GROWTH AND POLITICAL IDEOLOGY IN A DEMOCRACY: THE FRENCH EXPERIENCE, 1871 - 2009

Franois Facchini Mickal Melki -Centre d'conomie de la Sorbonne, Paris 1, France-

Abstract. This paper examines the influence of political ideology on economic growth in a democracy - such as the French one - from 1871 to 2009. It does so by addressing three main issues: the property and the reliability of an index of political ideology in the long-run, the robustness of the relationship between ideology and growth and the specific channels through which political ideology affects economic performance. The main conclusion is that there is strong empirical support that a rightwing ideology in parliament has been more favorable to economic growth. It appears that the three transmission channels from political ideology to growth are inequality, unemployment and public spending. By contrast, there appears to be less empirical support for explanations based on the effect of inflation on growth.

Key-words: Political ideology, partisan model, GDP growth, France.

1. Introduction The present paper examines the effect of political ideology on economic growth in a democracy in the long run. To do that, we use annual data on France on the period 18712009. More precisely, we propose to focus on the parliaments political ideology and its effects on the implemented policies and their economic performances. That raises several questions: does ideology matter in economic policy-making? If it does, which ideology is more favorable to economic growth? Finally how does political ideology work on growth?

The issue seems to have been traditionally neglected by economists. This is paradoxical, as economists were at one time quite ideologically active with regard to their preferred policies and institutions for implementation. Indeed political ideology is rarely taken into account in the explanations of economic growth. As proof, Sala-I-Martin et al (2003) rank, according to a Bayesian method, the 67 most common variables in the growth model literature, without referring to any ideology variable. Nevertheless, one can identify two kinds of literature dealing with the influence of political ideology in economy. The oldest one that dated from the 1970s lies within the framework of the partisan and business political cycle literatures while a more recent literature refers to political ideology to account for a wider range of phenomena such as public spending, inequalities, economic growth...

The original political business cycle theory (Nordhaus 1975) tends to deny any role of political ideology by assigning only electoral motivations to politicians and by taking the economic actors away of rationality. Therefore both leftwing and rightwing governments are supposed to artificially increase GDP growth before elections. As for the partisan approach (Hibbs 1977), it acknowledges a role for the governments political affiliation. Hibbs (1977) initially showed that in 12 post-war capitalist democracies, the leftist governments implemented expansionary fiscal and monetary policies to boost GDP growth and keep unemployment low, whereas the rightwing parties first and foremost struggled against inflation. But the influence of these policies is, at least in the long run, neutralised by economic actors. Under the rational expectation assumption, voters immediately anticipate the policies effects, making them ineffective. However Alesina (1987) showed that under several conditions, the government can work on the national economy. But this effect is never lasting because of the adaptive expectation assumption; the economic actors quickly 2

correcting their expectations. In this perspective, there is a partisan effect on growth only during the first two years of electoral terms.

Concerning the empiric tests, Alesina (1988) confirmed his theory with a study on the United-States. Democrats and Republicans would record similar growth rates during the second part of their terms, but the former would perform better during the early years of the term, as they implement monetary (higher inflation) and expansionist budget policies (higher deficit and spending) that are not immediately expected by voters (Alesina 1988, p.36). Alesina et al. (1997, chapitre 6, p.174) confirmed this finding for 18 OECD countries on the period 1960-1993. Finally, Dubois (2005) performed a similar test for the French case from 1979 to 2003 and established that the GDP growth decreased by 0.3% during the six quarters following the election of a rightwing government and increased by 0.5% during the same period under a leftwing government. Thus Dubois confirms the rational partisan cycle theory for France with a stronger partisan effect in favour of leftwing parties.

In a more recent literature, ideology has become the main variable to explain numerous phenomena. In a meta-analysis, Imbeau et al (2001) listed more than 600 cross-sectional studies over the effects of government ideology in terms of size of the State, social welfare, economic, foreign and defence policies to check the politics matters hypothesis (Imbeau et al 2001, p.1). But the huge number of studies over the governments policies contrasts with the few studies on their economic performance. Among these scarce works, the paper of Bjornskov (2005) is the first one to study the effect of political ideology on growth outside the framework of the political business cycle theories. He found empirical evidence for 58 OECD countries, on the 1975-2000 period, showing that countries to the right of the average experienced one quarter of a percent additional growth per year compared to the lefty countries (Bjornskov 2005, p.140). The positive effect of the rightwing ideology would stem from better legal systems and less government intervention. Accordingly, Bjornskov (2008) established for a panel of 74 democratic countries on the same period that the higher the income inequalities are, the more a government shift to a rightwing ideology improves growth. At the mean inequality level, a move from a center to a center-right government is associated with a .28 percentage points increase in the annual growth rate (Bjornskov 2008, p.306). One can also refer to the study of Bjornskov and Potrafke (2010) on the positive 3

effect of the rightwing political ideology on economic freedom and implicitly on economic growth for the Canadian provinces.

However all the current empirical studies in the long-run effect of ideology suffer from a common pathology due to the fact that they are cross-sectional. Indeed they compare the effects of governments ideology either for a group of countries or for a set of states, provinces or regions inside a country. The pathology would stem from fairly short observation periods that generally do not exceed 25 years and their focus on a small group of countries (Potrafke 2010, p.3). Imbeau et al. (2001, p.3) add that partisan effects would be too subtle to ensure sufficient robustness of cross-sectional statistical estimations. They refer here to the difficulty to measure the national differences between the various rightwing and leftwing ideologies of the countries of the studied panel. For instance, using the Database Political Institutions (DPI) (Beck et al. 2001) that measures the ideological orientation of the chief executive of 177 countries on the period 1971-2000, we find the shady result that France was perceptibly more rightwing than the US between 1975 and 2000 (Bjornskov 2005, p.144). Finally the last pathology stems from the unresolved identification issue. The bulk of the literature on the influence of ideology does not take seriously into account the problem of reversal causality, according to which GDP growth can also influence political ideology.

In view of these points, the present paper provides a different perspective from the current literature. It indeed proposes to adopt an historical approach through a time-series analysis, which focuses on a single country, France, to study the issue of the economic effects of political ideology. This is quite innovative because there is no long-period study on the topic for France and more generally, for the other countries. Moreover focusing on the French democratic experience of almost one and a half century (1871-2009) leads us to propose a methodology to build an ideology index in long period, not available in the current literature. Furthermore a time-series analysis is automatically immune to the above main pathologies. First the length of the observation period 1871-2009 is safe from any criticise. Secondly, focusing on a single country banishes the problem of the subtle differences in national political ideologies. Thirdly we suggest to discuss the identification issue and to properly tackle it. In this manner, this article paves the way for long-period analyses based on 4

quantitative history, with ever more reliable and accurate sources - such as the works of Maddison (1995, 2005), Barro and Ursua (2008) and Toutain for the French case.

The main conclusion of this article is to confirm the results of the modern literature over ideology and growth. Indeed it provides empirical evidence that rightwing governments have been more favorable to economic growth in France from 1871 to 2009. Three transmission channels explain the relationship between ideology and economic performance: the structure of income inequalities, unemployment and public spending.

This paper is structured as follows. In the second section, it proposes an historical background of political life and economic growth in France (2). In the third section, it presents the data, the ideology index and the used macroeconomic variables, as well as the empirical strategy (3). The forth section presents the tests and proposes to interpret the results (4). The fifth section concludes (5).

