The Dynamics of Executive Approval in Fifth Republic France: A Preliminary Empirical Analysis *

Transcription

1 The Dynamics of Executive Approval in Fifth Republic France: A Preliminary Empirical Analysis * Richard S. Conley Assistant Professor Department of Political Science University of Florida 234 Anderson Hall P.O. Box Gainesville, FL (352) x 297 rconley@polisci.ufl.edu Abstract This article is the first to develop a theoretical framework for systematically testing the dynamics of aggregate, monthly presidential and prime ministerial approval in Fifth Republic France ( ). The empirical model combines general economic trends, rally effects from presidential foreign policy/military actions, socio-cultural factors specific to French political life, including domestic strife and strikes, as well as the three experiences of cohabitation or divided government, to gauge changes in monthly public approval for the five French presidents spanning de Gaulle to Chirac and their prime ministers. The results of the time-series analysis suggest that many of the factors presumed to affect presidential approval in the U.S., notably the state of the economy, impact French presidents approval in a similar way. However, French presidents do not gain a significant short term boost from either foreign/military actions or from their annual Bastille Day interview. Instead, domestic strife, such as terrorist attacks in the homeland by international groups or regional independence movements (e.g., Corsica, Brittany), produce a more substantive short-term rally effect. National strikes, frequently provoked by governmental policies opposed by France s formidable unions, drive down approval significantly in the short-term for both the president and the prime minister. Finally, caeteris paribus, both Mitterrand and Chirac won the public opinion duel vis-à-vis opposition prime ministers. The condition of cohabitation when the opposition party or coalition of opposition parties controls the legislature enabled Mitterrand and Chirac to reposition themselves for their respective reelection victories in 1988 and * Author gratefully acknowledges grants provided by the Department of Political Science, University of Florida, and Dr. Gayle Zachmann, Director of the University of Florida Paris Research Center, for this research. The research was completed in Summer 2003 at the Université Aix-Marseille III (Aix-en-Provence, France) and Fall 2003 at the Centre américain de Sciences-Po and the University of Florida Paris Research Center in Paris, France. Paper prepared for the Southern Political Science Association Conference, New Orleans, LA, January 7-10, 2004.

2 Introduction Despite constitutional and developmental similarities between Fifth Republic France and the United States, little comparative analysis of presidential politics in the two countries exists. 1 In both nations executive approval is a constant subject of interest in the press and among the public. But do the dynamics of French presidents public approval approximate those for U.S. presidents? And do the factors that affect French presidents approval similarly structure prime ministers approval? To date, these vital questions have not been systematically analyzed by French or Anglo- American political scientists. The French political science literature, rooted in a tradition of constitutional studies and public law rather than quantitative analysis, lends largely anecdotal insight into individual presidents and prime ministers public approval trends. Moreover, the major French survey organisations, and the annual publications they spawn, provide mostly descriptive evidence to explain patterns sometimes with the aid of current and former government officials whose party leaders are the subject of the surveys under analysis. 2 This article is the first to develop a framework for testing the dynamics of aggregate, monthly presidential (and prime-ministerial) approval in Fifth Republic France ( ). The empirical model combines general economic trends, rally around the flag effects, sociocultural factors specific to French political life, and the three experiences of cohabitation or divided government, to gauge monthly changes in executive approval for the five French presidents spanning de Gaulle to Chirac, and their respective prime ministers. The analysis permits an implicit comparison of the variables affecting French presidents public approval with their American counterparts. The results of the time-series analysis suggest that many of the factors presumed to affect presidential approval in the U.S., notably the state of the economy, impact French presidents approval in a similar way. However, French presidents do not gain a significant, short-term boost from either foreign/military actions or from their 2

3 annual Bastille Day interview an analogy to U.S. presidents State of the Union Address. Instead, domestic strife, such as terrorist attacks in the homeland by international groups or regional independence movements (e.g., Corsica, Brittany), produce a more substantive positive rally effect. Negative events, such as national strikes and protests, frequently provoked by governmental policies opposed by France s formidable unions, drive down approval significantly in the short-term and also negatively impact evaluations of French prime ministers. Finally, caeteris paribus, the approval ratings of both Mitterrand and Chirac benefited from the incidence of opposition control of the legislature as did, ironically, their prime ministers under cohabitation. Yet the French condition of divided government apparently furnished greater advantages for Mitterrand and Chirac in their respective reelection bids in 1988 and 2002 compared to their prime ministerial challengers. This article unfolds in several stages. The first section provides an overview of the French dual executive and the import of presidential domination of the prime minister in the realm of public approval. The second section presents a comparative synopsis of French presidents and prime ministers approval from January 1959-August 2003 and theories that account for changes in public approval. The third section details the methodology for gauging the dynamics of monthly public approval for French presidents and their prime ministers. The fourth section discusses the results of the empirical analysis. The concluding section offers some additional thoughts on the prospects for extending comparative analysis of the French and U.S. presidencies. I. The French Fifth Republic: Towards a Presidential Regime The French Fifth Republic has often been referred to as semi-presidential or a dyadic given a double executive structure (Duverger 1959). In reality, however, the system has moved decisively in the direction of presidentialism, notwithstanding recent periods of divided partisan control of the presidency and the National Assembly (Ardant and Duhamel 1999). Presidents 3

