Home

Erik Moberg ©
 

Contents:

Powell's theme
Majoritarianism and proportionalism
Responsiveness in the emergence of policymakers
Representational congruence between voters and policymakers
My objections to Powell's results
The influence of the opposition in proportional systems
Another view of politics in proportional systems
Summary
Bibliography
 
 
 

Powell's theme

In his new book Elections as Instruments of Democracy Bingham Powell deals with an extremely important topic, namely the citizens' influence on governmental policies in a number of modern democracies. He deals with the effectiveness of the influence, and to some extent also with its mechanisms. But it is not only Powell's theme that is important; his treatment of it is also impressive. The arguments are straightforward and, on the whole, theoretically connected - his ambition to build theory is obvious. Furthermore his marshalling of relevant empirical data is quite compelling. All of this together clearly makes the book an important work in political science. Still some parts of Powell's work fail to convince me. In particular this is the case with his analysis of proportional systems. Before developing my own position I shall however present Powell's reasoning in some detail.
 
 

Majoritarianism and proportionalism

Obviously a discussion like Powell's must use some kind of classification of democracies, and Powell's main distinction is the one between majoritarian and proportional democracies. A proportional democracy is first of all characterized by proportional elections to the legislature - normally constitutionally stipulated. This is however not enough. Although the government usually represents only around half the legislature, there may nevertheless be rules, and customs, which give real influence to the opposition. All parties may for instance be proportionally represented in the committees of the legislature, the chairmen of the committees may come from all parties, and the committees may have substantial power. If this is so, and if further mechanisms with similar effects are operating, the ideas of proportionalism are permeating not only the elections to the legislature, but also the policymaking thereafter. Due to the combined effects of election rules, and policymaking rules, all groups in the society will thus be influencing the policymaking at all levels in proportion to their size. The power is dispersed. Such a system, according to Powell, embodies the proportional vision.

A majoritarian democracy, emanating from the majoritarian vision, is quite different. Here, at first, the elections to the legislature are majoritarian - the first-past-the-post system for instance - rather than proportional. Again, though, this is not enough. The same principles which govern the elections to the legislature are operating at higher levels as well. Thus the opposition is kept out from the policymaking rather than invited to take part. The power is concentrated to the incumbents who are, on purpose, left alone with the responsibility of governing. One important, or perhaps the important, rationale for this, in the majoritarian vision, is that it facilitates accountability and mandate. If the governing is bad it is easy to identify those responsible, and to hold them accountable by voting them out. Similarly it is easy to identify the opposition, and to give it a mandate for executing its alternative program, by voting it in. It may also be said, as Powell does, that the voters control the policymaking in this kind of system, whereas they influence it in a proportional system.

Powell's investigation covers all democracies for which relevant data have been available for roughly the last twenty-five years. These are the twenty countries in table 1 divided into three groups according to Powell's classification (p 41). During the period studied about 155 elections have been held in these countries (p 3).
 
 
 

Predominately Proportional Mixed Predominately Majoritarian
Austria
Belgium
Denmark
Finland
Germany
Italy
Netherlands
Norway
Sweden
Switzerland
Ireland
Japan
Spain
USA
Australia
Canada
France
Greece
New Zealand
United Kingdom
Table 1

Most of the countries, it should be noted, are parliamentarian. Only one country, the US, is purely presidential. Furthermore, some of the countries are not clear-cut in this respect. France, for instance, represents a mixture of parliamentarism and presidentialism. And some countries have changed their constitutions in relevant ways recently. The classification of New Zealand, for instance, applies to the time before 1996, when proportional elections were substituted for majoritarian ones.

Incidentally, it should by now be clear that Powell's basic distinction between majoritarianism and proportionalism has considerable similarities with Arend Lijphart's distinction between majoritarianism and consensus, and this connection is explicitly mentioned by Powell (p 21). Still, as Powell also makes clear, the two distinctions are not exactly the same (p 266, notes 10, 13 and 16).
 
