result, we can distinguish the impact of the mayors’ election factor from the impact of the country variable.
But we found that the method of mayors’ nomination remained insignificant when controlled for other independent variables. Other results are not very much different than what has already been presented in the party significance index. Proportionality in elections is confirmed as an important factor. This result may look surprising since it is not directly related to mayors’ party membership. But in countries where the mayor is elected by the council, the method of council election is definitely very important—a more partisan council elected in a proportional system is more likely to nominate a par-tisan mayor. The council election system may also have an indirect impact in countries in which the mayor is directly elected by all voters. Even in this case, a mayor needs to cooperate with the council. We should remember that the position of directly elected mayors in CEE countries is usually relatively weak. For example, they have to rely on council approval in appointing important staff members. However, party mayors have a better chance of success in cooperation with highly partisan councils elected in a pro-portional system. In countries with majoritarian electoral systems, councils are more likely non-partisan and they may cooperate more easily with an independent mayor.
There are also country-specific variables that remain important. Estonia, Romania, Slovakia, and to a lesser extent Bulgaria are more likely to have partisan mayors. In Latvia and Hungary, independent mayors are more likely to be elected, assuming that the impact of other (institutional) factors is eliminated. There is an interesting differ-ence between the impact of the “Poland” variable in both regression models. “Poland”
had a modestly negative impact on the 2003 party significance index, while the impact of the same variable on the 2000 mayors’ partisanship analysis was modestly negative.
The change of sign might be related to the 2002 reform of Polish mayoralty election methods (although, as mentioned earlier, this variable’s high significance has not been confirmed).
Do our results shed light on whether there are parties that govern in Central and East European local politics? The proper answer is—it depends where and when. We have been able to identify countries in which there is a tendency for strong party influence (Slovakia, Bulgaria, and Romania). But institutional factors also play a role—especially the territorial fragmentation of the local government system and the method of council election.
5. POLITICAL FRAGMENTATION WITHIN LOCAL GOVERNMENTS
dominant position that can easily implement its policies, or are there several small groups that necessitate stable or ad hoc coalitions for each decision that is made? To measure levels of political fragmentation we use an index called the Rae-Taylor fragmentation index. The formula for this index is:
FRA = 1 – ∑s(i)2,
where FRA is a fragmentation index for the council, and s(i) is a share of mandates possessed by the faction.
One can find two versions of this index (see Gabriel et al. 2002): the “electoral fragmentation index,” which reflects fragmentation of votes in the election, and the
“parliamentary/council fragmentation index,” which takes into account the division of seats among party groups. The former better reflects the variation in citizens’ political options. But we will concentrate on the latter, as it illustrates the environment for day- to-day local government management.
The index differs from 0 to 1, where 0 means a situation in which all seats in the council belong to the same political group (or all voters voted for one party, if we take the
“electoral” version of an index), while 1 means that each councilor is independent and there are no factions in the council (or every councilor belongs to different political party).
Table 1.16 illustrates the fragmentation of local councils in seven CEE countries.
The analyzed countries may be divided into those with a high or low fragmentation index. The overall fragmentation was by far the highest in Hungary with an average score above 0.9, then in Poland (0.818), and in Latvia (0.809). In the group with a low fragmentation index we find Romania (0.727), Slovakia (0.731), Bulgaria (0.744), and Estonia (0.747). But if we take into account differences in the size of local governments, the lowest score is for Slovakia.
We expected that fragmentation would be lower in the big cities, where parties play a more important role and there are fewer independent councilors who do not belong to any faction. The last row in table 1.16 shows that this is definitely the case in Hun-gary, to a slightly lesser extent in Poland, and much less in Slovakia. In the remaining countries the correlation is insignificant (Bulgaria, Romania, and Latvia) or even points weakly in the opposite direction (Estonia).
The data on a fragmentation index is static and cannot show changes over time.
But as the party system in Central and Eastern Europe is far from stable, we can expect that fragmentation within the council changes significantly during the term of the council. Some groups and coalitions break down after the election and new factions are formed, which increases the level of fragmentation. The opposite can also happen, but our observations suggest this is much less frequent. We cannot verify this in a systematic way, but the hypothesis can be partially confirmed by data from Poland. The index of fragmentation calculated in 1998 for all Polish cities above 100,000 citizens was 0.614
(Swianiewicz and Klimska 2003). The same index, calculated for the same group of cities two years later (still in the same term) was 0.680, i.e., it was considerably higher.