4 have come to dominate prime ministers, creating an imbalance that has prompted some scholars who take a strict constructionist view of the Fifth Republic constitution to call for a reevaluation of institutional powers (Lascombe 2002). Regardless, norms and expectations in the Fifth Republic have evolved in the direction of a presidency-centered, not parliamentarycentred, model of executive leadership that accentuates the centrality of public approval for presidents policy and electoral fortunes particularly in recent periods when the president and prime minister have been from opposing parties. Constitutionally, French presidents in the Fifth Republic were supposed to act as arbiters of republican institutions. But Charles de Gaulle, the first president of the Fifth Republic, quickly dispelled notions that a parliamentary-centered regime would predominate. De Gaulle swiftly reverted to France s long tradition of plebiscitarian tendencies. In 1962 he won a constitutionally-dubious referendum to replace a complex, territorially-based electoral college system with direct election of the president as a means of enhancing his and future presidents claims to be the only representative of all the people (Ehrmann 1983, 7-11). He solidified the presidency and the Palais Élysée, the president s Paris headquarters, as the locus of power and the institution to which the electorate looks for policy leadership. Unwittingly, perhaps, de Gaulle also opened up a greater possibility for divided control of the presidency and the legislature (see Duverger 1986). De Gaulle also introduced another imbalance to the co-ordinate positions of the president and the prime minister for their respective responsibilities. He established the precedent that the president could fire prime ministers at will as he did Michel Debré and Georges Pompidou, despite any such formal-constitutional authority (the Fifth Republic Constitution maintains that the president merely appoints the prime minister, but the latter s confidence rests in the parliamentary majority not the president). 3 In sum, de Gaulle effectively placed the prime minister and the legislature in an inferior institutional position in the Fifth Republic s 4

5 constitutional order and the impact on the relationship between presidential and prime ministerial approval has been palpable. As Parodi (1971) notes, presidents approval ratings typically dominate their prime minister s and subsequent presidents have followed de Gaulle s exemplar to maintain the dynamic. As fusibles 4 or fall guys prime ministers serve at the president s pleasure and are expected to act as lightening rods to deflect public anger for unpopular policies away from the Élysée. Valéry Giscard-d Estaing ( ) required his prime ministers, Jacques Chirac and Raymond Barre, to sign undated letters of resignation (Cole 1998, 77-78). And as the experiences of Pierre Mauroy and Édith Cresson under Mitterrand will be shown, when the prime minister loses public confidence presidents in the Fifth Republic have had few scruples about forcing their departure before their own popularity may be contaminated. It is under the condition of outright split-party control of the presidency and the legislature in the last two decades that the meaning of the double executive in France has become most meaningful. Both Socialist President François Mitterrand and Gaullist President Jacques Chirac endured divided government cohabitation in French parlance at some length. During periods of cohabitation, the system has approximated dominance by the legislature envisaged by many of the framers of the Fifth Republic (see Andrews 1982, 25-33). The president has been placed in a much more inferior position in domestic affairs and has occasionally sustained challenges by prime ministers to preeminence in de Gaulle s sacrosanct domaine réservé (reserved domain) of foreign affairs. Most importantly, the president s traditional domination of public approval has been inverted in periods of divided government. It is precisely because of the president s weak institutional position under cohabitation that the public opinion rivalry between the president and the opposition prime minister has taken on added importance in the last two decades. Like their American counterparts, French presidents hold a privileged position in terms of media attention with which other political actors 5

6 find it difficult to compete. Presidents have recourse to go public (Kernell 1997) and build grassroots support through strategically-timed use of the bully pulpit and selective criticism of the opposition prime minister (Bigaut 1995). Media relations have become more important and public approval figures more prominently in French presidents electoral and policy calculus (see Kaid et al. 1991). In their public relations duel with opposition prime ministers, Presidents Mitterrand ( ) and Chirac ( ) used institutional prerogatives and their symbolic position as head of state and rassembleur to overtake positive evaluations of prime ministers and reposition themselves for reelection victories. Cohabitation is just one of many factors that has influenced variations in executive approval since the 1980s. The essential task ahead is to unravel the complex puzzle of approval dynamics across time from multiple perspectives. II. Executive Approval in the Fifth Republic: A Synopsis The difficulties in assembling monthly data on executive approval in France are significant. No single organisation, such as Gallup in the United States, has tracked approval and publicly made available monthly data for the duration of the Fifth Republic. Historical data are not readily available on the Internet and are scattered across scholarly volumes. The time-series of approval ratings for this analysis was pieced together from several sources. 5 The Paris-based firm SOFRES provides monthly data on its Internet site from 1978 to the present. SOFRES also provided data for the period Monthly data for Charles de Gaulle and Georges Pompidou, and their prime ministers, were culled from Sondages: Revue Française d Opinion Publique published under the auspices of the Institut Français de l Opinion Publique (IFOP) between , and from Les Français et de Gaulle, also published in collaboration with IFOP. [Figure 1] 6

