Partisanship and Political Attention in France: Agenda Dynamics and Electoral Incentives

Home , French Fifth Republic

APSA, Boston 2008 Frank Baumgartner Pennsylvania State University Sylvain Brouard SPIRIT, Sciences Po Bordeaux Emiliano Grossman Sciences Po Paris

Partisanship and political attention in France: Agenda dynamics and electoral incentives

How does issue attention change over time and what determines such change? Are there leftwing or right-wing issues? Do partisan preferences affect issue attention or do the media or public opinion determine change in political attention? Or are institutional dynamics of political life more important? Do the proximity of elections, majority status or the years in office explain issue attention? This paper examines the dynamics of issue attention in France. In doing so, it will draw on the increasing amount of monographic or comparative work on the politics of attention and agenda-setting. It is based, moreover, on original data collected over the past three years on three separate and comprehensive agendas: presidential speeches (each quasi-sentence of the annual New Years Address by the President), bills (all government bills introduced in Parliament), and government statements and bill proposals (e.g., the agenda of the formal weekly cabinet meetings held each Wednesday throughout the Fifth Republic). For all of these indicators, we possess data for the period from 1986 to 2006. Each item in these three databases has been coded according to the common topic codebook of the comparative policy agendas project so we can easily assess the degree of attention across all policy topics, and we can do so in a manner that will later enable international comparisons as well. Our objectives are several. First and foremost, we want to renew the study of French institutions by developing original and innovative indicators of the life of political

1 institutions, as well as partisan and media indicators. While this paper is only a first step towards this wider objective, it already builds on a significant amount of data. Second, from a more comparative point of view, we intend to build on the literature in comparative politics and comparative public policy, as well as on the currently ongoing agenda projects elsewhere, to explore the dynamics of attention and to understand the specific determinants – institutional or other – of issue attention change. In this paper, we will concentrate on the importance of the partisan character of particular issues. To which extent are issues partisan: i.e. left-right or right-wing? To which extent, on the contrary, are they determined by other factors? In the rest of this paper, we will first discuss in further detail our research outline and hypotheses. We then look at static differences, before exploring more dynamic hypotheses on the importance of partisan factors.

I. Issue attention and partisanship In democratic theory, whether traditional or economic, people vote in accordance with policy preferences, and political parties propose and try to implement specific policies for instrumental or intrinsic reasons. So, the identity of the governing parties should matter in policy making. According to the partisan hypothesis,1 the partisan composition of governments affects policies. Several empirical studies support this perspective.2 In contrast, however, other studies demonstrate that political parties do not matter in government and more precisely in lawmaking.3 For example, Cutright (1965), Haniff (1976), Wilenski, (1975), and Dye (1976) point out that « Welfare state » expenses depend on economic growth and demographic variations rather than on partisan preferences. More recently, renewing the industrial society convergence theory (Pryor, 1968 ; Parkin, 1971), scholars explain that globalisation (Garett & Lange, 1991 ; Keohane & Milner 1996 ; Mair, 1995), the increasing levels of complex interdependence (Keohane & Nye, 1989) and international capital mobility (Simmons, 1998 ; Andrews, 1994 ; Moses, 1994) decrease

1 For example, CASTLES F. & McKINLAY R., « Does politics matter ? an analysis of the public welfare commitment in advanced democratic states », European Journal of Political Research, Vol. 7, 1979b, pp. 169-186, BLAIS A., BLAKE D. et DION S., « Do Parties Make a Difference ? Parties and the Size of Government in Liberal Democracies », American Journal of Political Science, Vol. 37, n° 1, February 1993, pp. 40-62. 2See for example ROSE R. & DAVIES P. L., Inheritance in Public Policy : Change without Choice in Britain, New Haven, Yale University Press, 1994, and WILENSKI H., The Welfare State and Equality, Berkeley, University of California Press, 1975. 3 SCHMIDT M. G., « When Parties Matter : A Review of the Possibilities and Limits of Partisan Influence on Public Policy », European Journal of Political Research, Vol. 30, 1996, p. 155-183.

2 sharply the probability of partisan influence. The historical work of R. Rose and P. L. Davies (1994) on Britain policies also concludes that parties do not matter in policy making. Thus, we face two essentially divergent views on the relationship between parties and policies. One way to overpass this contradiction is to focus on « when parties matter »4 and to set a research agenda for analyzing the interaction between partisan variables and institutional or socio-economic variables. In this paper, we will focus on the effect of partisanship and electoral competition on three different agendas to understand if and how patterns of issue attention are shaped by partisanship.