2. Historical background of political ideology and economic growth in France (1871-2009)

2.1 One of the oldest democracies The French democratic experience is particularly interesting for our kind of study because the current French regime is one of the oldest democracies in the world and thus provides a long period to analyse. France was the First European country to introduce universale male suffrage in 1848 and after having experienced, in the wake of the French Revolution, two empires, three constitutional monarchies and two attempts of Republic, France adopted de facto in September 1870 a stable parliamentary republic with the third Republic (18701940). Nevertheless we chose to start our study only in February 1871, the date of the first legislative elections of the Third Republic, for the National Assembly in charge of the preparation of a new constitution. But the regime is really established only in 1875 with the vote of the Wallon Amendement (bill) and the constitutional laws. The National Assembly is split in an Upper Chamber, the Senate and a Lower Chamber, the House of Deputies, the sole institution elected by direct universal suffrage.

That is the first reason why we focus in this study on the Lower Chamber to assess the political ideology of parliament. A second reason is that the Third Republican is known to be the Republic of the deputies, leaving almost no room for the executive power. This bicameralism system characterizes the working of the French democracy until now, except a brief interruption with the Vichy Regime (1940-1945). The forth Republic (1946-1958) is in the continuation of the previous Republic with roughly the same working of institutions whereas under the Fifth Republic (1958-nowadays), the Parliament is composed with the National Assembly, that becomes the Lower Chamber and the Senate. So after 1958, we focus on the National Assembly to characterize the parliaments political affiliation.

2.2 One of the oldest ideological divides The main difficulty inherent to a long period analysis is to identify a clear and constant divide between right and left. If after the 1930s and in the post-war society, the ideological divide is clearly structured around economic issues, that is far from being obvious for the early years of the French democracy. Indeed, during the first three decades of the Third Republic, until the Ralliement of Church to Republic in 1998, the main ideological opposition is between a republican left in favour of a republican and lay regime and a conservative right that supports a monarchist and religious government system. But in spite of the obvious ideological evolutions inside the right- and leftwing parties all along our period, the issue of the permanence of a leftwing ideological divide has been strongly debated among historians and analysts of the French political life. It appears that two trends have cohabitated all along the French democratic experience: on the one hand, the variety and the diversity of the political groups and parties and, on the other hand, the dualism of the fundamental (right/left) tendencies (Goguel 1986, p.19). The deepness of the divide is reflected in the electoral behaviour of the French voters, which tended to prove the existence and the permanence of a fundamental connection between all the lefts and all the rights. For instance, studying the votes in western France, Siegfried (1930) supplied us with striking examples of stability of the relative weight of the two main tendencies since 1871. And the left-right divide seems to remain a touchstone in the French political landscape (Mayer 1997, p.15) and in general (Budge et al. 2001, p.22; Markussen 2008, p.342).

In view of the history of the left in France, the left-right divide seems to be over the labour market regulation, the denunciation of the harmful effects of the mur de largent1 on social welfare, the necessity of a redistributive tax policy, the national interest of a colonial strategy with a civilizing purpose. On the contrary, the right, commonly called the group of the established order, as opposed to the movement group (Goguel 1956), rather tend to control and keep order and the main equilibrium.

A brief outline of the political choice of the various leftwing governments since the consolidation of the French democracy allows to confirm these divides and their historical permanence. The left has governed under the three republics, at the end of the 19th century with opportunist republican coalitions, after the First World War with the Cartel des gauches (1924 1926, 1932 1934), before the Second World War with the Front Populaire (1936 1937), under the Fourth and the Fifth republic with socialist governments (1981 1995 et 1995 2002). If we focus on the compositions of the Lower Chamber of the Parliament on the period 1871-2009, the leftwing parties governed during 79 years, as compared to 55 years for the rightwing parties. In spite of the evolution of the leftwing ideology throughout the time, one can identify similarities in the implemented policies.

In 1971, left and right are opposed upon the war debts, the financial costs of the colonial strategy (Fridenson 2005, p.587; Becker 2005, p.313). As early as the 1870s and from the Waldeck-Rousseaus government in 1899, the left developed its main issues (Duclert 2005, p.211) such as the regulation of working time and working conditions, wealth redistribution trough tax2 according to the Solidarist ideology (Audier 2010) and the denunciation of the mur de largent . Even though some governments as the Combess one or the Bloc des gauches are often presented as socially shy (Candar 2005, 223), all the main laws on labor market ware voted by lefty majorities in parliament: creation of a minimum wage in 1950 by a leftwing coalition, the working 6-hour day for the under 12 years old children in 1974, the 11-hour day for the 16-18 children and women in 1892, the general 10-hour day, the working 6-day week in 1906, the working 48-hour week and the 8-hour day (Bloc National) in 1919, the 40-hour week in 1936 (Front Populaire), the 39-hour week in 1982 and the 351

Expression of the 1920s that refers to the opposition of the banking and financial circles to any economic and social reform in France. 2 Creation of the IRPP: http://piketty.pse.ens.fr/fichiers/public/Grasset2001/Annexes/AnnexeC.doc

hour week in 1998 and 2000 and more generally the entrance and exit conditions in the labor market.

The left seems to have been rather in favor of public spending, inflation (Fridenson 2005, p.589), nationalizing of certain strategic public sectors, redistributive policies likely to decrease inequalities (Fridenson 2005, p.592 589) and a strict regulation of freedom of contract.

2.3 French growth and ideology In view of the differences in policy-making, we can reasonably make the assumption that leftwing and rightwing governments have not had the same economic performance (H1, figure 1). Indeed, the government in power does not react only to its environment but also make real choices of economic policies according to its ideology (H2, figure 1). And then the implemented policies, which stand for the transmission mechanisms, affect economic performance (H3, figure 1). In this perspective, it is natural to connect the economic and political histories of France and to wonder about the effects of the political powers ideology, supposed to represent the peoples ideology, on economic growth.

Figure 1: Transmission mechanism between ideology and growth H1

VOTE

PARLIAMENT IDEOLOGY

H2

ECONOMIC POLICIES

H3

ECONOMIC PERFORMANCE

The French economic history can be presented as follows (see Figure A.2 in appendix). The French economy experienced two main critical stages until the Second World War: 1860 1890, 1929 1939 (Asselain,1984a, 172). On the contrary, from 1945 to 1973, it experienced a steady growth period. After that, the GDP growth slow downed and was even negative in 1993 and 2009. War played an important part in the evolution of the French growth at the end of the 19th century as well as during the 20th century.

The 1860-1890 period was called before 1929 la Grande Crise3. It corresponded approximately to the 1873-1896 Kondratieff cycle of long-run prices decrease. In France, this stage began during the decade 1860 and wad confirmed by the decrease in the GDP per capita during the 1880s (Lvy-Leboyer 1971). The origin of the decrease was the American Civil War, the phylloxera plague that destroyed harvest, the decreasing investment return in railway, the weight of trade between France and Great Britain and the agricultural depression. After the prosperous 1890-1913 period (Asselain 1984a, Chapter 4), followed the 1914-1918 war and the rebuilding from 1918 to 1929. However, as early as 1924, the level of the French GDP and national income is the same as in 1913 but growth is quite fast (Asselain 1984b, p.25). 1929 is obviously a date of break down followed by a recession and imbalance period (1930 1935), which, contrary to numerous countries, lasted beyond.

The 1937 reforms, under the Front Populaire, of the Blums government had only temporary effects. The 1937 and 1939 increases did not make up the deep recession of 1938. The economic instability corresponds to a high political instability: fall of the Blums government and the accession of the Chautempss government in 1938. On the contrary, during the 1945-1973 period, the annual GDP growth is high. On the 1949 1969 period the average annual growth rate is 4,6 and 6,6 from 1969 to 1973 (Caron 1981, p.158). From 1975, as numerous European developed countries, France experienced very low growth rates compared to prior and to other big formerly industrialized countries like Great-Britain or United-States. All the main stages of the French growth are provided by the below table extracted from Fontvielle (1976, p.173).