7 Figure 1 presents executive approval data spanning all five presidents and seventeen prime ministers in the Fifth Republic. Several important distinctions are notable for the fortyfour year period covered in the charts. First, early presidents in the Fifth Republic maintained relatively stable public approval trends and finished their terms with a level of popularity nearly as high as when they took office. De Gaulle and Pompidou began and ended their terms above 50 percent, despite some notable peaks and valleys for de Gaulle. Second, greater volatility in the initial and final approval ratings for the subsequent three presidents Valéry Giscard d Estaing, François Mitterrand, and Jacques Chirac is clearly visible in Figure 1. Giscard narrowly prevailed over Mitterrand in the second round of the 1974 presidential election. Chirac s candidacy and divisions on the right had left him in second place after the first round of voting. Giscard subsequently took office with the lowest approval rating of all presidents in the Fifth Republic, at just 44 percent, and did not recover by the end of his septennat. A precipitous decline in public approval for Mitterrand and Chirac in their first terms is also noteworthy. Both presidents began their terms well above 60 percent but fell below 40 percent within 18 to 24 months. Mitterrand and Chirac gradually recovered to position themselves for respective reelection victories in 1988 and 2002, though Mitterrand left office in 1995 the most unpopular president below 40 percent a record for presidents in the Fifth Republic. Scandals surrounding financial dealings, the suicide of former Prime Minister Pierre Bérégovoy, and revelations about his role in the Vichy government and that he had a second family contributed to this dubious distinction (see Thody 1998, ). Third, the data show that across the Fifth Republic prime ministerial ratings tend to track presidential approval in the aggregate. Per se, however, presidential approval does not offer much of an explanation for changes in prime ministers public confidence over time (r=.20). 7 A more important tendency is that prime ministers appointed at the beginning of the president s 7

8 first or second septennat appear to benefit from the president s honeymoon period or état de grâce (Parodi 1997, 92-93). Another trend is that until the mid-1980s, prime ministerial approval was dominated by the president s ratings: The approval level of the president consistently outpaced the prime minister (Parodi 1971, ). Socialist Prime Minister Laurent Fabius broke the trend briefly from before the legislative elections of 1986 brought about the defeat of the Socialist government and the first cohabitation. Finally, it is only in the last two incidences of cohabitation with centre-right Prime Minister Édouard Balladur ( ) and Socialist Prime Minister Lionel Jospin ( ) that the prime minister s approval consistently trumped the president s ratings. Which factors explain the relationship between presidential and prime ministerial approval? Periods of cohabitation notwithstanding, the president s domination of the prime minister reflects the fundamental basis of Gaullist presidentialism. Elected by direct universal suffrage and able to change prime ministers at will, the president is the locus of decision-making and director of the contours of governmental priorities (Parodi 1997, 94). If prime ministers appointed at the beginning of the president s first or second septennat garner relatively higher levels of approval, it may be because they enjoy somewhat greater latitude compared to their successors. Presidents have typically dissolved the National Assembly following their electoral victories if legislative elections were not otherwise scheduled. Presidents tend to appoint prime ministers from the bloc of deputies who were most supportive in the second round of legislative elections. The appointment of the prime minister in these circumstances constitutes a more or less constrained choice on the part of the president. Later appointments between legislative elections offer the president far greater discretion. Subsequent prime ministers are often viewed as the men and women of the president (Portelli 1997, 22-23). 8

9 Periods of cohabitation may invert the president s domination of the prime minister because of the latter s own electoral legitimacy. The first two cohabitations (1986, 1993) occurred at regularly scheduled legislative elections. The third (1997) was the result of President Chirac s disastrous decision to dissolve the National Assembly prematurely without offering a sufficient reason for his actions (Goldey 1998). All three legislative elections that produced an opposition government were widely regarded as repudiations of the president and his parliamentary majority. It is under these conditions that the prime minister, as head of the parliamentary regime, appears to sustain a stronger electoral legitimacy than the president in the court of public opinion, though this dynamic did not enable Prime Ministers Chirac, Balladur, or Jospin to successfully challenge the incumbent president in their immediate electoral bids for the Elysée following periods of cohabitation. 8 Figure 1 alludes to a potential explanation. The data show that the first and third cohabitations benefited not only the prime minister s public approval but also the president s. A central puzzle is thus why prime ministerial and presidential approval have tended to positively co-vary under cohabitation. The French political science literature offers much anecdotal evidence to explain the highs and lows of individual presidents and prime ministers popularity. Many studies emphasize economic trends, domestic strife, and rally effects from international crises. Prime ministers are believed to bear the brunt of negative evaluations of the economic context, though presidents cannot escape responsibility for the broad contours of governmental policies (Fontaine 1994). Giscard and his prime ministers are thought to have been negatively affected by the combined effects of rising unemployment and the energy crisis crises respondents believed would only worsen in the mid-1970s (Sondages 1976, 22-23). Similarly, Parodi (1988, ) attributes Prime Minister Chirac s short-term slide in public approval during the first cohabitation to civil discontent in education and a national strike by the French National Railways in late 1986 and early In a like manner, the parallel erosion in support for 9

