An analysis of the non-economic forces affecting the paris stock market during the late 19th century
- Autores: Miguel Ángel Ortiz Serrano
- Directores de la Tesis: Carlos Santiago Caballero (dir. tes.)
- Lectura: En la Universidad Carlos III de Madrid ( España ) en 2020
- Idioma: español
- Tribunal Calificador de la Tesis: María del Pilar Nogués Marco (presid.), Vincent Lignon (secret.), Gabriel Mesevage (voc.)
- Programa de doctorado: Programa de Doctorado en Historia Económica por la Universidad Carlos III de Madrid; la Universidad de Barcelona y la Universitat de València (Estudi General)
- Materias:
- Enlaces
- Tesis en acceso abierto en: e-Archivo
- Resumen
The present dissertation analyses the performance of the Paris Stock Market in the late 19th century. In particular it focuses on the years 1888-1889 and 1894-1899, and the economic consequences derived from the socio-political turbulences in this historical period. This thesis approaches to the early years of the French 3rd Republic and its multiple challenges. The first chapter provides a descriptive explanation to the French political context during the last three decades of the 19th century, aimed to understand the progressive changes experienced by this regime which led to an extreme social polarisation. The second chapter focuses on the effect of political connections during the Boulangist campaign. The close relationship between politics and business has been largely studied by French historians such as Jean Garrigues , who depicted an extensive analysis on the close links between politicians and businessmen in the late 19th century France. The economic importance of political connections has captured a growing interest among scholars in the last two decades, after Fisman (2001) published his seminal paper on the value of connectedness under Suharto regime in Indonesia. Whereas most of the first studies focused on developing countries, such as Faccio (2007), economic historians have been progressively more attracted by this topic. Applying modern techniques in financial econometrics to historical cases allows us to compare outcomes across time and infer the behaviour of economic agents in both historical periods. Ferguson and Voth (2008) and Braggion & Moore (2013) shed evidence on the importance of political connections to determine firms’ value. The first case shows that the stocks from the companies linked to the Nazi party obtained positive cumulative abnormal returns after the elections of 1933. The second article display the differences in performance between political and non-politically connected companies in the London Stock Exchange after legislative elections. Other authors, however, find little evidence on the effect of political connections, such as Lehman-Hasemeyer & Opitz (2017) and Grossman & Imai (2016).
The Boulangism was a heterogeneous movement that directly affected to the stability of the 3rd Republic, when in 1887 the general Boulanger started a populist campaign against French government. Initially backed by radicals and socialists, with the support of newspapers such as “La Cocarde” or “Le Gaulois”, his movement soon captured the interest of his former adversaries from the far right, who saw on it an opportunity to destabilise the Republic and overthrow the government. The general’s campaign against the “opportunist” republic and corruption soon polarised French society and increased political tensions, which turning point was the by-election of 27 January 1889. This election was interpreted as a plebiscite about the future of the republican regime. After the announcement of Boulanger’s victory, some of his partisans encouraged him to stage a Coup d’état against French government. However, the general hesitated and finally rejected this option, and postponed seizing power after the legislative elections for September 1889. His lack of decision and hesitance made many of his partisans to abandon him. Some weeks after, the general auto-exiled in Brussels to escape arrest.
I find evidence on the importance of political connections the days after the by-election of 1889 for the companies linked to republican parties. I have collected a new dataset from the following primary sources: Firm-specific characteristics were obtained from the Annuaires des Agents de Change, the yearbook of the Compagnie des Agents de Change, the organisation in charge of supervising and organising the economic transactions at the Paris Bourse. These yearbooks can be found at the Archives of Crédit Agricole and at the Archives of the French Ministry of Finance (SAEF). I have compiled price data on a daily basis for 73 liquid stocks of the Paris Bourse, from 1 March 1888 to 16 February 1889. Prices were collected from Gallica, the online repository of the National Library of France. This includes the “Bulletin de la Cote”, the spreadsheet reflecting every daily financial operation published by the Compagnie des Agents de Change. I construct a panel-data logistic regression to observe the effects that political connections had on investors’ attention, finding that being linked to republican parties contributed positively to stocks’ trading frequency. The main goal of the article is to find whether, depending on the type of connection, firms’ performance varied after the elections of January 1889. This is calculated by applying Event-Study techniques through a market model, as in Campbell et al. (1997) and Kothari & Warner (2007). In a preliminary analysis, findings indicate no sectorial effect derived from the elections. The main results, however, show a significant outcome. Firms connected with republican parties obtained 0.6% and 1.5% positive cumulative abnormal returns two and ten days after the by-election of January 1889. Firms conservative boards and non-connected companies have no significant results. Additionally, a group including both conservative and republican connections shows no-effect either. A placebo test in a different date shows no effect for any of the groups. We confirm in a further robustness analysis that republican connections were the only driver of cumulative abnormal returns for stocks two and ten days after the by-election.