Table 1. Production long cycles

High growth Low growth 1850-1868 1870-1889 High growth Low growth 1894-1913 1913-1921 High growth Low growth High growth Low growth 1922-1941 1942-1946 1947-1973 1974-2009 Source: Fontvieille (1976), p.173

3. Data and empirical strategy To examine the effect of political ideology on the economic growth of France, we gathered annual data coming from the quantitative economic history works carried out with France as its subject. The list of variables (definition and sources) is provided in appendix, table A.1.
3

The Great Crisis

3.1 The building of the Ideology index and the explained variable First of all, we built our own index to measure the yearly parliaments political ideology over the period starting from the Third Republic to nowadays. This index is based on the composition of the Lower Chamber (elected by universal suffrage) of the Parliament, supposed to represent the ideology of the people and which possesses the legislative power. Thus we consider the composition of the Chamber of Deputies during the Third and Forth republic and the National Assembly during the Fifth Republic. As we put it before, even if the Third Republic started one year earlier, we start our index in 1871, the date of the first legislative elections in the Third Republic, for an assembly in charge of the new constitution. So we use a continuous variable (Parliament ideology) that indicates the percentage of the left-wing seats in the Chamber and that is distributed between 0 and 1; 0 reflecting the absence of leftwing members in the Lower Chamber, and 1, a Chamber totally filled with leftwing deputies.

Concerning the tricky issue of the ideology intensity of the various rightwing and leftwing parties, we do not use the Beck et all. (2001)s method which ranks the French parties along a left-right axis, therefore allowing for the distinction between extreme and moderate parties. By doing so, we could not identify the ideology of the majority in the Chamber because France was often governed by left-of-centre or right-of-centre coalitions. That would not be relevant to reserve a special treatment for the moderate parties since they enabled to form majorities and they fully governed inside these majorities. Furthermore we did not include in our index the few independent deputies and worked to ensure that their presence would not influence the ideology of the majority. And concerning the election years, during which the majority can shift, we decided as a convention to consider the composition of the incumbent Chamber, that is to say the percentage of leftwing deputies before election. But we made sure that this convention would not influence our tests by also testing the ideology variable that refers to the new elected Chamber during the election years.

Lastly, to tackle the other tricky issue of the Vichy regime (1940-1944), we decided to attribute this period as much to the right as to the left. This drove us to assume that, during these years, the Chamber was evenly shared between both political ideologies, and 10

therefore the variable Parliament ideology is coded 0,50 from 1941 to 1945, according to our convention on the incumbent Chamber. We also propose a dummy variable (Parliament ideology dummy) that shows the value 0 when the majority of the Chamber is rightwing and 1 when it is leftwing. Therefore this index is not available for the Vichy period. The building of the index allows to present on the figure 3 the French political life from 1871 to 2009 and the changes in power.

Figure 2. Evolution of the Ideology index for the French parliament from 1871 to 2009
100,00 90,00 80,00 70,00 60,00 Ideology 50,00 40,00 30,00 20,00 10,00 0,00 1871 1877 1883 1889 1895 1901 1907 1913 1919 1925 1931 1937 1943 1949 1955 1961 1967 1973 1979 1985 1991 1997 2003 2009

Source: Website of the French National Assembly: http://www.assemblee-nationale.fr/histoire/ index.asp

The explained variable is the annual GDP growth rate in France for the period of 1871-2009. To the best of our knowledge, we chose Maddisons website as the most reliable source from among the different long-term data available for the French GDP (figure 3). We propose to compare it with the data of J-C. Toutain (1997), available only from 1890 (figure 1, appendix). From our ideology index and the data on French GDP growth, it appears that the French economic growth was shown to be higher under the legislature with a right-wing majority during the period of 1871 2009. In fact, the average growth rate under a rightwing majority at the Lower Chamber is almost 4% as compared to 2.40% under a left-wing one. The trend is obvious with a basic regression of the average growth rate per term on the average ideology of each term (figure 4).

11

Figure 3. French GDP growth rate (1869-2009)

50.00% French GDP growth rate

S
40.00% 30.00%

20.00%

10.00%

0.00% 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 -10.00%

-20.00%

Source: Maddisons website

Figure 4. Average Growth per term and Parliament Ideology 1871 2009
12 10 8 Average Growth per term 6 4 2 0 0,00 -2 -4 -6 Ideology 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00

3.2 The explicative variables Concerning the explicative variables, we use a basic growth model proposed by Alesina et al. (1997, chapter 6) in which the national growth in t is explained by the national growth in t-1 and the external growth in t. We voluntarily do not use in our baseline equation variables that can be endogeneous in our model, such as human capital and the degree of openness of the national economy. The former can result from a governments political choice to invest in education and the latter can mirror the governments political ideology towards trade

12

policy, as Nollen and Iglarsh (1990) or Dutt and Mitra (2005) empirically showed. Thus, in our baseline equation, we propose to explain the annual French GDP growth by the national GDP growth of the previous year, the European GDP growth and the Parliament Ideology index. The variable European Growth that stems from Maddisons website, corresponds to the annual GDP growth rate of 11 European countries, which represents the main traditional trade partners of France. The two growth variables are naturally expected to positively influence the French growth rate:
FrenchGrowtht = 1 + 2 FrenchGrowtht 1 + 3 EuropeanGrowtht + 4 ParliamentIdeot (+) (+) (?)

Besides these variables that obviously do not depend on parliaments ideology, some other potential explicative variables have traditionally been in the heart of political debate. As it is presented in the previous section, the political divide was over issues such as protectionism, inflation, unemployment, public spending and income inequalities. Concerning alls these variables, we resorted to several databases (see the detail in the appendix; table A.1).

Firstly, to proxy the effect of trade policy on growth, we use a variable of Openness that designates the degree of openness of the French economy and that is given by the volume of trade divided by the GDP, the data being supplied by Asselain and Blancheton (2005). Secondly, the data on inflation come from Pikettys website4 and the OECD databases. Thirdly, data on unemployment being not available all along the studied period, we turn to a variable of total employment measuring the average number of workers, found in Bourlies et al. (2010) and highly correlated with unemployment. Fourthly, we integrate variables of income inequalities able to measure the structure of inequalities and which are said to be more appropriate than a single aggregate index such as Gini to capture the effect of inequalities on growth (Voitchocsky 2005). Based on Pikettys website5, we resort to an variable measuring the level of the national highest incomes that correspond to the highest centile of the declared revenues for the income tax and a variable of top end inequality that represent the national average wage divided by the average level of the highest incomes.
4 5

http://piketty.pse.ens.fr/fichiers/public/Grasset2001/Livre/TabChap1.xls http://piketty.pse.ens.fr/fichiers/public/Grasset2001/Livre/TabChap2.xls

13

To finish, we use variables of public spending that focus on the central State, excluding the expenditures of Social Protection and of local public authorities because they do not directly depend on the government, contrary to the former, and because various kinds of public spending seem to have different effects on growth. For instance, while the role of total public spending is much debated, the empirical studies of the 1980s and 1990s agreed to show that social protection increased growth while the tide turned from the late 1990s (Arjona, Ladaique and Pearson 2002). In this perspective, we propose a variable of public spending that designates the annual consumption on the central State, yearly voted in the parliament and that can be founding the bill on budget. Another variable of government share of GDP represents the share of Public Spending in GDP. All the variables are presented in table A.2 in appendix.