10 President Chirac and his first Prime Minister, Alain Juppé, from is believed to have been provoked by negative evaluations of economic policies, scandals, and national strikes (Portelli 1998). Only Prime Minister Édouard Balladur, widely hailed for his low-key and non-conflictual leadership style during the second cohabitation, appears to have held relatively higher public approval for much of his premiership despite an erosion in the national economy (Dupoirier and Grunberg 1993). Periods of international crisis or domestic terrorism are thought to yield rally effects, but cursory evidence suggests they are rather ephemeral in the French context. Ralliements autour du drapeau ( rally round the flag effects ) also do not appear to extend to prime ministers, who are charged much more with domestic affairs. As chief of state, the president is the symbol of French foreign policy and the guardian of republican institutions (see Zarka 1992, ). François Mitterrand apparently benefited from a short-term rally effect during the Gulf War crisis between late summer 1990 and early Spring 1991, though the phenomenon did not mitigate Prime Minister Michel Rocard s declining popularity for domestic policies (Duhamel 1992; Lellouche 1991). Mitterrand s subsequent decision to sack Rocard and hand Matignon to Édith Cresson despite any domestic crisis, followed by Cresson s willingness to include the French Communist Party in her government and a series of policy missteps and civil strife, rapidly left her the most unpopular prime minister in the history of the Fifth Republic (Duhamel and Jaffré 1992). These scattered hypotheses are comparable to well-established theories about changes in executive approval in the U.S., even if the American political science literature is far more rich in the empirical modeling of public opinion trends. Econometric models have shown that the general state of the economy substantively affects approval levels. Presidents Ford, Carter, and particularly George H.W. Bush, are notable examples of presidents app roval ratings suffering rather dramatically from the poor state of the economy (Brace and Hinckley, 1993). The 10

11 economic context weighs significantly on American presidents reelection chances, and Lewis- Beck and Rice (1992) have shown similar effects for French presidents. Moreover, it is a wellestablished fact that U.S. presidents receive a temporary boost during periods of military action abroad or international crisis. As Ragsdale (2000, 44) points out, however, short wars and military actions are most likely to produce positive effects while long, bloody conflicts like Vietnam may have an inverse effect. Third, negative events, often out of the president s control, can also reduce approval levels (Brace and Hickley, 1991). In the American context such nondiscretionary events may include domestic unrest, racial violence, and protests against wars like those suffered by Lyndon Johnson during Vietnam. Finally, presidential approval tends to suffer from a time/competence paradox. Decisions taken by presidents inevitably disappoint some, and their approval declines over the months. The president s administrative effectiveness and competence, however, increase with time. Thus, as Thomas Cronin and Michael Genovese (1998, 108) note, The president s po wer is usually at its zenith early in the terms, when knowledge is the lowest. The diffuse conjectures offered in the French literature on executive approval beg the question of which generalizable claims can be made across time about executive approval. The objective of the analysis that follows is to model the effects of economic conditions, rally effects, and negative domestic events on presidential and prime ministerial approval. The goal is to systematize the study of approval dynamics common across the Fifth Republic. The analysis makes implicit comparisons to U.S. presidents, taking into account socio-political differences in the French context. III. Methodology The empirical model for evaluating changes in French presidents approval replicates variables frequently employed in the analysis of public opinion of U.S. presidents. The structure 11

12 of the equation is ordinary least squares (OLS) regression combining continuous and dummy variables. The equation may be expressed as follows: PP=a+ß 1 (Unemployment)+ß 2 (Time)+ß 3 (Rally)+ß 4 (Strife)+ß 5 (Strikes)+ß 6 (Bastille)+ ß 7 (President)+ß 8 (Cohabitation) + e Where: PP is the president s monthly approval rating; Unemployment is the monthly unemployment rate, ranging from a low of 1.2% (February 1961) to a high of 12.07% (March 1994); Time is the president s month in office at the time of the monthly poll; the data range from 40 to 84 months; Rally is a dummy variable for the month in which the president takes a military action or there is an international crisis; Strife is a dummy variable for the month in which a domestic or international terrorist incident took place in metropolitan France or Départements d Outre-Mer/Territoires d Outre-Mer such as New Caledonia; Strikes is a dummy variable for the month(s) in which a major national strike or protest took place; Bastille is a dummy variable for the month of July in which presidents give their annual Bastille Day televised interview; President is a dummy variable indicating individual presidents first term in office; Cohabitation is a dummy variable for periods during which the opposition party of the president, or coalition of opposition parties, controls the National Assembly; e is the error term for the equation. Unemployment. The theoretical reasons for including unemployment in the model are straightforward. As this key indicator rises, presidential approval should decline. Since the mid- 1970s France has suffered a structural unemployment problem more grave than other members of the European Union (EU). Beginning with the oil crisis and ensuing high inflation of the 1970s, unemployment nearly tripled under Giscard s septennat from 2.7 percent at the beginning of his term to 7.5 percent by May Unemployment reached double-digits under Mitterrand in September 1986, despite early plans for nationalisations and the president s about-face with 12

