rose, The Interna-tional Almanac of Electoral History, 3rd ed. (London: Macmillan, 1991).

3. W. Arthur Lewis, Politics in West Africa (London: Allen and Unwin, 1965),

71.

Economist, 11 May 1991, 13.