3.4 Empirical strategy Our empirical strategy is based on two batteries of time-series regressions. In the first one, we assess the assumption 1 (H1, figure 1), according to which political ideology affects economic growth, providing by the way a series of robustness check for this relationship and then we test the assumption 3 (H3, figure 1), that is to say the effect of macroeconomic variables that are policy outcomes on growth. In the second battery, we cope with the assumption 2 (H2, figure 1) by testing the effect our ideology index on political outcomes.

In the first battery, we start to test with a 2SLS method6 our baseline equation (1) to check the role of our measure of the annual parliaments ideology on the annual French GDP growth rate by controlling with variables not connected to ideology, the lagged French growth and the European growth (H1). Then we check the robustness of the relationship by a series of additional tests. We make sure that the relationship is valid for the whole studied period and does not depend on specific terms or Republics by adding various period dummies. We also check that the relationship is homogeneous inside the terms and is not related to electoral effects. We complete this first set of regressions by adding to the baseline equation (1) macroeconomic variables (economy openness, inflation, total employment, highest incomes, top end inequality, public spending, government share of
6

We do not resort to a 3SLS method because we do not deal with a system of equations since the observation periods of both batteries of tests do not match exactly.

14

GDP) that stand for the potential transmission mechanisms through which ideology affects growth. To identify them as transmission channels, we check their effect on the dependant variable, the French growth (H3) and weather their inclusion in the baseline regression affects or not the role of the ideology variable on growth. The purpose of the second set of regressions is to confirm the transmission hypotheses, by checking the effect of ideology on the variables identified in the previous step as potential transmission variables (H2). That enables us to distinguish the policies of the left- and rightwing parliaments and thus know through which transmission channels political ideology affect growth.

However by using annual data, a concern arises from the fact that ideology is measured across the same period as growth. We are thus potentially exposed to the identification issue - that is to say ideology and growth may be endogenously determined. However the connected literature tends to discard and sometimes to say nothing about the issue by paying no attention to the possibility that the voters take the government s economic outcomes into account to cast their ballot, influencing by the way the composition of the parliament. In short, economic growth can in turn work on our ideology variable. However, the empirical literature on the effects of ideology has not given much credence to the economic voting assumption, often arguing that the vote mostly reflects the voters' irrationality (Patty and Weber 2007; Wolfers 2007). This assumption seems to be confirmed by the works on France. Franck (2010) supplies empirical evidence showing that the victories of the leftwing coalitions in the first five elections of the French democracy is not due to the national economic growth but rather to "luck", in other words the income shocks at the local level caused by rainfall (Franck, 2011p.22).

In short, the literature does not provide appropriate instrumental variables for ideology to treat the endogeneity issue, apart from scarce exceptions like Bjornskov and Potrafke (2010), who used a variable of government employment (to study the effect of political ideology on economic freedom). We personally use a variable of Working Time (average number of hours worked per year, per worker) from Bourlies et al. (2010). As the laws decreasing working time were systematically voted by leftwing majorities in parliament, we can expect our working time variable to be lower during the leftwing terms while economic growth depend more on productivity than to the number of hours worked. To test the 15

assumption of endogeneity, we perform a Hausman test as follows. We first regress Parliament Ideology on our instrument and the set of exogenous independent variables of our baseline regression. Then, after having checked the significance of our instruments (p=0.0001), we include the predicted values in the baseline growth equation along with the ideology variable and the other independent variables. As the predicted values are insignificant (p=0.88), the endogeneity assumption can be rejected.

4. Results

4.1 Does political ideology indeed affect economic growth? 4.1.1 Estimation of the baseline regression Before any test, it appeared that the French economic growth was shown to be higher under the legislature with a right-wing majority during the period of 1871 2009. This first stylised fact provides us with a first indication concerning the effect of political ideology on economic performance. But, we need to empirically test this relationship to make sure of the robustness of this first hint. Following our empirical strategy, we first provide one battery of regressions to evaluate the role of parliament ideology on the French growth (H1).

The results of this first battery of tests are given by the below table 3. The all ten regressions are controlled by dummy variables for the (First and Second World) war years and by other year dummies to correct the normality of the residuals. The first column of the table provides the estimation of our baseline regression. As expected and according to Alesina et al. (1997), the French growth of the previous year and European growth have a significant and positive effect on the French growth while the parliament Ideology variable turns out to have a significant and negative influence. A 10% increase in the leftwing seats in the Lower Chamber of the parliament (that is to say 57 seats out of 575 nowadays) would reduce the French growth by around 0,2%.

In column 2, we include the degree of economy openness, supposed to reflect the governments trade policy. It actually has a strong positive impact on the national growth and keeps and even strengthens the effect of ideology in the explanation of growth. That first is in accordance with the voluminous empirical literature suggesting that trade has a 16

quantitatively large and robust positive effect on income (Frankel and Romer 1999) and that trade liberalisation has a positive effect on growth in the long run (Foster 2008, p.545; Sala-IMartin et al. 2003; Sachs and Warner 2000; Wacziarg and Welch 2003). Secondly we can deduce that there is no left/wing ideological divide over trade policy. The former does not seem to be a transmission channel throughout which political ideology traditionally works on growth in France, contrary to evidence provided by the empirical literature on U.S. (Dutt and Mitra 2005). Therefore we keep the variable in our baseline regression. Table 2: Results7
Dependant Variable French Growth (t-1) European Growth Parliament Ideology Economy Openness Inflation Public Spending Government Share of GDP Total Employment Top End Inequalities Highest Incomes Observation 138 Adjusted R-squared 0.955 F-statistic 80.27 1 0.06** (2.02) 2 0.10** (2.12) 3 0.10** (2.03) 4 5 6 7 Annual French GDP growth rate 0.10** (2.02) 0.06 (1.45) 0.84*** (5.97) -1.64 (1.28) 0.39*** (4.08) 0.09*** (4.63) 0.81*** (9.89) -0.48 (0.58) 0.13** (2.52) 8 0.24** (2.49) -0.13 (0.50) -2.56 (0.89) 0.76*** (3.72) 9 0.13* (1.77) 0.03 (0.20) 1.84 (0.88) 0.81*** (6.76) 10 0.15*** (6.55) 0.43*** (4.32) -1.74 (1.64) 0.39*** (6.35)

0.10* (1.78)

0.99*** 0.85*** 0.85*** 0.85*** 0.84*** (11.49) (8.02) (7.95) (7.83) (7.65) -1.95** (2.17) -2.39** -2.23** -2.24** -2.25** (2.40) (2.12) (2.08) (2.51) 0.24*** 0.25*** 0.25*** 0.24*** (2.91) (2.98) (3.00) (2.95) 0.01 (0.02) -1.22 (0.79)

-15.62*** (5.38) 81.46*** (4.38) 234.58** (2.15) 43.57*** (4.55) 91 0.868 46.70

-9.63*** (5.05) 69.30*** (2.89)

138 0.941 67.68

119 0.945 80.17

119 0.945 76.36

119 0.941 65.46

119 0.901 61.24

119 0.973 126.67

91 0.715 26.09

11.63** (2.57) 91 0.957 90.98

Note: Absolute value of t statistic in brackets; * significant at 10 percent; ** significant at 5 percent; *** significant at 1%.

4.1.2 Robustness check We performed a series of additional tests the robustness of the established relationship between political ideology and economic growth. First we replied the regression of the column 2 by slightly changing our ideology index. We first use an alternative index that takes no longer into account the composition of the incumbent Chamber, but the composition
7

Schooling, Openness and Government share of GDP being the stationary variable of ln(Schooling), Openness and Government share of GDP respectively.