Parties and issue attention The “partisan difference hypothesis” remains common. It is based on equally forceful arguments. First, given that parties have different policy priorities, partisanship will induce variation in attention allocation. This hypothesis has the great advantage that it is easily testable. Moreover, we assume that certain issues are historically linked to particular political camps. A specific set of issues should be associated with leftwing and rightwing incumbents. These issues are of course not the same in all countries and over time. Without going into too much detail as to how we come to this hypothesis, we expect the following for the French case: - Rightwing parties should dedicate more attention to issues like law and order, immigration, defense, agriculture; - Leftwing parties should focus more on Rights, solidarity, housing, environment, education, labor.

Let us now turn to the “partisan similarity hypothesis”. In the above mentioned literature, there are several types of arguments to explain the perceived or real convergence of partisan preferences. A first argument is very straightforward: given that portfolio holders want to attach their name to a policy initiative, each government, president, or legislature will try to legislate on each topic. Put differently, parties are made of individuals and party leaders, once

4 SCHMIDT M. G., « Political performance and types of democracy : Findings from comparative studies », European Journal of Political Research, Vol. 41, 2002, p. 147-163; SCHMIDT M. G., « The impact of political parties, constitutional structures and veto players on public policy », in KEMAN H. (ed), Comparative democratic politics, London, Sage, 2002 ; BOIX C., « Privatizing the Public Business Sector in the Eighties : Economic Performance, Partisan Responses and Divided Governements », British Journal of Political Science, Vol. 27, 1997, pp. 473-496, and OATLEY T., « How Constraining Is Capital Mobility ? The Partisan Hypothesis in an Open Economy », American Journal of Political Science, Vol. 43, n° 4, 1999, pp. 1003-1027.

3 in power, will attach more importance to their own success than to the values and the ideology of their party. Second, even assuming, on the contrary, that parties are indeed characterized by strong (ideological) preferences, this does not necessarily affect issue attention: newly elected governments will want to erase changes from outgoing governments. This will thus induce a similar attention allocation after alternation. Provided that alternation has been the rule for every general legislative election in France between 1981 and 2002, i.e. all the elections in our sample, this hypothesis is paramount for our paper. Finally, a third line of argument posits “partisan neutrality”. According to this point of view, governments are driven by the general dynamics of public problems that arise from the media agenda, public opinion, international events, the economy… Political attention is thus first of all shaped external events. Those alternative hypotheses are of course fundamental. Given our data, we will not be able to test all hypotheses equally. In particular, for the time being, we lack systematic data on external variables. Hence, the third hypothesis will only be tested residually, i.e. only in as far as the two preceding hypotheses are not confirmed. And even this will not be straightforward, as it will be close to impossible to distinguish it clearly from the “similarity” hypothesis. Moreover, it will not be possible – within the limits of this paper – to explore in more details the eventual sources of partisan similarity, as this would require a separate and specific research design. In sum, then, we will only look at whether differences between parties are strong and significant.

Despite their importance, these hypotheses can hardly be said to cover all of the potential significance of parties. It is true that all of these hypotheses may be confirmed at different points in time. We therefore need some additional elements about how these different hypotheses may apply at different moments of political life. Put differently, after an essentially static comparison, we will then turn to a more dynamic analysis.

Political cycles and relevance of partisanship in attention allocation We will focus here on the importance of elections. We assume that the attention to policy issue before elections will depend on the prioritization of issues in (the targeted part of the) pubic opinion. But electoral campaigns have also the function to focus attention on specific issues in order to alter the prioritization or the salience that the public ascribes to them. In fact, out hunch is that by shaping prioritization of issues parties will be better able to attract voters and to win elections.