13 austerity measures by 1983 (Christofferson 1991). Since the mid-1980s unemployment hovered between 8.5 and just above 12 percent for most of Mitterrand s second term and Chirac s first term. The EU stability pact has arguably complicated French attempts to redress economic stagnation through fiscal policies (de Bossieu 2002, 6-12). Member countries in the Euro zone must maintain a budget deficit of less than 3.0% of gross domestic product (GDP) which provoked a crisis between the Raffarin Government and the EU Council of Ministers in Fall 2003 when it was clear France would exceed that threshold for another consecutive year. Monthly unemployment rates come from several sources. For the period , data are not readily available from the Ministry of Labor (Ministère du Travail) or standard governmental statistical compilations. The figures for this period were calculated by the author from raw data provided in the Statistiques du Travail et de la Sécurité Sociale, which was published monthly during that period by the Ministère des Affaires Sociales. The data are housed at the INSEE (Institut National de la Statistique et des Études Économiques) headquarters in the Parisian suburb of Malakoff. Data for are from the Annuaire rétrospectif de la France, (1990) and from the INSEE web site. Time in office. Measuring presidents time in office gauges the impact of decisions over time and gauges relative honeymoon effects. In the American context, Kernell (1978) contends that time only measures time and has dubious explanatory power. However, presidential decisions taken over the course of their terms are widely thought to disappoint many in the electorate and contribute to a decline in popularity even as the president s policy competence may increase. The metric employed in the model is each president s months in office for each term. Before constitutional change by popular referendum in 2000 that reduced the president s term to 5 years (quinqennat) effective with the 2002 election, presidents in the Fifth Republic served 7 years (septennat). The minimum value is 40, and the maximum value is 84. Not all presidents 13

14 have served out their full terms. De Gaulle resigned abruptly in April 1969, four years into his second term, the day after voters rejected his referendum on Senate reform in April Georges Pompidou died in office in 1974, 5 years into his term. 9 Rally effects. Rally effects comprise major international crises and presidential events, including autonomous troop deployments (except for the Gulf War). Foreign observers may be surprised by the number of independent French military actions abroad in the Fifth Republic. French presidents have been particularly active in sending troops to former colonial interests in francophone Africa, including recent peacekeeping missions to the Côte d Ivoire and the Congo. Data for rally effects from were culled from La Cinquième République : L Histoire au jour le jour, Le Monde Dossiers et Documents. Data for were assembled using the searchable archives of the Paris daily Le Monde and annual Dossiers et Documents published by Le Monde, which traces major events of the year For this period, events which received the attention of at least 3 articles were included. 10 A complete list of the data is provided in Appendix 1. Domestic strife. Domestic and international terrorism has been a much greater and continuing problem in France than in the U.S. The phenomenon began in the mid-1970s. Attacks on metropolitan French soil have come not just from international terrorist organisations but also frequently from regional groups, most notably Corsican and Bretagne nationalists. Perhaps the most stunning event in the last decade was the assassination of the Prefect Claude Érignac in Corsica in In similar fashion to the coding for rally effects, data for international and domestic terrorist incidents were assembled from La Cinquième République : L Histoire au jour le jour, Le Monde Dossiers et Documents and the Paris daily Le Monde. Like international crises and military actions, domestic terrorism incidents are expected to have a short-term, positive effect on presidents monthly approval. The list of domestic strife incidents is furnished in Appendix 2. 14

15 Strikes. France has a long history of étatism which is beyond the scope of this article (see Descamps 1972; Moriaux 1982). But it is clear that France s formidable unions (syndicats) frequently view presidents and prime ministers attempts to reform state functions as fundamental attacks on acquis sociaux (entitlements) attained over time. The recurrent problem of national strikes, student demonstrations, and civil strife in the public sector reflects an ingrained suspicion of public officials that complicates governance (Safran 2003, 60-61; Hoffman 1994). One central difficulty for presidents and prime ministers is that public sympathy typically sides with the protestors, not the government whether it is a question of the student protests of May 1968 (Kravetz 1968) or the intermittents (theatre performers ) actions that canceled cultural festivals around the country in Summer Strikes and major demonstrations should therefore have a significantly negative short-term impact on presidential and prime ministerial approval. Strike data were obtained from Annuare rétrospectif de la France, séries longues published by INSEE and from Lenormand Céline, Chronologie indicative de l histoire du movement ouvrier français de la Révolution française à la fin des années These data include major strikes and demonstrations. A complete list of the strike events is provided in Appendix 3. Bastille Day Interview. The president s annual, televised Bastille Day interview with select national news reporters on 14 July is the closest analogy to American presidents State of the Union Address. The occasion casts the president into the media spotlight and provides an opportunity for him to announce new proposals and take credit for accomplishments. It also furnishes a forum to criticize the opposition prime minister under conditions of cohabitation (Elgie 2002, 306). American presidents usually receive a short-term bounce from their State of the Union Address. The Bastille Day interview is included to examine whether the same dynamic holds for French presidents. 15