17

after election, that does not modify the relationship. We also measured the political ideology with a dummy variable (0 for a rightwing majority and 1 for leftwing one), as Alesina et al. (1996) and the effect of ideology becomes less significant. It would imply that the percentage of leftwing and rightwing deputies is more important to explain growth than the ideological affiliation of the majority of the Chamber. That confirms the hypothesis according to which the strongest a majority is in the parliament, the more it influences growth, probably because it has the means to govern according to its ideology.

We also checked that this relationship is valid for all the studied period. We controlled the role of the three Republics with dummy variables in our equation and no one of these dummies reaches significancy. We also check that what we consider as the effect of ideology does not stem from an electoral cycle. We mean that the incumbent government, whatever its political ideology, could artificially increase growth before election to be more popular. Therefore we integrate an electoral variable in the baseline equation that takes the value 1 during the election years. This variable turns out to be not significant and the parliaments ideology always keeps its part. More generally, we checked the effect of including dummy variables for each year of the terms and we found that no one was significant, that confirms the absence of electoral cycle.

4.1.3 The potential transmission variables Still in this first battery of regressions, we now study the effect of other macroeconomic variables on growth (H3). Before adding any variables, we perform again the baseline regression (of the column 2) on a shorter period, from 1891 to 2009, since this is the biggest common period on which all our variables are available and we need to compare the coefficient of ideology in the baseline regression and in equations including potential transmission variables. We can notice in column 3 that the relationship is unchanged on this shorter period. In the column 4, the coefficient of inflation is negative but does not reach significancy. However, according to the theory of the inflation costs, inflation has negative effect on growth firstly since it reduces the value of money and leads actors to hold lower real money balances than they would in the absence of inflation (Benabou 1991), secondly since unanticipated inflation reduces the amount of credit that lenders supply (Alchian et Kessel 1977) and thirdly since it modifies the hierarchy of the relative prices and prevent 18

investors from knowing if a prices variation comes from a variation in the demand or if it results from a monetary policy that misadjusts money supply and demand and fourthly since it increases the menus-costs , in other words the time spent in remarking inventory or reprogramming computers (Caplin and Spulber 1987). But, whereas some empirical results do suggest that inflation is harmful, the evidence is not overwhelming (Barro 1998, p.90; Gylfason and Herbertsson 2001). The absence of effect of inflation on growth in the long run in the French case is not necessarily opposite to theory. Indeed, according to Barro (1998, p.95), an increase by 10 percentage points in the annual inflation rate is associated on impact with a decline in the annual growth rate but the level of inflation in France has never been durably higher than 10 percent.

According to the column 5, the central States spending has a negative but not significant effect on growth. But if we use the share of the spending in the total GDP, the negative effect becomes significant (column 6). That is in accordance with a big deal of the empirical literature on public spending (Bjornskov 2005). That also corroborates the theoretical literature, according to which public spending initially promotes growth when it serves to secure individuals rights (police, law, defense). But beyond a certain level, it becomes unproductive (Mueller 2003, p.550) and harmful for growth, that is proved by numerous studies (Tabellini 2005; Gwartney et al. 1998; Bernholz 1986, 2000; Barro 2000, p.27). That is also theoretically justified by the lower productivity of the public sector (Baumols law) and by the theories of the failure of State that explain the governments inability to set the appropriate quantity of public spending (Wolf 1979, 1988; Krueger 1990; Le Grand 1991; Facchini 2000; Evangelopoulos 2007; Boettke, Coyne, and Leeson 2007).

We also found that the share of total public spending (central State, social protection and local public authorities) in the total GDP had an insignificant impact on growth. That can be explained by the fact that, from 1945, the effect of the central States spending is counterbalanced by the ambiguous effect of social protection spending on growth, which stands for the huge majority of the total spending after the war. And the bulk of the social protection is composed of health spending, supposed to improve the workers productivity and therefore the GDP level. Our results tally with Reinhart (1999) who provides evidence showing that longer lives, resulting from health spending, are associated with faster 19

economic growth whereas higher government spending lowers growth for any life expectancy. We can actually notice in the column 6 that the ideology loses significancy. Thus the government share of GDP can be assumed to be a transmission variable.

As for the variable related to the level of employment, we find, according to the Okuns law, that variable of total employment has a significant and positive effect on growth (column 7). On a shorter period that excludes the war periods, we found that the unemployment rate had indeed a significant negative influence on growth. Our employment variable has also as effect to cancel the role of the ideology variable and can consequently be identified as potential transmission variable. Finally our inequality variables have the role predicted by Voitchocsky (2005, p.274): top end inequality appears to have a positive effect on growth while inequality further down the income distribution appears to be inversely related to growth. That suggests that redistributive policies such as progressive taxation and social welfare are likely to facilitate growth through their impact on the bottom of the distribution and to inhibit growth through their impact on the top of the distribution (Voitchocsky 2005, p.290).

Indeed, in our case, data on the lowest income is unfortunately not available but the average level of the highest incomes and the top end inequalities play a significant and positive role on growth (columns 8 and 9). Our results match Padovano and Galli (2001) who empirically showed a negative relationship between growth and effective marginal income tax rates for 23 OECD countries from 1951 1990. As we point that these variables totally remove the role of ideology, we can reasonably assume that the political ideology works on the domestic product by the channel of the top end inequalities by playing on the marginal tax rates. Thus our top end inequality variable, albeit not as complete as an aggregated index, allows to escape the debates in empirical studies on the virtues of inequalities (Person et Tabellini 1994; Alesina and Rodrick 1994; Barro 1999, 2000).

In view of this first battery of tests, we can reasonably conclude that parliaments political ideology matters in economic growth all along the French democratic experience, from 1871 to nowadays (H1). More precisely, we found robust evidence showing that a leftwing majority in the Lower Chamber of the parliament would be noticeably less favourable to 20

growth and, more than the majority, the size of the majority would account for growth level. From the significant effect of unemployment, income inequalities and government share of GDP, we now know which macroeconomic variables have influenced the French growth (H3). In the second battery of regression, we make sure that political ideology does affect these variables (H2).

4.2 How does political ideology affect economic growth?

We know check the effect of political ideology on the macroeconomic variables to know if rightwing and leftwing parliaments led to different political outcomes (H2) that influenced economic performance. The results of this second battery of regression are provided in the table 3. The column 1 strengthens the idea that inflation has not been a transmission channel during the studied period in France. By controlling the effect of the imported inflation on the level of domestic inflation with the oil prices (data provided on the website: http://www.ioga.com/Special/crudeoil_Hist.htp) and the openness of the national economy, we found no relationship between political ideology and inflation. That is in contradiction with the main predictions of the partisan theory according to which, under leftwing governments, unemployment is lower and inflation higher (Hibbs 1977) or, in other words, leftwing parties attach a higher weight to unemployment relative to inflation (Alesina 1987).

On the contrary central States spending appears to depend on the ideology of the parliament. In column 2, we can notice the significant and positive effect of parliaments ideology variable on the public spending variable when we control by inflation and a variable of total population proposed by Maddisons website. That would vindicate the thesis according to which the parliaments ideology works on economic growth more through the State spending than the social protection and public local authorities spending. Indeed the social protection spending steadily evolved, whatever the government in power, after the creation of the French Social Security system in the wake of the Second World War. Therefore, while a consensus emerged on the necessity of the social protection among left and rightwing governments after 1945, an ideological divide on the central States spending would have always existed since 1871 according to our findings and would supply an

21

additional explanation of the different outcomes of left- and rightwing parliaments in terms of growth.