4 If a given election deals first of all with national security, voters will be more confident in conservative parties. Thus conservative parties do have incentives to focus attention on national security. Generally speaking, conservative parties have an objective interest in prioritizing issues that may induce a conservative vote, provided that the conservative camp has a better reputation in dealing with such issues. The list of issues or political objectives that are unambiguously associated with a particular camp are of course not very many and not entirely stable over time. Yet, historically, national security, fighting inflation, reducing taxes or toughness on crime are issues where the electorate has trusted conservative or right-wing parties more than left-wing parties. And as far as economic issues are concerned, a large comparison of EU member states has recently shown that these are associations continue to work very well for large parts of the electorate (van der Brug, van der Eijk & Franklin 2007). The same type of logic may of course apply to leftwing parties. But parties do not only propose different solutions to the same set of issues. Parties prioritize differently the various issues and dedicate different amount of attention to these issues. Before elections, parties should focus attention on their privileged policy domain in order to frame the electoral campaign and signal their preferences to the (targeted part of the) electorate. Being in office gives the opportunity to use the government meeting agenda, the new year address and the legislative agenda in order to weight issues favorably and to signal the credible commitment of the current majority to fix these issues. In sum, the expectation is that pattern of attention before election should be different according to partisanship. Or, partisan differences should be reinforced and even exacerbated in the run-up to general elections. According to the same logic, pattern of attention just after elections should be different, too. Each party will address the key issues as soon as it comes to office in order to signal that it complies with the electoral promises and its issue priority. Further into the legislature, the impact of partisanship on attention allocation should be weaker before increasing again when the next election becomes closer. New problems arise, as the reality of the management of the state and the existing public policy constrain the government to allocate attention on a wide range of issues, as the relationship to the electoral promises become looser. Three phases of the attention cycle may be distinguished that should give rise to the following sequence concerning issue attention: the pre-electoral agenda setting period, the post-electoral promise fulfilling period, the governing period, and then, again, the pre-electoral agenda

5 setting period, the post-electoral promise fulfilling period, the governing period… The duration of the legislative term has of course an impact on this cycle. The fact that given its electoral system and party system, a political system is centripetal or centrifugal (Cox 1990) may decrease or increase the effect of the partisanship. When competition focuses on the median voter, the attention cycle should be less important than when gaining the median voter is not sufficient to win the election and the winning party (or coalition) should be able to attract far-rightwing or far-leftwing voters.

Given this theoretical understanding of the relationship between electoral competition, party and issue attention, several hypothesis should be made and tested. - the effect of partisanship is not fixed but varies according to the incentives derived from the electoral cycle and give rise to a three phases attention cycle. - leftwing and rightwing incumbents are expected to allocate attention (more) differently in the two phases of the attention cycle prone to partisan impact.

Data and classification As mentioned in the introduction, we draw on three different series: - bills - presidential new years’ address - weekly government statements. The three series are very different, of course, in that they vary in frequency, type of content and style. The number of bills introduced each year in the French Parliament is much lower than, say, in the US. It is only about 100 on average in the post-war period, which comes close to the average in most EU member states. Another peculiarity is the very high share of “international affairs” bills, which make up about forty percent of the bills on average. This is mainly due to the fact that all international conventions and agreements – including bilateral agreements – have to be adopted by a bill. The presidential new year’s address can hardly be compared to US-equivalent either . Unlike the State-of-the-Union speech, it is rather short and really more of new year’s address than a political programme. It is true, though, that major political issues are always mentioned and sometimes discussed during this particular speech. Finally, the weakly government statement have to be put into the context of parliamentary government, where the prime minister is usually the most important political actor. Yet those weakly statements are not very common elsewhere in Europe, even if similar statements exist in Belgium. Those statements basically list all items that have been dealt with during the

6 Wednesday government plenary meeting at the presidential palace. During this meeting, several types of items are discussed. The most important ones are government bill proposal: provided that on average 95 per cent of adopted bills are government bills, this element is crucial. The second most important item – in quantity – are government statements on any area of policy making. Finally, two other items are relevant: decrees and ordinances are law- like decisions without direct parliamentary approval5. All items of the three series have been assigned a topic code. The French code book is a translation and adaptation of the original US topic code, developed by Baumgartner and Jones6. The French code book, like its equivalents in the US and elsewhere, contains about 21 general topics which are on average sub-divided into ten more categories. This leads to a total of more or less 220 topic codes. For the purpose of the agenda comparisons in this paper, we will limit the analysis to the 21 more general topic codes. Eventually, the different “agendas” will have to be compared systematically and confronted with newspaper data, partisan programs and other elements to study the relative permeability to outside influence of each of those agendas. This will help us to understand the degree of “friction” that characterizes a particular agenda (cf. Jones, Larson-Price & Wilkerson 2008). For the purpose of this paper, we will concentrate on the “partisan hypothesis”, its strengths and its weaknesses, only.