The third chapter of this dissertation studies the weeks around the economic collapse of the Panama Company. In particular, we discuss the effect of fake news circulated by French newspapers receiving bribes from Compagnie Universelle du Canal de Panama to support the issue of bonds for the 26 June 1888. This works can be commonly subject to several caveats such as that reputable sources of information can prevent from fake news, or that they may have little effects on overall investors’ decision-making. However, this is not the case of the present chapter, since investors’ at the Paris Stock Exchange in the late 19th century were tightly linked to each other, and information was shared by hearsay and printed media (Ortiz-Serrano, 2018). We expand the database from Chapter 2, to include daily data on prices for four additional months. Thus, the time span covered by the new database goes from 10 November 1887 up to 16 February 1889. We include firm-specific characteristics for the year 1887, such as capital, dividends, or number of shares, from the same primary sources as in Chapter 2. Data on newspapers was collected on a daily basis as well, from 1 March to 1 September 1888, from Gallica. The links are available in the primary sources’ section of the dissertation. The four newspapers are the following and are known to have received abnormal payments the months before the issue: Le Petit Journal, Le Temps, Le Figaro and Le Gaulois. We observe a sharp increase in coverage regarding the Panama Company during the weeks under study, which is in line with the findings of the accounting experts studying the Scandal in 1892-93. To identify the role that news played in the evolution of the Compagnie’s stock price, we perform linear regressions using two standard asset pricing models: The Capital Asset Pricing Model (CAPM) and the Fama-French three factor model (FF). We control for potential “noisy dates”, defined as those events that we are unable to discern between true and fake news. These dates are 1 March 1888, 14 March, 23 April, 28 April, 8 June and 26 June. use both models using data from 1 March 1888 to mark Lesseps’ decision to go along with the issue of obligations à lots until 26 June 1888 when the issue was fixed. We observe in the results provided by both models some interesting outcomes. First, we observe that “noisy dates” variable has a positive significant coefficient, suggesting that for these dates, the returns of the Panama Company were abnormally high compared to other days in the sample. Second, both coverage and the tone of coverage played a role in explaining company’s stock price, although it only remains significant after 23 April. Third, we find that news and tone has a negative effect on company’s returns. Running a Bai & Perron (1998, 2003) structural breaks approach led to a break on 21 April, further warranting studying the effect of news before and after this date. Results remain significant after some robustness checks. For instance, we run a two-stage least squares regression to control from potential omitted market variables. Even after controlling for elements explaining about two-thirds of the variance of the Compagnie’s stocks, coverage remains as an explanatory factor.
Results show that the market accounted for the contents appearing in French newspapers, although interpreting them in a different way. It signals is that positive or additional coverage in the news, particularly after 23 April 1888 was taken as a signal to drive prices down. In our view, this can happen for one of two reasons: First, it is possible that the market believed that the additional coverage was an attempt either by the company or by the newspapers to trick them into buying the new stock issue. Second, it might be that a market manipulation scheme was in place and that those who paid for the additional positive coverage were purchasing the stock before the news came out and selling as the news was revealed to the whole market. This would follow the market adagio of “buying with the rumour and selling with the news”. Other robustness test accounts for the effect that coverage had on dividends’ yields. This is based on the efficient market hypothesis, which states that prices incorporate all available public information. Thus, dividend yields should reflect all of the information that prices incorporate as they evolve and that should be related to our news variables if investors take them to offer true and believable information for them to act upon. We regress the dividend yield on a linear time trend and our noisy dates dummy. Additionally, we include the average coverage and average tone of coverage variables, and an interaction between the coverage and the tone of coverage variables with a dummy taking value one on or after 23 April. Once controlling for the differences in behaviour before and after 23 April, news have no effect before and a have a significant negative effect after that date, confirming our first set of results.
The fourth and the fifth chapters explore the ethnic affiliation of companies listed at the Paris Stock Market. Chapter 4 analyses the economic effects of Antisemitism, in an attempt to capture both short and long run effects during the Dreyfus Affair. The captain Dreyfus was arrested in late 1894 accused of espionage, giving rise to a substantial growth of Antisemitism in France in the subsequent years. Newspapers played a key role in the spread of xenophobia, such as La Croix and La Libre Parole, being the main Catholic and Antisemitic newspapers, respectively. The Affair captured high attention levels since its early days, although its interest decreased substantially after Dreyfus’ degradation in 1895. However, the case regained interest when, in late 1897, the identity of the real culprit was revealed. Most of the proofs indicated that the Commandant Esterhazy was the real spy, and that Dreyfus had been falsely accused. It gave rise to a campaign to rehabilitate Dreyfus, advocating for a reopening of the case, which intensified after the publication of Zola’s article, J’Accuse!, the 13 January 1898. Our study connects naturally with recent historical studies in economics on the phenomenon of antisemitism in Europe, such as those of Voigtländer & Voth (2012) and Grosfeld et al. (2017). However, unlike these studies, which focus on the causes of discrimination and its persistence, we explore the economic outcomes of Antisemitism. This links directly to literature studying the economic effects of xenophobia under the Nazi period, such as Ferguson and Voth (2008), Lindenthal (2017) or Huber et al. (2018). We test for both short and long run effects of Antisemitism, as in Acemoglu et al. (2011) and D’Acunto et al. (2018).