Table 3: Results the transmission channels

1 Dependant variables Parliament Ideology Oil Price Economy Openness Inflation Total Population Labour Regulation Capital Tax Inflation -0.04 (1.50) 0.0004 (1.04) 0.003*** (2.93) 0.87*** (15.70) 6.46*** (4.20) 0.005*** (2.66) -0.002 (0.25) -0.20*** (3.16) 2 Public Spending 0.04** (2.16) 3 Total Employment -0.013*** (3.18) 4 Highest Incomes -0.07** (2.37)

0.60** Tax (2.50) Observations 103 96 108 78 Adjusted R-squared 0.960 0.998 0.555 0.499 Note: Absolute value of t statistic in brackets; * significant at 10 percent; ** significant at 5 percent; *** significant at 1%.

In the column 3, there is a clear negative and significant effect the leftwing parliament affiliation on the employment level. This effect can be explained by the empirical study of Botero et al. (2004, p.1365) who found for 85 counties during 70 years that the political power of the left is associated with more stringent labour regulations and that heavier regulation of labour is associated with higher unemployment. Siebert (1997, 2003, Chapter 4, p.125) also argues that the regulation of the labour market rigidifies it and thus hinders employment creation. And the French leftwing governments have tended to more regulate the labour market since the beginning of the Third Republic. In fact, six out of eight legislatures that decreased working time were leftwing. And one of the two rightwing chambers that legislated on working time did it during under a leftwing republican government in 1874. The labour market regulations implemented by leftwing governments would have protected the insiders but made the entrance of the outsiders more difficult. Therefore we can confirm that employment policies and especially working time regulations

22

are part of the transmission mechanisms through which political ideology influences economic growth.

To finish, we regress the ideology variable and other variables related to total and capital tax levels (Pikettywebsite) on the variable of top-end inequalities (column 4). Leftwing ideology turns out to have a significant negative impact on the highest incomes and thus on the top end inequalities. Whereas we do not know the impact of ideology on an aggregated inequality index, this result suggests that ideology influences the structure of inequalities by playing on the highest incomes by means of redistribution policies. We can thus reasonably deduce from the two steps of tests that leftwing ideology contributes to slow down growth through redistribution policies disadvantageous for the highest incomes.

In summary, the empirical evidence suggests that differences in parliaments political ideology led to differences in policy-making and in policy outcomes. In almost one and half century of the French democratic experience, economic growth had been lower under a leftwing legislative power. That can be explained by the different policies voted by the left and rightwing parliaments. Indeed the leftwing majorities had tended to promote working time regulation therefore hindering the employment creation, foster redistribution policies to reduce the top end inequalities and finally increase the central States spending. All these measures turned out to be negative for economic growth. But a consensus on the total public spending and especially the social protection and on the issue of inflation appeared between the left- and rightwing governments at different periods.

5. Conclusion The paper wondered about the influence of the political ideology of the legislative power on economic growth all along the French democratic experience. Some of the findings were expectable and confirms the existing literature while other results are more surprising and require additional studies for a better understanding. Even though we totally share the caution expressed by Bjornskov (2005, 2008) when he presented empirical evidence that left was less favorable to growth, the present long-time analysis on France confirms the role of ideology as full determinant of growth and the inferior performance of leftwing majorities in terms of growth. Even if this paper remains one of the first empirical attempts to tackle a 23

highly controversial issue, using a time-series method makes us relatively confident in our findings. First that considerably increases the observation period and secondly that hinders the problems of comparability between the various national economic and political contexts of a cross-sectional series and to take into account the institutional singularities of one country.

This paper is also different from the connected literature on the issue of the emphasized transmission mechanisms between ideology and growth. Bjornskov (2005) insists on the prominent role of the institutional canal and of the public spending. Without denying the effect of institutions on growth in the long-run for France, that is difficult to study in the absence of appropriate data on institutional quality, it appeared that leftwing ideology had traditionally implemented economic policies to reduce top end inequalities, to increase public spending and especially central States spending and labor regulations for the protection of workers but in fine unfavorable to total employment. The empirical results also put into question the assertion, often sustained by the historians, according to which the lefty governments went for more inflationist policies rightwing ones. By the way it also distinguishes itself from the whole literature that maintains that the left makes inflation and the right unemployment. It emphasizes either a French specificity or merely the fact that this opposition in policiy-making is historically situated in the time and is not valid on more than one century.

Besides this contribution to the literature on the determinants of growth, it provides a new way to tackle another controversial issue of the responsibility of the political decisions in the economic decisions of a country. It opposes the thesis that governments cannot work on national macroeconomic variables (Wagner 1977) and the purely opportunist behavior of politicians of the rational business cycle theory. It redeems general interest in policy-making through the normative dimension of ideology (North 1990, p.23, note 23)8. Indeed the

By ideology I mean the subjective perceptions (models, theories) all people possess to explain the world around them. Whether at the micro-level of individual relationships or at the macro-level of organized ideologies providing integrated explanations of the past and present, such as communism or religions, the theories individuals construct are coloured by normative views of how the world should be organized (North, 1990, 23, note 23).

24

political action is not a mere matter of redistribution or of expressive behavior or rhetoric but does modify the citizens lives.

6. Bibliography Alchian, A. and Kessel, R.A. (1977, 1959) Redistribution of Wealth Through Inflation, reprinted in Armen Alchian, Economic Forces at Work, Indianapolis: Liberty Press. Alesina A. (1987) Macroeconomic Policy in a Two-Party System as a Repeated Game. Quaterly Jounral of Economics, 102, pp. 651-78 Alesina A., Roubini N. and G. Cohen (1997) Political cycles and the Macroeconomy MIT Press, Cambridge Massachusetts, London. Andr, C. et Delorme, R. (1978). Lvolution sculaire des dpenses publiques en France , Annales, Histoire, Sciences Sociales, 33 anne, n2, 255 278. Arjona R., Ladaique M. and Pearson M. (2002) Protection sociale et croissance, Revue conomique de lOCDE, numro 35, 2 Asselain, J.-C. (1974b). Histoire conomique de la France du XVIII sicle nos jours. 2. De 1919 la fin des annes 1970, Histoire, Point Seuil. Asselain, J.-.C. (1984a). Histoire conomique de la France du XVIII sicle nos jours. 1. De l'ancien Rgime la premire Guerre Mondiale, Histoire, Point Seuil. Asselain, J.C. et Blancheton, B. (2005). Dynamique de louverture internationale. Paradoxes, enjeux, lments dinterprtation partir du cas de la France , Economies et Socits, Srie histoire conomique quantitative , AF, n32 : 49 179. Barro R. (1999), Inequality, growth and investment , NBER WP 70038. Barro R.J. (2000, p.27), Les facteurs de la croissance conomiques. Une analyse transversale par pays. Economica, traduction franaise de Richard Layard de Determinants of Economic Growth. A Cross-Country Empirical Study, MIT Presss edition, Cambridge, Massachussetts. Barro R.J. (2000), Inequality and Growth in a Panel of Countries, Journal of Economic Growth, 5. Barro R. and Ursua J. (2008). Macroecomomic Crises since 1870. Brookimgs Papers on Economic Activity. Beck, T. Clarke, G., Groff A., Keefer P. and Walsh P. (2001). New tools in comparative political economy: The database of political institutions. World Bank Economic Review, 15(1), 165176. 25