II. Left-wing and right-wing issues? This section will quickly discuss the importance of left-right differences for the aggregate data, i.e. for all entries between 1986 and 2006. Tables 1 through 3 present our three different series splitting left and right presidential or legislative majorities. Without going into the specifics of the French political system, we simply recall that the most important election is that of the president, but that the president is hardly powerful without a favourable majority in the Assemblée nationale, France’s lower chamber. During cohabitation, i.e. opposing legislative and presidential majorities, it is the former that prevails7. Table 1 provides the proportion of attention as mirrored by adopted bills. We’re thus looking here at the outcome of partisan preferences. Provided that it takes between several months and

5 For a longer detailed of law-making and its evolution in France, see Grossman (2008). 6 For the original codebook, see http://www.policyagendas.org. The French codebook is available from the author by request. 7 For a more detailed discussion of France’s political system, see Elgie (2005) or Appleton, Brouard & Mazur (2008).

7 several years to adopt a bill, it is of course difficult to interpret differences between camps. Yet, it is true that given the close to complete control of the parliamentary agenda by the French government8, the latter is free to adopt particular bills when it best suits it, rather than following the particular cycle of legislative procedure. Therefore, the existing differences are far from meaningless. Table 1. Partisan differences in the areas of lawmaking activity in France, 1986 to 2006. President of the: Government of the: Category Left Right Sig. Left Right Sig. 1 Economics 4.54% 4.83% 0.823 4.38% 4.97% 0.654 2 Civil Rights 1.36% 3.62% 0.018 2.92% 1.91% 0.294 3 Health 5.02% 1.70% 0.030 4.63% 2.31% 0.155 4 Agriculture 2.60% 1.02% 0.142 2.43% 1.30% 0.316 5 Labor 4.77% 4.57% 0.898 5.38% 3.97% 0.395 6 Education 0.98% 0.79% 0.668 0.53% 1.26% 0.084 7 Environment 3.12% 1.24% 0.028 3.37% 1.11% 0.010 8 Energy 0.08% 0.18% 0.634 0.09% 0.16% 0.684 9 Immigration 1.48% 1.11% 0.660 1.21% 1.40% 0.823 10 Transport 2.18% 2.47% 0.673 1.88% 2.75% 0.219 12 Crime 7.59% 4.15% 0.006 5.85% 6.12% 0.838 13 Social Welfare 1.06% 1.67% 0.250 1.29% 1.40% 0.831 14 Housing 1.82% 1.22% 0.404 1.35% 1.73% 0.616 15 Commerce 8.18% 3.94% 0.017 6.17% 6.24% 0.972 16 Defense 3.03% 2.19% 0.456 3.16% 2.12% 0.377 17 Science 1.27% 0.67% 0.344 1.40% 0.59% 0.225 18 Trade 3.04% 5.61% 0.062 4.21% 4.26% 0.972 19 International 25.14% 45.70% 0.001 30.20% 39.26% 0.165 20 Government 14.12% 7.71% 0.022 10.93% 11.33% 0.896 21 Lands 6.46% 3.53% 0.068 6.70% 3.48% 0.049 23 Culture 2.16% 2.08% 0.913 1.92% 2.33% 0.589 Total number of laws 873 984 873 984 Number significant at .01 2 0 Number significant at .05 7 2 Number not significant 14 19

Rather surprisingly, legislative majorities appear to have less influence on issue attention than the president. This may be explained by the greater importance of the president in the executive, during “normal” times, i.e. in the absence of divided majorities. Yet, it is also true that the president does have some influence on issue attention. In fact, he always heads the Council of ministers and, thus, all major issues are discussed in his presence, even under divided majorities. Moreover, it has generally accepted under the 5th Republic that the

8 On this point, cf. the vast comparative study directed by Strøm, Müller & Bergmann (2006).

8 President is to play an important role in the “reserved domains”, i.e. foreign affairs and defense, whatever his political color. Yet the results presented in table 1 are not straightforward: while differences are significant on crime, the relation is not as expected. Mitterrand (the only left-wing president during the period under consideration) appears to have cared more about crime than Chirac (the rightwing president for our period). Only the environment and health appear as left-wing issues, as expected. Moreover, Chirac appears to have had a much stronger interest for international affairs than Mitterrand. And this is the single most important difference in this table. Another interesting difference concerns the attention paid to government regulation. It appears that the left has legislated more in this area than the right. This is at least partially due to the fact Mitterrand was in office between 1981 and 1995, while Chirac was in charge thereafter. The “reform of the state”, still a very prominent term in France, led to much more legislation in the context of the privatizations that began in 1986 than in the second half of ours sample. Finally, the main conclusion of table 1 is the relative insignificance of partisan preferences for the adoption of bills. The fact that the legislative majority has hardly any impact clearly points to the similarity or neutrality theses. Moreover, the presidential differences should not be over-interpreted, provided that we are dealing with a “n” of just two cases here. Much can be due to the evolving dynamics of political issues more generally.