We build a new economic database of 140 liquid stocks for the period 1894-1899. We use similar sources as in the previous chapters, such as the Annuaires des Agents de Change for firm-specific information, or the Bulletin de la Cote for the daily prices for each stock. In addition, to identify the religious/ethnic affiliation of the members of the board, we use the analysis from Cyril Grange (2016) and Paul Levy (1960), to identify the firms in two different ways. However, only Grange’s method represents a valid way to identify firms. This is reasonable, since his work provides an extensive approach to the most prominent French Jewish families for the period 1870-1940. The individuals belonging to these families were commonly known by French society and were among the most powerful Jewish branches in late 19th century France. The second procedure, based on Levy (1960), represents a more generalist approach, including every French surname with Jewish origin. Thus, some individuals could be identified as to be J-connected whereas contemporary public opinion did not perceive them in that manner.
We collect data on five contemporary newspapers that extensively covered the Dreyfus Affair for 1894-98, to obtain a reliable measure of the intensity of the Affair and Antisemitism in France during the period under study. The selected newspapers played an important role in the evolution of the Affair: Le Siècle, Le Petit Journal, La Croix, La Libre Parole and L’Aurore. The attention paid to the Affair was high the weeks around the arrest, trial and deportation of Dreyfus (1894-95). Moreover, we observe an impressive growth of coverage by late 1897, which remained very high during 1898.
The short-term analysis uses standard methodology in financial econometrics, previously described, to calculate the economic effects of the following events: For the years 1894-95, Dreyfus’ arrest, conviction and degradation. For 1897-98, J’Accuse!, Henry’s suicide, and the reopening of the process in 1899. We perform Event-Studies using the CAPM and the three-factor model (Fama-French). Findings indicate a strong negative effect for Jewish-Connected firms after Dreyfus’ degradation. The long-run analysis was performed through a standard Differences-in-Differences estimation. We observe that the publication of J’Accuse! and the subsequent social campaign to rehabilitate the captain had a positive effect on Jewish-connected firms’ returns, which persisted for the whole year 1898. These results are consistent when the same experiment is applied to firms with Jewish CEOs. Additionally, we do not observe a negative effect on Jewish stocks from Antisemitic press. However, generalist newspapers such as Le Siècle could have more influence on French public opinion due to its lower ideological bias. We observe a negative effect from the coverage of the Dreyfus Affair on J-connected returns for the period 1894-96. The sign changes and the effect becomes positive for the same group of firms, for the period 1897-98. Last, finding indicate a positive effect of government changes after the arrival of Brisson in June 1898, which could coincide with a change of social expectations about the Affair and the perception of public opinion regarding French Jews. Thus, results suggest that both in the short and the long-run, negative and positive information on French Jews and Dreyfus may have affected investors’ decision-making at the Paris Stock Exchange.
Chapter 5 of the present dissertation complements Chapter 4, by analysing the relations of the economic agents at the Paris Stock Market during the Dreyfus Affair (1894-98). We attempt to complement descriptive analysis on the Paris Bourse such as Gallais-Hamono & Hautcoeur (2007) or Le Bris (2011) and provide an introductory study to the existing connections between politicians and the members of the most prominent Jewish families in France, as argued by Birnbaum (2014) and Grange (2016). On average, we observe that politically connected firms were more profitable, enjoyed from a much larger trading frequency and had higher market capitalisation. Additionally, through a panel date logistic regression, findings show that Jewish-connected firms were more likely to have political connections. That is, despite the rising Antisemitism in France from the late 1894, members of the most prominent Jewish families may have believed that participating a wide range of business activities could help them to be more socially accepted, at least among the high bourgeoisie. Nevertheless, these results need further research on how these relations affected to firms’ performance in the long run.
In general, this thesis contributes to the existing literature by showing the importance of non-economic factors on investors’ decisions. It demonstrates that French investors based many of their decisions on their political or personal views and were affected by media contents -not always in the expected way, such in the case of the Panama company-. The significance of these events helps to expand the existing research strand on French financial history. The Paris Stock Market was affected by similar factors as modern financial markets. That is, by media coverage, political connections, ethnic bias, discrimination, etc. In addition, it demonstrates that French investors were capable to distinguish between fake and real information, and they classified it to take their decisions. Last, the present dissertation shows how events that were historically considered as social problems, such as the Dreyfus Affair, had significant economic consequences for companies at the Paris Bourse.