16 First term presidents. As Figure 1 showed, not all presidents begin or end their terms with equivalent levels of popularity. Some benefit from a larger état de grâce (honeymoon). Caeteris paribus, the inclusion of dummy variables for presidents first term tests knowing whether de Gaulle or Chirac, for example, is president accounts for variations in approval above and beyond economic, rally, and time-decay effects. Dummy variables offer a stylized or aggregate account of individual presidents approval compared to their cohorts. Cohabitation. The French context of divided government epitomizes the battle for public opinion between the president and prime minister. As noted in Figure 1, prime ministers approval surpassed that of Mitterrand ( ) and Chirac ( ) in the last two incidences of cohabitation. Yet these presidents saw their approval increase compared to periods of unified control. Dummy variables for periods of cohabitation are employed to gauge more precisely the relative effect of a divided executive on presidential (and prime ministerial) approval, controlling for other variables. Prime Ministers Approval Analysis The approval model for prime ministers in the Fifth Republic is identical to the presidential model with only a few exceptions. First, dummy variables for periods of cohabitation were dropped. Dummy variables for individual prime ministers capture not only cohabitation effects but other effects not measured by the continuous variables in the model that are particular to the tenure of several occupants of the Hôtel Matignon. Second, the prime ministerial model was divided across two periods the de Gaulle/Pompidou era and the Giscard- Mitterrand-Chirac era precisely to avoid colinearity problems in light of the large number of prime ministers in the Fifth Republic (17). As the narrative takes up in greater detail below, the inclusion of dummy variables for individual prime ministers enables an evaluation of which leaders were far more or less popular than economic conditions, strike activity, or the time-decay function would suggest. In 16

17 particular, political scandal and prime ministers challenges to presidential status and authority explain precipitous declines in public confidence for such prime ministers as Raymond Barre and Édith Cresson, noted in Figure 1. IV. Results The results of the models for presidential and prime ministerial approval are shown in Tables 1 and 3, respectively. 12 For both tables, Newey-West standard errors are reported to correct for a violation of assumptions in the regression analyses. For the presidential approval model the Durbin-Watson statistic of.58 and the Cook-Weisberg statistic of 1.96 (p <.16) following standard ordinary least squares (OLS) regression suggest evidence of autocorrelation of the residuals and moderate heteroskedascity. Visual inspection of the residuals plotted against the predicted values confirms the presence of autocorrelation. Similarly, for the prime ministerial approval model divided between the de Gaulle/Pompidou and Giscard/Mitterrand/Chirac eras, the Durbin-Watson statistic is.30 and.49, respectively. The Newey-West procedure assumes that the error structure of the models is heteroskedastic and possibly autocorrelated up to some lag. Standard errors are generated with a heteroskedasticity - consistent covariance matrix that corrects for the problem of under-estimating the standard errors (see Davidson & McKinnon 1993, 552). From Stability to Instability: Presidential Approval from de Gaulle to Chirac The results in Table 1 confirm the majority of hypotheses of the study. The state of the economy drives down presidential approval rather significantly, and presidents do tend to lose public confidence over time. Similarly, major strikes and demonstrations carry a negative effect on presidential approval. Rally events from major presidential foreign policy events and military actions show a positive correlation to public approval, however, they do not reach statistical significance. Incidents of domestic strife, on the other hand, yield a short-term gain in public confidence. Finally, the analysis confirms that Mitterrand (1 st term) and Chirac saw their public 17

18 approval increase under conditions of cohabitation, controlling for other factors. Let us now examine these findings in more detail. [Table 1] The constant in the model in Table 1 suggests that all things being equal, presidents in the Fifth Republic can expect an initial approval rating in the mid-60s. The coefficients for individual presidents show that knowing whether de Gaulle, Pompidou, Giscard, Mitterrand, or Chirac held the presidency does not explain much of the variation in approval dynamics (though Chirac was more unpopular than the rest at the beginning of his first term). Instead, economic conditions, as measured by unemployment, and have become major factors in the instability of public approval ratings for later presidents, compounded by negative public evaluations due to strikes. Each increase of 1 percent in monthly unemployment yields a net loss of just under 1.2 percent in presidents public confidence. The effect is most pronounced for Giscard, Mitterrand, and Chirac. To highlight the importance of the economic context, Table 2 presents the mean effects of unemployment on monthly approval ratings across time by multiplying the coefficient by the maximum, minimum, and average values for the indicator. The data accentuate that during the immediate post-world War II economic boom, de Gaulle and Pompidou reaped the benefits in terms of public approval. With unemployment consistently under 3 percent from , public confidence in the first two presidents of the Fifth Republic was barely affected. This scenario contrasts mightily with Mitterrand and Chirac, in particular. Using the constant of 66 percent in Table 1 as a baseline measure and holding all other variables at 0, Table 2 indicates that on average, structural unemployment alone drove their ratings down to the mid- 50s. The average, net loss for Mitterrand was just above 11 percent and for Chirac slightly greater than 12 percent double the effect for Giscard. Alas, negative evaluations of the 18

