Master Thesis
Universiteit van Amsterdam Department of Economics
The Impact of the Socio-Political Background on Politicians’ Work Effort: Evidence from the European Parliament over the Years of Crisis
Philipp Hafner Student Number: 10603344 Supervisor: Dr. Carmine Guerriero
2
Abstract
Does a politician’s socio-political background have an impact on his work effort? Exploit-ing data on the7th European Parliament (2009-2013), I correlate seven work indicators of Euro-pean Parliamentarians with two variables from their domestic socio-political background. Pre-cisely, I employ a survey capturing European citizens’ attitude towards the European Parliament. A second explanatory variable indicates the democratic qualities prevailing in a Parliamentari-an’s home country. Using a fixed effects regression framework, I find that a decrease in the do-mestic level of trust towards the European Parliament induces Parliamentarians to exert more work effort. The small changes in countries’ democratic quality, on the other hand, had in gen-eral no impact on their work effort. These results are then confirmed by sevgen-eral robustness checks.
3
1. Introduction ... 4
2. Related Literature ... 6
3. Data and Empirical Strategy ... 12
3.1 Measuring the Public Attitude towards the European Parliament ... 13
3.2 Measuring the Quality of Political Institutions ... 16
3.3 Measuring Work Effort ... 18
3.4 Empirical Strategy ... 21
4. Results ... 25
4.1 Results of the Fixed Effects Regression Model ... 25
4.2 Results of the Pooled OLS Regression Model ... 27
5. Robustness Analysis ... 30
5.1 Alternative Measures for Trust ... 30
5.2 Alternative Measures for Political Background ... 33
5.3 Application of Standard Errors Clustered by Countries ... 34
5.4 Country-Fixed Effects Regression Model ... 35
5.5 Inclusion of a Quadratic Term of Trust ... 36
6. Conclusion ... 37
4
1. Introduction
A decline in the public level of trust towards political institutions has become a common global phenomenon in recent years. The importance of high public support for the functioning of political institutions has been pointed out by several scholars (e.g. Gamson, 1968; Putnam, 1993). The present study adds new evidence to this literature by looking at the impact of public level of trust towards the European Parliament (EP) and its members’ work effort. The work effort of a Member of European Parliament (MEP) is operationalized by seven inner-parliamentary work indicators (see section 3.3). While previous research is focused on national Parliaments, the assessment of the European case is novel and topical.
First, the increase of skepticism towards the European Union (EU) and its institutions has been substantial over the investigated period of time. Symptomatic is the decline in voter turnout for the elections of the EP, reaching an all-time low in 2009 of 43%.1
Second, this wave of skepticism is in contrast with the evolution of the political power in Europe. Whether European citizens like it or not, the EP is gradually gaining in influence. In the very beginning (1979), the EP served only as a discussion forum that consulted the decision-making powers of the European Council and Commission. Nowadays, the EP has evolved into an emancipated co-decider who – together with the Council – deals with three quarters of the European legislation process.2 Especially since the implementation of the Treaty of Lisbon on December 1st, 2009 many formerly national competences of Parliaments have been ceded to the EP (for an overview and a critical evaluation of the treaty see Hofmann, 2009). The calls for an even higher political integration within the EU in order to overcome the ongoing crisis are be-coming louder (Guiso et al., 2013). Therefore, the Treaty of Lisbon could mark only the begin-ning of a much profounder shift of responsibilities towards the EP.
Third, the tendency of dwindling institutional approval is intensified in consequence of te-nacious economic turbulences which keep Europe in suspense. During the last couple of years, the controversy over the right policy measures has increased the tension among Europeans. The debate was oftentimes characterized by polemic and ignorance. Moreover, scholars have not exempted the EU and its institutions from criticism (Follesdal and Hix, 2006). They point to the
1
This figure remained nearly the same for the recent elections in May 2014:
http://www.results-elections2014.eu/en/election-results-2014.html, accessed on June 15th, 2014.
2 The European Parliament provides information on its responsibilities in great detail:
http://www.europarl.europa.eu/aboutparliament/en/007c895f4c/Powers-and-procedures.html, accessed on June
5
lack of transparency and democratic legitimization.3 Any further evidence that helps to get a better understanding of a European institution and abolishes prevailing half-truths is therefore worth pursuing.