 

Responsiveness in the emergence of policymakers

Powell studies the relationship between the voters, on the one hand, and the selected influential policymakers, on the other, in two different ways. The first one, which is described in this section, focuses on selection patterns and decision making mechanisms, whereas the other one, which is dealt with in the next section, is concerned with the political opinions of the policymakers in relation to those of the voters.

When studying the selection patterns Powell assumes that the voters vote for political parties and their candidates. In this process, generally speaking, some voters may succeed well in getting represented by influential policymakers, whereas other may be less successful. In Powell's terminology (p 97) the authorized representation may vary - it may be high or low. This being so, it is natural to ask the normative question about the ideal pattern of representation, and also the positive question about what actually happens in reality.

As for the first question it is obvious that the majoritarian and proportional ideals differ considerably. The proportional ideal is that the authorized representation is high, i.e. that all voters are represented by influential policymakers in proportion to the size of their parties. In essence this means not only that all parties should be represented proportionally in the legislature, but also, and even if the government is formed on a narrower base, that there should be rules which give influence to the opposition parties in the current policymaking process. The majoritarian ideal, on the contrary, is that only those voters who, in a context of two parties, vote for the majority party, should be represented by policymakers with real influence. There may, of course, be representatives of the minority party in the legislature, but ideally they should have no roles in policymaking. Or, in Powell's words (p 125): "... the specific expectations of each vision are dissimilar: the proportional vision assumes a steady increase in share of government or policy making as a party wins more voter support; the majoritarian vision assumes an all-or-nothing switch as a party crosses the line of 50 percent backing from the electorate."

In his examination of the realization of these ideals in different constitutional systems Powell uses not only empirical data, but also coefficients indicating the influence of the opposition in various situations. I will discuss these coefficients further in the section "The influence of the opposition in proportional systems". Here I will just take them for granted and relate Powell's main finding, which is that each type of system succeeds well according to its own norms. The majoritarian ideal is thus, on the whole, well realized in countries with majoritarian constitutions, and the same is true for the proportional ideal in countries with proportional constitutions. "(W)e find", says Powell (p 154), "that each type of design is relatively successful on the terms established by its own vision, much less so by the alternative criterion." Furthermore Powell reports that if a system fails to realize its own ideal completely, the ideal of the other system is usually realized to some extent. There is thus a kind of trade off between the majoritarian and proportional ideals. So far everything is therefore quite good. Mostly some ideal is realized, and mostly that ideal is the one which should be realized. There is, however, one important exception from this general pattern.

This exception, when it appears, does so in majoritarian systems. In such a system, as is well known, it is perfectly possible, although not very likely, that the party which get most votes, that is the majority party, becomes the smaller party in the legislature. If so the government, and all influential policymakers, will be recruited from the minority party. This clearly violates both majoritarian and proportional ideals. In his empirical findings Powell reports a number of cases like this. He talks about "unmitigated failures" and an obvious "design flaw in majoritarian systems" (p 148).
 
 

Representational congruence between voters and policymakers

When comparing the political opinions of the voters and the influential policymakers Powell uses all through the left-right scale, and he focuses on the median voter position. The reason is the normative relevance of this position. "Staying at purely the normative level", says Powell (p 164), "the position of the median voter thus seems privileged over any other position in both majoritarian and proportional visions. It is the position that can defeat any other position in a simple majority vote; that minimizes the number of opposing voters."

For each country Powell measures essentially two things. First, he measures the distance between the voters' median position on the left-right scale and the government's position on the same scale. Second, he measures the distance between the voters' median position and the average position of all influential policymakers taken together. If there are no influential policymakers outside the government, as may be the case in a majoritarian system, the two distances are, of course, the same. If, however, parties outside the government also take part in policymaking, as could be the case in proportional systems, then the two distances differ. Anyway, it is the distance between the median voter position, and the average position of the influential policymakers, which Powell uses in his definition of representational congruence (p 169). The smaller this distance is, the greater (or better) is the congruence.