Benabou, R. (1991). Comment on The Welfare Cost of Moderate Inflations, Journal of Money and Credit and Banking, 23, 504 513. Bernholz P. (1986), Growth of government, economic growth and individual freedom, Journal of Institutional and Theoretical Economics, 142: pp.661 683. Bernholz P. (2000, p.10), Democracy and capitalism: Are they compatible in the long run?, Journal of Evolutionary Economics, 10: pp.3 16. Bjornskov, C. (2005). Does Ideology Affect Economic Growth? Public Choice, 123: 133 146. Bjornskov, C. (2008). The Growth-Inequality Association: Government Ideology Matters , Journal of Development Economics, 87, 300 308. Bjornskov C. and N. Potrafke, Forthcoming (2010) Political Ideology and Economic Freedom across Canadian Provinces Eastern Economic Journal. Boettke, P.J., Coyne, C.J. and Leeson, P.T. (2007). Saving government failure theory from itself: recasting political economy from an Austrian perspective, Constitutional Political Economy, 18, 127 143. Bourlies, R., Cette, G., Lopez, J., Mairesse, J. and Nicoletti, G. (2010). Do Product Market Regulations in Upstream Sectors Curb Productivity Growth ? Panel Data Evidence for OECD Countries, NBER Working Paper Series, 16520, http://wwwnber.org/papers/w16520. Budge, I., Klingenmann, H.-D., Volkens, A., Tannenbaum, E., Bara, J. (2001). Mapping policy preferences: Estimates for parties, electors and governments 1945 1998, Oxford University Press, Oxford. Caplin, A.S. and Spubler, D. (1987). Menu Cost and the neutrality of Money, Quarterly Journal of Economics, 102, (4), 703 725. Caron, F. (1981). Histoire conomique de la France, XIX - XX sicle, Paris, A. Colin. Carry, A. (1999). Le compte satellite rtrospectif de lducation en France (1820 1996) , Economies et Socits, Srie AF, n25, fvrier mars. Dubois, E. (2005). Economie politique et prvision conjoncturelle.Construction dun modle macroconomtrique avec prise en compte des facteurs politiques. Thse pour le Doctorat, Universit de Paris 1 Panthon Sorbonne. Dutt P. and D. Mitra (2005) Political Ideology and Endogeneous Trade Policiy: An Empirical Investigation, The Review of Economics and Statistics, vol. 87, n. 1, pp. 59-72.

26

Evangelopoulos, P. (2007) Towards a Synthesis of Theories of State Failure, Int. Rev. Econ., 54, 13 34. Fontvielle, L. (1976). Evolution et croissance de lEtat Franais 1815 1969 , Economies et Socits, Srie histoire conomique quantitative , septembre octobre-dcembre. Foster, N. (12008). The Impact of Trade Liberalisation on Economic Growth: Evidence from a Quantile Regression Analysis, Kyklos, vol.61, n4: 543 567. Frankel, J. and Romer, D. (1999). Does trade cause growth?? American Economic Review, 89: 379 399. Foster, N. (12008). The Impact of Trade Liberalisation on Economic Growth: Evidence from a Quantile Regression Analysis, Kyklos, vol.61, n4: 543 567. Franck R. (2010) "Political Consequences of Income Shocks: The Consolidation of Democracy in France", Bar Ilan University working papers Galor O. and Tsiddon D. (1997), Technological Progress, Mobility, and Economic Growth, American Economic Review, June 87, (3), pp.363 382. Goguel F. (1986) La politique des parties sous la IIIme Rpublique. Edution du Seuil. Gwartney J., Holcombe R; and Lawson R. (1998), The Scope of Government and the Wealth of Nations, Cato Journal, vol. 18, n2, pp.163 190. Gylfason T. and Herbertsson T.T. (2001), Does inflation matter for growth?, Japan and the World Economy, 13, pp.405 428. Hibbs, D.A. (1977). Political Parties and Macroeconomic Policy, American Political Science Review, 71, (4), 1467 1487. Imbeau L., F. Petry and M. Lamari (2001) Left-Right Ideology and Government policy : a Meta-Analysis European Journal of Political Research 40: 1-29 IFM (2003). World Economic Outlook, Chapter 4. Kaldor N. (1956). Alternative Theories of Distribution, Review of Economic Studies, 23 (2), pp.94 100. Lee, J. (2000). The Robustness of Okuns Law: Evidence from OECD Countries, Journal of Macroeconomics, 22, 331 356. Levy-Leboyer, M. (1971). La dclration de l'conomie franaise dans la seconde moiti du XIX sicle , Revue d'histoire conomique et sociale, 4. Markussen, S. (2008). How the left prospers from prosperity , European Journal of Political Economy, 24, 329 342. Mayer (1997) 27

Mueller, D.C. (2003). Public Choice III, Cambridge University Press. Nollen S. and Iglarsh H.(1990) Explanations of Protectionism in International Trade Votes Public Choice ???. North, D.C. (1990). Institutions, Institutional Change and Economic Performance, Cambridge University Press. Padovano, F. and Galli, E. (2001). Tax Rates and Economic Growth in the OECD Countries (1950 1990), Economic Inquiry, vol. 39, n1, January, 44 57. Patty J and R. Weber (2007) "Letting the good times roll: a theory of voter inference and experimental evidence" Public Choice, 130, 3-4, 293-310." Persson T. and Tabellini G. (1994), In Inequality Harmful for Growth?, The American Economic Review, 84 (3), pp.600 620. Alesina A. and Rodrik D. (1994), Distributive Politics and Economic Growth, The Quarterly Journal of Economics, 104, pp.465 490. Potrafke, N., (2010) Forthcoming, Political Cycles and Economic Performance in OECD Countries: Empirical Evidence from 1951-2006, Public Choice. Reinhart, V.R. (1999). Death and taxes: their implications for endogenous growth, Economics Letters, 92, 339 345 Rodrik D. and Rodriguez F. (2000), Trade Policy and Economic Growth: A Skeptics Guide to the Cross-National Evidence, in Ben Bernanke and Kenneth Rogoff, eds, NBER Macroeconomics Annual 2000, Cambridge: MIT Press. Sachs J. and Warner A., (1995), Economic reform and the process of global integration, Brookings paper on economic activity, 1, pp.1-118. Sala-I-Martin, X., Doppelhofer, G. and Miller, R.I. (2003). Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach , American Economic Review, vol.94, n1, 813 835. Sala-I-Martin, X., Doppelhofer, G. and Miller, R.I. (2004). Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach , American Economic Review, vol.94, n1, 813 835. Siebert, H. (1997). Labor market rigidities : at the root of unemployment in Europe , Journal of Economic Perspectives, 11, (3), 37 54. Siegfried, A. (1930) Tableau des partis de France, Grasset. Tabellini, G. (2005), The Role of the State in economic Development , Kyklos, vol.58, n2, pp.283 303. 28

Toutain, J.C. (1997). Le produit intrieur brut de la France, 1789 1990, Economies et Socits, Histoire conomique quantitative, Srie HEQ, n1, n11, 5 136. Toutain, J.C. (1987). Le produit intrieur Brut de la France de 1789 1982 , Economies et Socits, Srie Histoire conomique quantitative, numro 15, mai. Voitchovsky, S. (2005). Does the Profile of Income Inequality Matter for Economic Growth?. Distinguising Between the Effects of inequality in different Parts of the Income Distribution, Journal of Economic Growth, 10, 273 296. Wacziarg R. and Welch Karen Horn (2003), Trade Liberalization and Growth: New Evidence, NBER Working Paper Series, December. Wagner R.W. (1977), Economic Manipulation for Political Profit : MAcroeconomic Consequences and Constitutional Implications, Kyklos, 30, pp.395 410. Wolf, C. (1979). A Theory of Nonmarket Failure, Journal of Law and Economics, 22, 107 139. le Grand, J. (1991). The Theory of Government Failure, British Journal of Political Science, 21, October, 423 442. Wolfers J. (2007) "Are Voters Rational? Evidence for Gubernatorial Elections" University of Pennsylvania working paper.