In table 2, presidential new year’s addresses show more and stronger differences. These are all the more interesting as they contradict, at least partially, the conclusions of table 1.

9 Table 2. Partisan differences in the areas of Presidential speechmaking activity in France, 1986 to 2006.

President of the: Government of the: Category Left Right Sig. Left Right Sig. 1 Economics 10.49% 17.42% 0.037 12.04% 15.41% 0.334 2 Civil Rights 3.13% 7.22% 0.056 5.19% 4.88% 0.885 3 Health 0.81% 2.74% 0.060 1.55% 1.86% 0.758 4 Agriculture 0.00% 0.34% 0.112 0.24% 0.07% 0.397 5 Labor 4.04% 9.07% 0.073 4.06% 8.71% 0.074 6 Education 0.86% 4.34% 0.000 1.84% 3.12% 0.187 7 Environment 0.51% 4.36% 0.029 2.93% 1.69% 0.458 8 Energy 0.00% 0.15% 0.336 0.00% 0.14% 0.334 9 Immigration 0.22% 0.45% 0.440 0.36% 0.29% 0.837 10 Transport 0.23% 0.83% 0.141 0.29% 0.73% 0.280 12 Crime 0.83% 5.49% 0.005 3.50% 2.50% 0.537 13 Social Welfare 7.65% 6.84% 0.639 7.22% 7.32% 0.953 14 Housing 0.29% 1.29% 0.036 0.37% 1.14% 0.092 15 Commerce 0.00% 0.58% 0.043 0.00% 0.54% 0.044 16 Defense 7.86% 1.51% 0.000 5.17% 4.62% 0.768 17 Science 0.23% 2.86% 0.005 1.25% 1.67% 0.644 18 Trade 1.63% 2.77% 0.310 1.83% 2.50% 0.563 19 International 47.07% 18.07% 0.000 39.40% 27.68% 0.112 20 Government 13.35% 13.08% 0.939 11.93% 14.52% 0.474 21 Lands 0.22% 0.00% 0.333 0.23% 0.00% 0.334 23 Culture 0.58% 0.61% 0.946 0.59% 0.59% 0.996 Total number of speech activities 599 941 713 827 Number significant at .01 5 0 Number significant at .05 9 1 Number not significant 12 20

Generally speaking, it appears that the president speaks more about what he does not do than about what he does. While we have not tested for this here, there even appears to be negative relationship between legislation and the new years address on some key issues. This is most visible for international affairs, where the relation between left and right is simply inverted. This also appears to apply to crime, where we have seen that more legislation had been adopted under Mitterrand. One president legislated actively in the area but spoke rarely of it, while his successor, Jacques Chirac talked often about it but legislated little. Finally, the differences on education – that we defined as a left-wing topic – are also interesting. Chirac mentioned it several times more often than Mitterrand in his NYAs. This also applies to the environment. Again, it is interesting to note the relative absence of other major issues in this list: labor or crime and immigration, housing or social welfare.

Table 3. Partisan differences in the areas of Council of Ministers activity in France, 1987 to 2006. President of the: Government of the: Category Left Right Sig. Left Right Sig. 1 Economics 5.21% 3.37% 0.027 4.36% 4.28% 0.926 2 Civil Rights 2.09% 2.93% 0.141 2.30% 2.67% 0.527 3 Health 2.46% 3.98% 0.010 3.28% 3.11% 0.781 4 Agriculture 1.32% 2.38% 0.111 1.41% 2.23% 0.196 5 Labor 8.15% 5.23% 0.007 6.94% 6.55% 0.739 6 Education 4.63% 5.43% 0.443 6.32% 3.81% 0.014 7 Environment 4.19% 4.95% 0.199 4.51% 4.60% 0.885 8 Energy 0.66% 0.96% 0.272 0.67% 0.93% 0.349 9 Immigration 0.92% 1.32% 0.220 0.91% 1.30% 0.242 10 Transport 3.17% 3.96% 0.230 3.33% 3.76% 0.505 12 Crime 5.95% 8.49% 0.059 6.91% 7.42% 0.712 13 Social Welfare 3.79% 3.73% 0.931 3.80% 3.73% 0.932 14 Housing 2.53% 2.82% 0.601 2.51% 2.82% 0.576 15 Commerce 7.00% 6.75% 0.837 6.29% 7.44% 0.329 16 Defense 2.47% 3.95% 0.010 3.43% 2.95% 0.417 17 Science 2.78% 2.38% 0.558 3.02% 2.18% 0.224 18 Trade 2.28% 1.40% 0.045 1.89% 1.82% 0.883 19 International 15.08% 16.06% 0.608 15.65% 15.46% 0.919 20 Government 19.15% 13.72% 0.002 15.68% 17.32% 0.383 21 Lands 2.17% 1.96% 0.698 2.87% 1.32% 0.003 23 Culture 4.01% 4.23% 0.695 3.92% 4.30% 0.502 Total number of actions 2118 2332 2003 2447