For his measurements Powell uses different kinds of empirical data. As for the voters he uses results from surveys in the different countries, in which the voters have been asked to place themselves on a left-right scale with ten positions. From these data it is of course easy to find the median position, it is just a matter of counting. For the policymakers he uses the results of investigations in which experts on the different countries have been asked to place the political parties on the same scale as the one used for the voters. For the government the position is obtained by just taking the average of the positions of the member parties, weighted by their shares of seats in the lower house of the legislature (p 173). In a majoritarian system, as we have seen, this may also be the position of all influential policymakers. Furthermore, if the government in such a system consists of a single party, as it sometimes does, the position of that particular party constitutes the government position. In a proportional system things are usually different. First, since the government in such as system usually consists of several parties, all of these parties should be taken into account. Moreover the opposition parties, which may have considerable influence, should also be taken into account. Powell does this by considering the parties' positions on the scale as given by the experts, the parties' sizes, and, again, their coefficients of influence (p 173).

Now, Powell's first finding (p 222, figure 9.3) is that the distance between the median voter and the governmental position is somewhat greater in majoritarian systems than in proportional ones. This is hardly what one should be expect. Rather, as Powell says (p 171), "the majoritarian model predicts good congruence", whereas the proportional model, even if median outcomes are possible, "may also lead to policies somewhat off the median" (p 172). Powell's explanation for his finding (p 221) is that "The majoritarian vision works only fitfully. The simplifications and distortions necessary to create decisive elections frequently exact a heavy cost in distance between the citizen median and the position of the government." This explanation, which to me seems plausible, is, I think it is fair to say, more in terms of deficiencies in majoritarian systems than in terms of merits of proportional ones.

Powell's next finding (p 222, figure 9.3), and main finding with respect to representational congruence I should say, is that the distance between the median voter and the position of all the influential policymakers is substantially greater in majoritarian systems than in proportional ones. The representational congruence is thus much better in proportional systems than in majoritarian ones. Powell (p 217) talks about "the superior performance of the proportional designs in comparison to the majoritarian ones, especially in creating good congruence between citizens and policymakers." He also says (p 18) that "(t)he proportional influence designs enjoy a surprising advantage."
 
 

My objections to Powell's results

We may thus summarize Powell's results as follows. With respect to the responsiveness in the emergence of policymakers he concludes that, on the whole, and usually, both majoritarian and proportional systems perform satisfactorily. Majoritarian systems do however sometimes fail seriously by giving the power to the minority rather than the majority party. For this reason, and in this respect, Powell concludes that proportionalism is the better system. With respect to the representational congruence between voters and policymakers he again finds proportionalism better. The distance between the median voter position, and the average position of all influential policymakers, is much smaller in proportional systems than in majoritarian ones. Proportionalism is thus the far better system in this respect. Altogether, and taking both aspects into account, Powell comes out strongly in favor of proportionalism.

My first comment to this is that I find Powell's analysis of majoritarian systems, and their deficiencies and fitfulness, interesting and valuable. By reading this I, for one, have certainly become less inclined to believe that these systems necessarily tend to give policy results very close to the median voter. I think I have got a better understanding of these systems and I appreciate that.

Having said that, I have now come to my main objection to Powell's book, which is about his analysis of proportional systems. First, I think that Powell severely exaggerates the authorized representation of these systems. They do not, I submit, embody the proportional ideals to the extent that he affirms. Second, I think that Powell severely exaggerates the systems' representational congruence. In my opinion they are not as superior to majoritarian systems in this respect as he says. These two things together mean that the merits of proportional systems are considerably smaller than Powell holds. In addition to this I also think that proportional systems have other important properties, which perhaps are less attractive, and which Powell ignores. In the next section I will give my reasons for saying that Powell overestimates the systems' merits, and in the section after that I will bring in the properties ignored by Powell.
 