19 employment context have been a principal albatross around the necks of recent presidents of the Fifth Republic. [Table 2] Presidents must also be wary of provoking major strikes and demonstrations with unpopular policies led by their prime ministers. Such events, often at the behest of public sector unions, yield a short-term decline of over 3 percent in the president s approval. Of course, the coefficient reported in Table 1 is a standardized effect across time. Two presidents Charles de Gaulle and Jacques Chirac were subject to the most fluctuation in public confidence owing to strikes and demonstrations. Charles de Gaulle s presidency was a period of substantive modernization of the French state and arguably significant progress in terms of social policy. But his two terms were also a time of intense social conflict. Of the 45 strikes and major demonstrations catalogued between 1959 and 2003, 19 (41%) took plac e during de Gaulle s reign from an average of one strike or major demonstration every five and one-half months. The most memorable set of strikes took place in May The student uprising in protest of crowded university settings subsequently sparked public sector unrest. Inexplicably, de Gaulle departed for Germany to meet with his confidant General Massu, leaving Prime Minister Pompidou to negotiate a peaceful settlement to the chaos (Thody 1998, 33-34). De Gaulle s public approval fell by some 8 points over previous months. Indeed, the model helps explain some of the peaks and valleys in de Gaulle s relationship with the public. Visual inspection of Figure 1 shows that general strikes in April 1961, May 1966, May 1967, and the miners strike of March 1963, were associated with decreases in public confidence from 2 to 5 percent for the enigmatic president, consistent with the stylized coefficient for strikes in the regression model. Jacques Chirac ranks second in terms of the frequency of strikes from 1995 through mid A total of ten events took place over these eight years for an average of one every ten 19

20 months. The most pivotal events occurred in 1995 and 1996 the beginning of Chirac s septennat when strikes by transportation workers and truckers in protest of the Juppé government s proposed social security reforms paralyzed the nation. Once again, visual inspection of Figure 1 underscores the validity of the generalized effect of strikes in the model (Table 1). The advent of these strikes erased Chirac s état de grâce, already in jeopardy with the resumption of nuclear weapons tests in the Pacific, dropping public confidence between 2 and 4 percent in late By early 1996 Chirac s turn toward budgetary austerity provoked negative evaluations of his plans to combat unemployment among no less than 87 percent of survey respondents (Portelli 1998, 15). The effects of economic conditions and strike activity underscore that like their American counterparts, French presidents in the Fifth Republic must hit the ground running (Pfiffner 1996) to maximize leverage during their honeymoon period while they benefit from a surfeit of public confidence. If the general economic climate takes a heavy toll on approval ratings, time in office also yields a significantly negative effect. 13 For approximately every 6 months in office presidents lose a full percentage point in public confidence. Put another way, a president serving out his full septennat of 84 months can expect a net loss of about 14 points as the next election approaches. French political analysts often make reference to the president s 100 days, and for good reason. The difficult, ineluctable decisions presidents must make are bound to alienate some in the electorate and disappoint others. As in the American case, presidents resources are greatest when their competence level is arguably weakest, at the beginning of their terms. All told, the time-decay function, combined with difficult economic conditions since the 1970s, explains the greater propensity for the steeper valleys in presidential approval in the post - Pompidou era. Presidents cannot count on either symbolic television appearances or international crises and military actions to reverse the negative impact of economic conditions, strikes, or time-decay 20

21 effects. American presidents major televised addresses often lead to a short-term increase in public support (see Edwards 2004, 38-39). For French presidents, their annual Bastille Day interview, however symbolic, does not yield a similar effect. Moreover, presidents in the Fifth Republic do not benefit from events thought to produce a rally round the flag effect. The coefficient for rally events of just over 1 percent does not reach statistical significance. Employing interaction terms for individual presidents and rally events does not change the outcome. The failure of such actions to boost public approval constrains presidents ability to rebound from negative evaluations of the economy and the impact of civil unrest. Which factors explain why, across the history of the Fifth Republic, neither military actions, largely confined to the African continent, nor significant international crises bolster confidence in the president? Mitterrand s support for the American-led coalition in the Gulf War of 1991 is a case in point about public ambivalence toward military actions. The fourth French president of the Fifth Republic was successful in mobilizing public support behind the American-led coalition. His public approval increased from the upper 50s during the crisis in Kuwait and peaked in Spring 1991 in the low-60s. Nevertheless, this bounce was effectively far less compared to President Bush and Prime Minister Major (Zarka 1992). Analyses of opinion toward the Gulf War specifically suggest that a general contradiction specific to French political culture may explain the dynamic. On the one hand, Mitterrand masterfully employed the symbolism of his chief of state role through public outreach to marshal unity behind the allied coalition (Duporier 1992, 126; 144). On the other hand, many French simultaneously evidenced a strong desire to avoid participation in any war that might entail potentially negative economic consequences (Lellouche 1991). A strong majority of French expected major protests against the war at the outset of allied military actions. Moreover, a majority of younger French (18-25) expressed solidarity or sympathy with the pacifist movement, which has roots in France s libertarian as well as Christian political traditions 21