Number significant at .01 4 1 Number significant at .05 6 2 Number not significant 15 19

Finally, the data from the Council of ministers does not fundamentally alter the general image. It is true that labor, for the first time, appears as expected as a left-wing issue, but health appears to be more important to the right. Defense appears more important to the right, as expected, but the differences are feeble. All in all this static comparison does not yield very consistent results. While there are some differences, these rarely concern the expected areas and the differences, more over rarely point into the expected direction. Table 4 summarizes our results on left-wing and right-wing topics. The table shows clearly that neither the left nor the right is on average more active on any area that is associated with their camp. There are of course a few exceptions, but all in all the hypothesis according to which partisan preferences structure issue attention is disconfirmed.

11 Table 4 – Summary of results for by presidential and legislative majority (legislative in parentheses) Issues Bills Government agenda New year’s address

Leftwing Rights - (0) 0 (0) - (0) Labour 0 (0) + (0) - (0) Education 0 (0) 0 (+) - (0) Environment + (+) 0 (0) - (0) Housing 0 (0) 0 (0) - (0)

Rightwing Agriculture 0 (0) 0 (0) 0 (0) Immigration 0 (0) 0 (0) 0 (0) Law and order - (0) 0 (0) + (0) Defence 0 (0) + (0) - (0)

Note: + : in the expected direction; 0 : no difference between left and right; - : relation opposite to that expected.

Yet, we also see that it might be interested to explore the differences between different agendas as there may be trade-offs between legislation and speeches of government statements. The correlations between the different agendas (not reproduced) here are hardly significant. More interestingly, as the last column of table 4 shows, presidential speeches appear to work systematically against our expectations. Mitterrand preferred to talk about supposed right-wing issues, while Chirac talked a lot about left-wing issues. While much of this may, of course, be explained by the personalities of the two presidents. It is certain that the presidential function leads to a general interest in all topics. They will go even further to show that they are able to handle issues that are usually associated with the opposite camp. This leads us to a more general question of the relation between the different agendas. Governing is made up of many things: legislation, communication, agenda-setting etc. All these activities or functions do not appear to follow the same logics. This is probably the main result of this first analytic section. The agendas seem to move rather independently. Even in a presidentialized political system with rationalized parliamentarism, governing, legislating and speaking appear only loosely connected.

III. Parties, elections and the dynamics of issue attention

In this section, we will examine the explanatory power of our interactive hypothesis: do partisan preferences play a stronger role in the wake and immediate aftermath of elections?

12 In order to study this, we will first look at the overall evolution of issue attentin over the 20 year under study, irrespective of which camp is in office.

General dynamics Figures 1 and 2 show the evolution of issue attention for laws and government stattements. Figure 1 presents the policy topics (numbered from 1 to 23 at the bottom of hte figure; see Tables 1 through 3 for an explanation of these policy topic labels) for the different parliamentary leggislatures from the 7th (1986-) to the 12th (ending in 2006). It shows that over the past 5 legislatures the distribution of issue attention has become more even. Peaks on labour (5) and government (20) have significantly declined. This appears to be due to more general changes in the preferences of politicians and/or the elecorate. Labour rights and social security were very important at the beginning of the period under study, as the outgoing socialist government in 1986 tried to adopt several measures to signal its commitment to leftwing policies. Several of these policies were later reformed by the new right-wing government won the 1986 elections. A similar logic applies to government (20). The issue of “state reform” has been a central issue since the 1980s, but in the late eighties the debate dealt with very general and highly publicized question. In the 1990s and 2000s the issue did not disappear, of course, but became more technical and regulatory, leaving the general political agendas. A third decline concerns legislation on health. This issue has followed a logical comparable to that of state reform, but changes were less radical. A fourth decline took place on macroeconomic policy (1). This decline is again more subtle than for government or labour, but nonetheless significant. While it would take too much time to break down this decline into the different subcategories, this evolution is clearly related to the change of the role of the state in the economy. With the decline and – ultimately – the disappearance of monetary policy or the decline of industrial policy, France became a more mainstream country in terms of economic policy making (cf. Loriaux 1991, Schmidt 1996).