 

The influence of the opposition in proportional systems

My critique of Powell's praise for proportionalism emanates from one single, simple notion. He seriously overrates, I think, the influence of the opposition in proportional systems. When saying so I am however well aware of our limited knowledge about these matters. Powell writes (p 93) that "(e)mpirically it is very important ... to determine the extent to which the opposition is effectively represented in policy making. As we have rather little research in this area to guide us, it is necessary to be inventive." On this particular point I agree, and I will therefore try to comment constructively.

When dealing with the influence of the opposition Powell considers essentially two factors, which he describes like this (p 103f): "One is the political strength of the government. The great contrast is between majority and minority governments. Government majorities usually need bargain only with their own factions. Minority governments must bargain with the opposition. The second major factor is structural - the nature of the committee system. A committee system that largely rubber stamps the government's proposals offers little role for opposition. A committee system that is a genuine area for policy debate and modification offers important scope for meaningful representation." In order to describe Powell's method for taking these two factors into account I will use table 2, which shows the responsiveness in the emergence of policymakers, or the authorized representation, in some countries.
 
 

Country
Share of
electorate
supporting
the govern-
ment, %
Effective
represen-
tation of the
opposition
Authorized
represen-
tation
Finland
60
24
84
Switzerland
78
5
83
Denmark
35
47
82
United States
51
26
77
Sweden
46+4
21
71
Germany
56
15
71
Austria
53
15
68
Canada
36
32
68
Netherlands
52
16
68
Norway
41
27
68
       
       
United Kingdom
42
4
46
New Zealand
39
4
43
Table 2

The table consists of parts of a big table entitled "Contributions to Effective Representation after the Election, Twenty Democracies in the Early 1980s" (p 110). I have omitted some columns in the original table and rephrased the headings of the remaining columns somewhat. The second column in the table here shows the percentage of voters who, in the election considered, supported those parties who formed the government after the election. The third column, which is the really interesting one in this context, and to which I will return, shows the influential part of the opposition as determined by Powell. The fourth column, finally, shows the sum of the figures in the two former columns. These figures indicate the authorized representation as defined in the section "responsiveness in the emergence of policymakers" above. In principle the figures can take values between 0 (for dictatorships) and 100 (which implies a government formed of all the parties that received voting support). The countries in the table are ordered according to the figures in the fourth column. I have included the first ten countries in Powell's table, and the last two ones. Eight countries in the middle of Powell's table are omitted. For clarity I should add that Powell, as the reader has perhaps already noticed, in his calculations sometimes deals with shares of the electorate, and sometimes with shares of the legislature. This distinction is however not important here, in particular as we are interested in proportional systems for which the two shares usually are of roughly the same size. It is therefore not necessary to go into the details of this aspect of Powell's method.

Now, let us look at the important third column. Powell arrives at the figures there by multiplying, for each country, the opposition's share of the legislature, expressed in per cent, with a sum of two coefficients. These coefficients are related to Powell's two factors mentioned above, and represent the opposition's influence via bargaining with the government, and via the committee system, respectively. Let us consider Denmark as an example. For this country the coefficient for the opposition's influence via bargaining with the government is determined to 0.5, and the coefficient for influence via the committee system to 0.25. Then, when multiplying the sum of these two coefficients, 0.75, with the opposition's share in the legislature after the election considered, which is 62 %, we get 47. This is the figure for Denmark in the third column above.

Obviously the two coefficients are of crucial importance, and hence also the methods by which they are determined. I will start with the coefficients for the opposition's influence via the committees. Here, Powell's method is quite simplistic. "I assigned", he says (p 106), "a weight of .25 if the legislature has both strong committees and divides the committee chairs among all the larger parties, not just the governing parties". And (p 106) "I assign a weight of .125 if the legislature has a mixture of strong committees chaired by the governing parties or weak committees with shared committee chairs". Powell thus operates with just two distinct coefficients here, and obviously it is in order that the former coefficient is greater than the latter one. Still Powell presents no arguments for the absolute values of the coefficients but rather comments (p 106) that "(a)dmittedly, the specific weights are somewhat arbitrary". It is easy to agree.