22 (Dupoirier 1992; Winock 1991). The lesson of the Gulf War is that while presidents may be able to effectively galvanize public support behind military actions, there remains a more profound, underlying skepticism of such actions among the French compared to Americans that undercuts significant rally effects. Domestic strife, however, does yield a moderate increase in public confidence in French presidents. Terrorism on French soil perpetrated by international- or domestic-based groups spurs a short-term gain of 2 percent in public confidence following such incidents. The effect is entirely consistent with the president s status as the symbolic and constitutional guardian of republican institutions in the Fifth Republic. When the integrity of the nation has come under attack, the public has rallied around presidents calls for national unity and the redoubling of efforts to bring the architects of terrorist attacks to justice. During the numerous terrorist incidents in Paris the mid-1980s and civil unrest in New Caledonia Mitterrand actively sought to reassure French citizens. Similarly, the increase in terrorist activities by Corsican nationalists in the late 1990s prompted steadfast condemnations by Jacques Chirac, a return to the leitmotif of the social fracture in France, and a focus on law and order. Both Chirac and Interior Minister Nicolas Sarkozy were widely hailed in the métropole for the ultimate arrest in summer 2003 of Yvan Colonna, the suspected assassin of Corsican Prefect Claude Érignac five years earlier (Gurrey 2003). If the advent of the first cohabitation in the Fifth Republic raised concerns about a crisis of the regime, by the third cohabitation there was little question that the French electorate not only approved of a divided executive but believed the arrangement would function well (Quermonne 1987). The data in Table 1 confirm that the first and third incidences of cohabitation boosted presidents public approval, controlling for economic conditions, strikes, and time-decay effects. During the first cohabitation from public confidence in François Mitterrand increased by nearly 19 percent. Similarly, in the extended period of 22

23 cohabitation from Jacques Chirac benefited from a net increase of nearly 13 percent, compensating for the otherwise negative evaluations that had plagued his early first term (noted by the individual dummy variable indicating a loss of a little more than 7 percent compared to other presidents). Only the second cohabitation, with Édouard Balladur as prime minister, did not reinforce public confidence in the president (discussed in the next section). Mitterrand s increase in public support during the first cohabitation stems from several plausible sources. The president s popularity had decreased substantially following his volteface on economic policy and the austerity measures adopted beginning in His job approval reached a nadir during his first septennat in early 1984 (Figure 1). With his popularity effectively halved compared to 1981, Mitterrand s côte de confiance was among the lowest for presidents up to that point and had likely bottomed out among all but his most fervent supporters. Replacing the ever-more unpopular Prime Minister Pierre Mauroy with the young, technocratic Laurent Fabius did not immediately benefit Mitterrand s own fortunes, though the appointment was strategic and aimed at repositioning the Socialist Party in the center of the electorate (Duhamel 1985). The payoff came with Fabius ouster and the introduction of proportional representation in the legislative elections of 1986, which brought about the first cohabitation. Mitterrand s popularity immediately rebounded by some 10 points a seeming contradiction of the unprecedented institutional arrangement. His popularity remained well above 50 percent, and occasionally peaked above 60 percent, for the rest of the first cohabitation (Figure 1). With the introduction of proportional representation Mitterrand and the Socialists sought to maximize their own partisan support and simultaneously set a trap for opposition Prime Minister Chirac. Chirac was forced to manage a very narrow centre-right coalition and the proportional scheme enabled Jean-Marie Le Pen and Front national to capture 35 seats. Chirac and his majority coalition were plagued initially by questions about whether to work with Le 23

24 Pen, and later vexed by the frequently disruptive tactics by the feisty, extreme right leader. Mitterrand refused to go along with the prime minister s desire to expedite controversial legislation by ordonnance (which requires the president s signature), forcing Chirac to submit major elements of his programme to the National Assembly through the standard legislative process. These projets de loi concerning privatizations and employment drew the intense opposition of the Left and created a forum in which Mitterrand had ample opportunity to critique the Prime Minister and his majority. The net effect of the first cohabitation was a negative media focus on Chirac while Mitterrand dutifully played the role of statesman on foreign and defense policy and selectively censured the government s domestic policies (Bigaut 1995, 12). Persistent divisions on the Right, and the rebound in public confidence in Mitterrand, facilitated the president s reelection campaign. The first round of the 1988 election split the vote on the Right between Chirac (19.95%) and Raymond Barre (16.5%), with Mitterrand culling 34 percent of the v ote. In the second round Mitterrand prevailed easily over Chirac with 54 percent of the suffrage. Cohabitation did not work to Mitterrand s advantage, however, in his second septennat. Table 1 shows that divided government from had no substantive impact on Mitterrand s approval. As Figure 1 showed, Prime Minister Édouard Balladur s job approval though it declined over time consistently outpaced Mitterrand s between 10 and 30 percent. The second cohabitation accentuates the relative political legitimacy of the president and the prime minister under a transformed electoral context. As noted earlier, scandals certainly beleaguered public confidence in Mitterrand during his second term. And the economic downturn, an increase in unemployment, and internecine warfare in the Socialist camp scarcely aided the president s standing (Jaffré 1994, 147). Yet the electoral circumstances of cohabitation cannot be underestimated in analysing Mitterrand s foundering public approval. The legislative elections of 1993 constituted a far greater rejection of the president and his party compared to 24