13 Figure 1. Laws by policy topic, 1986 to 2006.

Beyond those declines there is at least one significant increase (see Figure 2), concerning the attention devoted to crime (12). While we defined it as a right-wing issue in section 1, Figure 2 shows clearly that increase in interest for law and order has taken place irrespective of parliamentary or presidential majorities. The exact story of this sudden rise in the interest for law and order still needs to be written. Preliminary analyses disconfirm in any case that this happened after increased public opinion interest for those issues.

14 Figure 2. Government agendas by polity topic, 1986-2006.

Let us now turn to political cycles and the possible effect of elections and electoral timing on these dynamics.

Divided government, elections and issue attention

Divided government has been a common political situation (9 years) in France since 1986. A lot has been said and written about divided government in France although no systematic studies have been done so far. Our data shows results in contradiction with common wisdom. In fact, one of the main argument is that French President dedicate more attention to International Affairs during divided government than during unified government. Unfortunately, during Chirac presidency, the difference in attention to foreign affairs is insignificant whereas during Mitterrand presidency, the President talked more about the International Affairs during unified government. Historical conditions (democratization of the former East-European countries for example) of course largely affect attention to foreign affairs. Nevertheless, there is no obvious effect of divided government on International affairs. Figures 3 through 9 compare levels of attention in the three agendas (Bills, Presidential Speeches, and Council of Ministers activities) by divided government (Figure 3) and then for

15 those issues traditionally associated with the left (topics 5 labor, 6 education, 7 environment, 13 social welfare, and 14 housing) and the right (topics 4 agriculture, 9 immigration, 12 crime, and 16 defense). Each shows the percentage of total attention for the topic in question.

Figure 3. Presidential attention by divided government status, 1986-2006.

Figure 4. Percentage of Bills Introduced on « Leftwing » issues, 1986-2006.

16 Figure 5. Percentage of Bills Introduced on « Rightwing » issues, 1986-2006.

Figure 6. Presidential Attention to “Rightwing” issues, 1986-2006.

17 Figure 7. Presidential Attention to « Leftwing » issues.

Figure 8. Council of Ministers Attention to “Rightwing” issues, 1986-2006.

18 Figure 9. Council of Ministers Attention to “Leftwing” issues, 1986-2006.

Our hypotheses of the effects of elections is only partially confirmed. As we have seen earlier, some issues, such as law and order are no longer the monopoly of a particular camp. On other issues, however, electoral periods appear to change attention significantly. In general, while it is easy to find isolated examples of clear partisan effects, more general patterns are much harder to discern when we look at the entire set of issues. We turn last to Table 5 in which we look at the timing of attention by governments of the left and the right. The table shows the average attention overall to each of the policy topics in the dataset, then breaks it down into pre- and post-election attention. Finally, it shows the same data separately for governments of the left and right. If governments are manipulating their activities for electoral gain, there may be higher levels of attention to certain issues in the pre- election phase and then less attention afterwards. Further, governments of the left and right may choose to highlight different issues.

19 Table 5. The electoral timing of policy attention. post- Topics average pre-election election pre-left post-left pre-right post right 1 6 5 7 4 7 8 6 2 4 5 6 3 3 5 4 3 5 6 8 3 4 4 7 4 3 3 5 7 2 7 5 10 14 15 6 6 15 12 6 8 6 9 6 9 7 7 4 9 2 7 6 8 8 2 1 3 1 1 1 4 9 2 2 3 1 1 2 4 10 6 4 10 2 7 5 6 12 11 12 14 14 7 11 13 13 6 4 7 4 7 5 9 14 4 3 7 3 5 4 8 15 11 16 20 4 12 14 16 16 5 3 7 3 6 2 11 17 5 3 7 2 3 4 9 18 3 3 4 2 3 3 6 19 22 19 23 18 21 18 28 20 24 26 34 14 27 26 30 21 4 4 7 4 1 3 9 23 7 5 13 4 9 5 14

Table 5 shows a number of areas where governments of the left and the right have significantly shifted their attention from the pre- and post-electoral period. Further, left and right governments choose different foci during the electoral campaigns. For example, in the case of crime (topic 12 above), the left was more active in the pre-electoral phase and less so after the elections whereas the opposite is true to some extent for the right. A number of differences are apparent in the table, and more detailed analysis than this paper allows will be needed to explore these questions in more depth.