The examining of Powell's coefficients for influence via bargaining with the government has to be more elaborated. Here he uses the three coefficients in table 3 (p 105f).
 
 

the opposition's influence in case of a minority government 0.5
the opposition's influence in case of a supported minority government 0.2
the opposition's influence in case of a majority government 0.1
Table 3

Now, when determining coefficients like these it is helpful to start from an idea about what they stand for. This is however not perfectly clear, even if the general intention is obvious. Sometimes Powell talks vaguely about things like "potential bargaining influence weight " (p 116), but sometimes he is also more precise. The closest he comes to an exact meaning is when he says (p 105) that "(i)n assigning a .5 probability to the opposition parties, I assume that the average representative of an opposition party facing a minority government is half as likely to have influence as a representative of a governing party." His explanation for a corresponding figure for the government, 1.0, is also illuminating. He says (p 105) that "(t)he government parties can at least negotiate on every issue; they will probably be part of a nearly every winning coalition" and (p 106) that "the government parties ... will always be able to give their supporters substantial influence on policy making." Hence the figure 1.0, which is the government's probability of participation.

Using these clues to the meaning of the coefficients I will now consider a simple example of the first type of opposition in table 3, the one standing against a pure minority government. I will assume that we have a legislature which represents the electorate proportionally. The legislature has 100 seats, out of which 44 belong to a few major parties which together form the government. Since there are no non-governmental parties which regularly support the government, all the remaining 56 seats belong to the real opposition. I will assume that this opposition consists of eight parties of equal size, which means 7 seats each. I furthermore assume, as Powell does, that a decision usually requires a simple majority. The support of one of the opposition parties is thus enough for the government for reaching a decision, since that will give a majority of 51. If there are no differences between the opposition parties, which would make some of them more likely supporters of the government than the other ones, then the probability for joining the government in decisions will be the same for each opposition party. In this case this means a probability of 1/8 or 0.125. Now, this figure, I submit, is the coefficient representing the real influence of an opposition party in the situation described. This coefficient is obviously much smaller than Powell's corresponding coefficient of 0.5.

These two coefficients, Powell's and mine, may now be used for deriving results for a table of the same kind as table 2 above. For each of the two cases I shall thus deduce figures for the authorized representation. For simplicity, and for making the calculations clear, I will only consider influence via the government and thus disregard influence via the committee system. Doing so, and remembering that the government has a support of 44 %, and that there are 8 opposition parties, each supported by 7 %, Powell's authorized representation is 0.44´ 1.0 + 8´ 0.07´ 0.5 = 0.72. The result in the fourth column of a table like table 2 would thus be 72. With my coefficient of 0.125 we rather get 0.44´ 1.0 + 8´ 0.07´ 0.125 = 0.51. This is obviously much lower than 0.72 and, in fact, not higher than the 51 % just needed for taking a decision. This last aspect is interesting since it means that the opposition's influence is limited to what is necessary for taking decisions. The coefficient 0.125 thus gives a result in agreement with William Riker's theorem about minimal winning coalitions. This, of course, depends on my way of deriving the coefficient, but it is still worth noticing.

Now, leaving the numerical example above, we may consider the relationships involved more generally. Let us again assume that we have legislature which represents the electorate proportionally, and that the decisions in that legislature are taken by simple majority (represented by the share 0.51). Let us also assume that we have a minority government supported by the share G of the legislature, and that the probability for being participants in decisions is the same for all legislators outside the government. If so, following the logic in the example above, the coefficient of influence for the opposition is:

(0.51 -  G) / (1 -  G)

In words this means that we get the coefficient we are looking for by dividing the part of the opposition, which is needed as a complement for making the government decisive, with the total opposition.