Polish data also suggest that the level of fragmentation has increased in the current term compared to the situation after the previous elections. At the end of 2002, the mean score for cities over 100,000 was 0.670, while after 1998 it was only 0.614 (Swianiewicz and Klimska 2003). Other observations tend to confirm this. For example, in the 1998 elections 54 committees or factions won seats in 42 big cities, while in 2002 the number increased to 98. The other indicator of increasing fragmentation in Polish cities is the diminishing number of councils in which one party received an absolute majority of votes. Of 42 cities over 100,000 citizens, 22 had a clear majority in the council after the 1998 elections; after the 2002 elections this number decreased to 11 (Swianiewicz and Klimska 2003). Interestingly enough, the increase in fragmentation was usually the highest in cities with a high electoral turnout. A well-known observation of Lipset (1981)—that a high level of voter activity is often correlated with a demand for political change—might offer an explanation. Unfortunately, we cannot compare this data with changes in other CEE countries and in smaller local governments.
How does the level of fragmentation in CEE countries compare with the situation in EU countries? We can analyze this only in an indirect way. A study published in 2002 includes fragmentation indices for big cities in eight EU countries (Vetter 2002).
But those indices were calculated for electoral data, not on the basis of information on the distribution of seats in the council. An electoral fragmentation index will always be higher than parliamentary fragmentation, so direct comparison with our scores in table 1.15 is impossible. But in 2002 an electoral fragmentation index was calculated for the 20 largest Polish cities (Swianiewicz and Klimska 2003). The score ranged from 0.669 in Gdynia and 0.760 in Sosnowiec to 0.873 in Krakow and 0.878 in Szczecin. With the exception of Gdynia, the scores were much higher than those for cities of a similar size in the EU. Of all the cities analyzed by Vetter, only Helsinki (0.82) and some Dutch
Table 1.16
Council Fragmentation Index (2000–2001)
Bulgaria Estonia Hungary Latvia Poland Romania Slovakia
Mean 0.744 0.747 0.967 0.809 0.818 0.727 0.731
Size-standardized mean 0.730 0.770 0.918 0.798 0.830 0.731 0.693 Correlation between
fragmentation and size of local government
* (–)**** (–)*** (–)*
Note: * means correlation significant at 0.05 level, ** at 0.01 level, *** significant at 0.001 level, and
**** significant at 0.0001 level. Blank spaces mean insignificant correlations. (–) means negative sign of the correlation coefficient.
Source: Own calculations based on ILDGP survey of CEOs (2000–2001).
cities (e.g., Amsterdam) showed values similar to those in Poland. But the fragmenta-tion of votes in cities in Denmark, Sweden, Norway, the UK, France, and Germany was much lower. For example, in the six biggest English cities the score ranges from 0.60 in Manchester to 0.65 in Liverpool. It is also very low in several French cities—0.569 in Tours, 0.572 in Nantes, and 0.584 in Bordeaux (Hoffman-Martinot 2002).
From our data we know that the parliamentary fragmentation index in Polish cities over 100,000 citizens is usually lower than in other big cities of Central and Eastern Europe (except in Slovakia). Indirectly, therefore, we may conclude that the level of po-litical fragmentation in CEE cities is much higher than in most of the EU countries.
It appears that weakness in the party system in CEE countries leads to consider-able fragmentation of local councils and can potentially produce complications in local management. Another potential complication is that the mayor does not always have a majority in the council—the opposite of the situation resulting from recent reforms in southern Europe. This may lead, and actually often does lead, to serious management problems when a mayor faces a majority opposition on council. Recalling the connection between low stability and high fragmentation of political parties in CEE countries, we can expect numerous cases in which effective decision-making may be difficult. Only seven mayors in the 42 Polish cities with a population over 100,000 have the clear support of the majority of councilors. In most of the remaining cases the mayor needs to rely on more or less stable coalitions. In five cities the situation is even more complicated—the majority in the council is held by a group that is definitely in opposition to the mayor (Swianiewicz and Klimska 2003). Similar problems are reported in other countries.
Sometimes they concern major cities such as the Romanian capital, Bucharest, or the Albanian capital, Tirana. But this issue goes beyond the scope of our analysis here.
6. COUNCILORS’ PARTY AFFILIATIONS AND POLITICAL CULTURE