It has to be added here that any concentration of issues far above the average in the pre- or post- electoral period of course has significant consequences for the rest of the legislature. Put differently, provided that the right deals with immigration especially in post-electoral years, this means that it will deal very little with this during the rest of the legislature, probably even less than their left-wing opponents. The latter appear to keep this issue away from election periods, provided that it does not sell to well to their core voters.

CONCLUSION

20 Estimating the effect of partisanship on attention in three French agendas gives us evidence about the dynamics of political competition in France. Parties generally characterize their opponent as blind to public problems outside their partisan ideology or clients. The underlying idea is that political competition is segmented, with different parties dealing with different problems for different groups. The results presented in this paper underscore that conversely, even in centrifugal political system, political competition is driven by common focus in attention. Different parties draw similar attention to the various policy issues. In such way, parties do not allow to any opponents the monopoly of any policy issue. The environment issue exemplifies this idea. Even if leftwing parties are allied with the Greens, as of environment is settled on the agenda, rightwing incumbents dedicate a substantial amount of attention to environmental issues. At least in France, policy issues are contested. In this logic resides the dynamics of political competition. So parties do not always matter. Around election time, parties use political agendas as leverage to push the voters in their preferred direction. But the effect is only a change in the policy attention queue. There is little static difference. The difference is located in the different ordering in the treatment of policy issues. Whatever is the party, a similar level of attention will be dedicated to a policy issue, only the timing of the attention will change. This paper has allowed us to introduce a new dataset and approach to the study of policy agendas in the French Fifth Republic but of course we have been able only to scratch the surface of the many questions we will need more time and space to explore in the future.

REFERENCES

Badie, Bertrand & Birnbaum, Pierre, 1979, Sociologie de l’Etat, Paris: Hachette.

Brouard & Tiberj, 2005, Français comme les autres ?, Paris: Presses de Sciences Po

CESARI, Jocelyne, 1997, Etre musulman en France aujourd’hui. Paris : Hachette.

CESARI, Jocelyne, 2004, « Islam français : croyances et pratiques ». Cités,hors-série. Paris : PUF, pp.401-410.

Falacci Oriana, 2002, Rage and Pride, Rizzoli Publications ;

Fetzer,Joël & Soper, Christopher, 2005, Muslims and the State in Britain, Franceand Germany, Cambridge: Cambridge University Press.

Fourest, Caroline, 2005, La Tentation obscurantiste, Paris: Grasset & Fasquelle.

21 Geisser, Vincent& Zemouri, Aziz, 2006, Marianne & Allah, Paris: La Découverte.

Huntington, Samuel, 1996, The Clash of Civilizations and the Remaking of World Order, New York: Touchstone.

Klausen, Jytte, 2005, The Challenge of Islam: Politics and Religion in Western Europe, Oxford: Oxford University Press.

Laurence, Jonathan & Vaïsse, Justin, 2005, Integrating Islam: political and religious Challenge I Contemporary France, New York, Brooking Institution Press.

Leveau, Gilles & Kepel, Gilles,1988, Les Musulmans dans la société française, Paris: Presses de Sciences Po.

Lewis, Bernard, 1994, Islam and the West, Oxford: Oxford University Press.

Lewis, Bernard, 2005, From Babel to Dragomans: Interpreting the Middle East, Oxford: Oxford University Press.

March, Andrew, 2007, Islamic Foundations of a Social Contractin non-Muslim Liberal Democracies, APSR,101:2, pp. 235-252.

Norris, Pippa & Inglehart, Ronald, 2004, Sacred and Secular. Religion and Politics Worldwide, Cambridge: Cambridge University Press.

ROY, Olivier. « Fondamentalisme et laïcité en miroir ». Cités, hors-série. Paris : PUF, 2004, pp.143-150.

Sniderman & Hagendorn 2007,

Sniderman & Piazza 1993

Tiberj, Vincent, 2008, La crispation hexagonale, Paris: Plon.

TRIBALAT, Michèle, 2004b, « Le nombre de musulmans en France, qu’en sait-on ? ». Cités, Hors-série. Paris : PUF, pp.21-31.

TRIBALAT, Michèle,2004a, « Une estimation des populations d’origine étrangère en France en 1999 ». Population, vol.59, n°1, pp.51-82.

Venel, Nancy, 2004, Musulmans et citoyens, Paris: PUF, 2004.

Ye’or Bat, 2005, Eurabia : the Euro-Arab Axis, Fairleigh Dickinson University Press

22 Zarka, Yves Charles, 2004, “Introduction”,Cités, Hors-série. Paris : PUF.