In order to see how the formula works we may start by using the value for G in the example above, namely 0.44. It is easily checked that we get, as we should, a coefficient of 0.125. Let us then take a rather big G, say 0.5. In that case the coefficient becomes very small, namely 0.02, which reflects the fact the complement which the government needs in order to become decisive, is minimal. Let us then imagine a very small G, say 0.02, which means that the legislature almost entirely consists of policymakers in opposition. In this extreme case the opposition's coefficient of influence becomes 0.50. One thing will, however, remain constant in all these examples, namely the authorized representation of 0.51. This, of course, should cause no surprise by now. Even so, however, and since parliamentary parties usually behave as indivisible lumps, the authorized representation may be somewhat higher in real situations.

My main conclusion of this reasoning is that Powell's coefficients in table 3 for influence via the government are much too high. I have focused on his first coefficient of 0.5 in that table, and it is only in the rather extreme case with a virtually non-existent government that the logic presented here gives a coefficient as high as Powell's. This logic applies to the other coefficients as well. It implies, for instance, that the coefficient in the third case with a majority government should be 0 rather than 0.1. Now, I think that this logic is quite compelling, and that deviations from it at the very least should be well argued. There are however no such arguments in Powell's book.

Rather, as a validation, he invokes (p 107f) an independent investigation by Laver and Hunt in which they have asked experts to assess the impact of the opposition on government policy in a number of countries. The experts could give answers on a scale from 1 (no impact) to 9 (high impact)."The comparison of the Laver and Hunt results with my analysis is remarkably reassuring", says Powell. In order to be able to compare the results he must however transform the Laver-Hunt scale to his own, and in so doing he assumes that "high impact" means "as much impact as parties in the government". This, it seems to me, is quite extreme. Since, after all, an opposition is an opposition, I think that most people would consider its impact high even if it was considerably lower than the government's. In fact, as Powell also says, "(w)e cannot, of course, be sure just what the respondents had in mind by 'high impact'...". This fact, I think, reduces the value of the validation considerably.

Here it is interesting to note that Powell, in a context where he has presented some data at odds with the median voter theorem, still affirms (p 183) that "(i)t would be foolish to write off in general the powerful center-seeking incentives that do operate in clearly two-candidate situations." Similarly, I think, it is prudent always to keep Riker's theorem about minimal winning coalitions in mind when discussing politics in proportional systems. Surely Powell alludes to Riker in a few places, as for example when he says (p 149) that "(m)inority and 'oversize' governments ... tend to be less durable than their minimal majority counterparts". Still I have a feeling that Riker to a large extent is forgotten.

I will now return to table 2. All ten countries in the top of that table except the United States and Canada are, as we saw in table 1, classified by Powell as proportional, whereas the two bottom countries are majoritarian. Now, the figures in the last two columns of the table, for the ten top countries, are quite high. This is important, since these figures constitute the evidence by which Powell supports his claim of high authorized representation in proportional democracies. It is these figures which make possible his conclusions that proportional systems behave so well according to their own norms. As I see it, and as I have tried to show above, the figures are however severely inflated. Powell's conclusion about the effective emergence of policymakers in proportional systems is thus, in my view, not valid.

Powell's result about the very good representational congruence in proportional systems suffers, as I see it, from the same deficiency, since it depends on the same inflated figures. When Powell determines the average left-right position of all influential policymakers he uses the same coefficients as when determining authorized representation. This is clear from his description of his method (p 173). Thus, my conclusion is that these results are unsubstantiated as well.
 
 

Another view of politics in proportional systems

In this section I shall indicate some aspects of proportionalism which I consider important, and which are neglected by Powell. What I am thinking about is a tendency for politics in proportional systems to focus on special interests rather than on ideological issues of common or general relevance. This tendency may be further clarified by distinguishing between two conceptions of politics which usually, although they are quite different, are not kept apart (the two conceptions are described in more detail in Moberg, 2000).