29

7. Appendix Table A.1. Data Sources

1-Parliament Percentage of the leftwing PM in the Lower Chamber of the Parliament, French overseas ideology departments et territories excluded and without taking account of the non-party MP Website of the French National Assembly: http://www.assemblee-nationale.fr/histoire/index.asp: Component 1: distribution of the seats in the in the Lower Chamber of the Parliament (Chamber of Deputies for 1871-1941, 1946-2009 2-French Growth Component 1: 1871-2008 Maddisons website: France GDP, million 1990 International Geary-Khamis dollars France annual GDP growth rate the 3 and 4 Republic and National Assembly for the 5 Republic)according to the different political parties
rd th th

Component 2: National accounts- INSEE (National Institute of Statistics and Economic Studies) : France GDP, 2009 3-European Growth Europe annual GDP growth rate Maddisons website: GDP Total 12 Western Europe (Austria, Belgium, Denmark, Finland, Germany, Component 1: Italy, Netherlands, Norway, Sweden, Switzerland, United-Kingdom), million 1990 International 1871-2008 Component 2: 2009 EUROSTAT: European Union GDP, million 2000 euros Geary-Khamis dollars million 2000 euros

4-Schooling Total domestic education spending Component 1: 1871-1996 Carry (1999): Total domestic education spending, million current euros

Component 2: Data provided by the Ministry of National Education, on the website: media.education.gouv.fr: 1984-2009 Total domestic education spending, million current euros Degree of openness of the French economy, given by the sum of the importations and exportations 5-Openness as a percentage of the GDP Component 1: Asselain and Blancheton (2005): goods importations as a percentage of the (current price) GDP, 1871-2002 Component 2: 1960-2009 6-Inflation Component 1: 18??-1999 Thomas Piketty's website: http://piketty.pse.ens.fr/fichiers/public/Grasset2001/Livre/TabChap1.xls World Bank: the volume of trade as a percentage of the GDP Inflation rate goods exportations as a percentage of the (current price) GDP

Component 2: OECD website

30

1999-2009 71Unemployment rate=PDRE/POPAC (PDRE = Unemployment (BIUT) - in annual average

Unemployment and in 1000, POPAC= Active Population) Component 1: 1884 1985 CEPII long-series bases (MACRO- LONG.XLS) Component 2: 1985-2009 8-Total Employment Average number of workers 1 Component Bourlies, R., Cette, G., Lopez, J., Mairesse, J. and Nicoletti, G. (2010) 9-Income Inequalities Average income of the top end decile 1 Component: 19001910/19201998 http://piketty.pse.ens.fr/fichiers/public/Grasset2001/Livre/TabChap2.xls INSEE website

10- State Size Amount of central States spending, voted by the parliament in the bill on the yearly budget. projets de loi portant rglement dfinitif du budget de lexercice de chaque anne depuis 1871. Larticle 7 de cette loi indique la fois le montant des recettes et des dpenses votes par le 1 Component: parlement. 11- Public Spending Component 1: 1871-1913 Component 2: 1869-1912 (discontinuous), 1920-1938, 1947-1974 Andre and Delorme (1983): total public spending/ (constant price) GDP Levy-Leboyer and Bourguignon (1985, 1990): government consumption Share of the total spending (State, Social Protection and local public authorities) of total GDP

Component 3: National accounts- INSEE (National Institute of Statistics and Economic Studies ) : (Billion constant 1959-2009 12-Tax 1 Component: 1896-2009 Piketty's website Total tax rate euros) public administrations spending/(constant price) GDP

31

13-Capital Tax Capital tax rate 1 Component: 1896-2009 Piketty's website

14-Labour tax Labour tax rate 1 Component: 1896-2009 15-Average Wage 1 Component: 1896-2009 16-Labour regulation 17-Oil price Dummy coded 1 for the legislatures that implemented working time regulations Crude oil price barrel. Real (Constant 2005 dollars) http://www.ioga.com/Special/crudeoil_Hist.htp Pikettys website Average Wage per worker Piketty's website

Table A.2. Data Presentation

Variable French Growth European Growth Parliament Ideology Economy Openness Inflation rate Public Spending (million 1998 constant francs) Government Share of GDP Total Employment Income Inequalities (million 1998 constant francs) Mean 2.40 2..28 0.54 28.96 6.22 Std. Deviation 7.34 3.28 0.21 8.79 11.17 Observations 139 139 139 139 139

573169

598893

119

24.54 20160634

0.12 1955656

139 120

671979

276167

91

32

Figure A.1 French GDP level (1869-2009)

French GDB - million 1990 International Geary-Khamis dollars 1,400,000 1,200,000 1,000,000 800,000 600,000 400,000 200,000 0 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Source: Maddisons website.

Figure A.2 French GDP growth rate (1869-2009)

50.00% French GDP growth rate

40.00%

30.00%

20.00%

10.00%

0.00% 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 -10.00%

-20.00%

Source: Maddisons website.

33

Figure A.3 French GDP growth rate (1891-1985) according to two different sources
60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% 1890 -10.00% -20.00% -30.00% French Growth (CEPII) French Growth (Maddison)

1900

1910

1920

1930

1940

1950

1960

1970

1980

Source: Madidison (2005), CEPII database

Figure A.4 Western Europe and French GDP growth rate (1871-2009)
30.00% 25.00% 20.00% 15.00% 10.00% 5.00% 0.00% 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 -5.00% -10.00% -15.00% Western Europe Growth French Growth

Source: Maddisons website.

34

Figure A.5 Western European and French GDP levels, million 1990 International GearyKhamis dollars (1869-2009)
6,000,000

5,000,000

4,000,000

3,000,000

2,000,000

1,000,000

0 1870

1890 1910 France GDP level

1930

1950

1970 1990 Western Europe GDP level

Source: Maddisons website.

Figure A.6 Western Europe GDP levels, million 1990 International Geary-Khamis dollars (1869-2009)
6,000,000 Western Europe GDP level

5,000,000

4,000,000

3,000,000

2,000,000

1,000,000

0 1870

1890

1910

1930

1950

1970

1990

Source: Maddisons website.

35

Figure A.7 Western Europe GDP growth rate (1871-2009)

10.00% Western Europe Growth

5.00%

0.00% 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

-5.00%

-10.00%

-15.00%

Source: Maddisons website.

Figure A.8 Ln (Total domestic education spending in France), million current euros (18712009)
12.00 LN (Total domestic education spending, France)

10.00

8.00

6.00

4.00

2.00

0.00 1870 -2.00

1890

1910

1930

1950

1970

1990

Source: Carry (1999), Ministry of National education.

36

Figure A.9 Degree of openness of the French economy (1871-2009), percentage of the GDP
50 Degree of openess of the French economy (%) 45 40 35 30 25 20 15 10 5 1870

1890

1910

1930

1950

1970

1990

Source: Asselain and Blancheton (2005), World Bank

Figure A.10 Governments share of total GDP in France (1871-2009), percentage of the GDP
55 50 45 40 35 30 25 20 15 10 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Government share of GDP (%)

Source: Levy-Leboyer and Bourguignon (1985, 1990), Andre and Delorme (1983), INSEE

37