The basic tool in the first conception is the spatial model. Within that model political positions are described by means of points on scales - for instance points on the ideological left-right scale. These positions are the subject matter of politics according to this conception, it is them that politics is about. Redistributions of resources among citizens may occur, if implied by the positions, but they are by no means a necessary part of the conception. The main theoretical result within this conception of politics is the median voter theorem. The other conception is rather based on zero-sum game theory. This conception is unthinkable without redistributions - they are a necessary part of it. Some groups of citizens may, for instance, use the majority rule for taxing other groups and share the proceeds. The main result within this conception is Riker's theorem of minimal winning coalitions. Now, my point is that politics in parliamentary proportional systems, at least to a large extent, has to be described according the latter conception. Still Powell, as I understand him, relies almost entirely on the former conception.

This may be illustrated by Powell's remarks about the bargaining which regularly takes place in proportional multi-party systems. In many places he talks about bargaining between policymakers, and about bargaining associated with government formation in particular (p 170, table 7.2). He says, however, very little about what the bargaining is about. When he says something he gives the impression that the bargaining just is about positions on the left-right scale. In a context mainly dealing with slightly different matters (p 173) he mentions for instance a government's "chosen position" on the left-right scale. It is easy to interpret this as the position chosen by the government itself, as a result of the bargaining preceding the formation of the government. In other places though, it is only fair to say, he rather professes ignorance. He says, for instance (p 92), that "... we do not know a great deal about the way government coalition partners make substantive policy decisions among themselves, an important problem for empirical research." I agree, and I will just add that theoretical research may be equally urgent.

In fact, my point here about bargaining may be substantiated with a theoretical result of my own, namely a theorem saying that it is impossible to account for the existence of a coalition government within the framework of a one-dimensional spatial model. The reason, basically, is that an agreement about a common position on the scale will always, because of the necessarily considerable size of the intended coalition, be worse for parts of the coalition than some prior, or otherwise obtainable, position (a complete proof of the theorem is given in Moberg, 2000). The agreement will therefore not come about. In this context this means that the bargaining, which Powell talks about, cannot possibly be only about positions on a left-right scale. The bargaining has, at least to some extent, to be about other matters as well. If these other matters are special interests of various groups in society it becomes, on the contrary, very easy to account for the emergence of a coalition. The members of the coalition just support each others special interests at the expense of the outsiders. If, so, this is just an ordinary case of exploitation by means of the majority rule.

Since special interests of various groups are easily handled when governing coalitions are formed, it is also likely that they will affect the political campaigns in proportional multiparty systems (this topic, as well as coalition formation, is further discussed in Moberg, 1999). A party striving to get the votes of some group of marginal voters may thus promise to support some special interest entertained by that particular group. It is my hypothesis that campaign promises of this kind are more frequent in parliamentary, proportional systems than in majoritarian systems. Powell does not work with this kind of data, but if he did so I guess that his picture of proportionalism would be a different and less harmonious one.
 
 

Summary

My main objection to Powell's clear and thought-provoking book is about his treatment of proportional systems. I think that his conclusions about the good properties of these systems with respect to responsiveness and representational congruence are unsubstantiated. The conclusions, as I see it, depend more on his unduly inflated coefficients than on real empirical data. Furthermore, I think that he misses some important, and less attractive, properties of proportional systems by sticking too closely, and too single-mindedly, to the spatial model and its related concepts. By that choice of model an examination of the role of special interests in the political process, which could be very important in proportional systems in particular, is excluded from the beginning.
 
 

Bibliography

Moberg, E, 1999, "The Expanding Public Sector - A Threat to Democracy?", in The Limits of Government - On Policy Competence and Economic Growth, (ed G Eliasson & N Karlson). New Jersey: Transaction Publishers. - A link to a pdf-file of this paper will be found in the section "Erik Moberg: Printed publications" on this site (www.mobergpublications.se).

Moberg, E, 2000, "Ideological Politics and Interest Politics".This paper is published in the section "Arguments" on this site (www.mobergpublications.se).