Finally, the environment surrounding MEPs is peculiar. In general, politicians have diffi-culties with gaining reputation among their voters. MEPs suffer from this phenomenon presum-ably to an even greater extent. The fact that their workplace4 is typically far away from their home constituencies raises suspicion. In addition, there is a considerable share of European citi-zens who are still unaware of the EP’s responsibilities.
All in all, why and how should a MEP adjust his work effort due to changes in the domes-tic level of trust? As the upcoming section illustrates, polidomes-ticians are in general susceptible to incentives. The survey employed in the present study captures a country’s periodical attitude towards the EP. From a MEP’s point of view, it indicates the degree of support and satisfaction of the voters he is responsible to. Assuming the constant exchange with his home constituency (see section 3.3), a MEP is likely to adapt his work effort in consequence of changes in the do-mestic level of trust.
Based on different theoretical foundations, there are two possible explanations concerning this adaption of a MEP’s work effort. On the one hand, the employed survey among European citizens can be interpreted as a country’s degree of “cultural acceptance” (see section 2). In this context, the survey symbolizes a country’s capability of holding its MEPs accountable to high work standards [Nannicini et al. 2013, p.223]. An increase in cultural acceptance therefore sug-gests an increase of a MEP’s work effort. On the other hand, the literature provides also a basis that argues the opposite. For example, the concept of “political capital” (Bourdieu, 1981) sug-gests that a lower level of trust induces a MEP to exert more work effort. Precisely, little support implies that domestic voters are not pleased with the political direction taken by the EP. This, in turn, induces the MEPs from such country to increase their work effort. These MEPs might for example try to reshape or even repeal existing laws in order to reestablish trust among their skeptical voters at home. In order to provide a holistic framework of a MEP’s socio-political background, the institutional quality prevailing in his home country serves as a second inde-pendent variable.
By focusing on a novel panel data set spanning from July 2009 to December 2013, I ex-ploit the exogenous economic crisis shock on the level of trust towards the EU in general, and
3 The lack of democratic legitimization is less true for the EP. In fact, it is the only EU institution that is directly
elected by its member states’ citizens and therefore represents the most prominent democratic element of the EU [Hurrelmann and DeBardeleben 2010, p.230-231].
4
6
the EP in specific. This particular period of time approximately captures the beginning of the crisis and its immediate aftermath.5 Furthermore, the use of a fixed effects regression model fa-cilitates the focus on the impact the socio-political background had on a MEP’s work effort.
The key findings of the study are the following. First, there is a relevant and significant negative effect of a country’s level of trust towards the EP on its MEPs’ work effort, i.e. MEPs from low supportive countries exert more work effort. Second, the small changes in the quality of political institutions had in general no impact on their work effort.
The rest of the paper is organized as follows. The next section provides an overview of the relevant literature and a theoretical framework. Section 3 explains the approach of how to meas-ure the single variables as well as the empirical strategy. Section 4 contains the main empirical results. Robustness checks are the subject of section 5. Finally, section 6 concludes.
2. Related Literature
The screening of politicians is a recent but varied research area. The strand of literature which aims to infer desirable incentives from the investigation of politicians’ behavior is the most relevant for the purposes of this study.
Lott and Bronars (1993) examine U.S. House of Representatives Members’ attendance at legislative voting sessions for a time period of 15 years. They find no evidence of shirking members who are in the final term of their parliamentarian membership. This finding contradicts theories based on the assumption of opportunistic politicians.6 According to this assumption, “the level of shirking should be greatest when a politician decides to leave office” [Lott and Bronars 1993, p.127]. Parker and Powers (2002) come to the conclusion that members of the American Congress, who are at the end of their congressional career, tend to engage more in so-called “foreign junkets”, i.e. trips which are assumed to be unrelated to their actual tasks.
Bernecker (2014) shows that members of the German Parliament, who are facing close electoral competition in their district and whose reelection is therefore precarious, have a lower absence rate in parliamentary voting sessions. In Italy, politicians who enter the House of Repre-sentatives via majority vote show lower absenteeism rates than those who enter via proportional vote (Gagliarducci et al., 2011). Also Nannicini et al. (2013) point out that the electoral
5 The collapse of the American investment bank Lehman Brothers in September 2008 is generally considered the
start of the subsequent global economic recession. The first symptoms of grave troubles in the EU became visible in 2010, when Greece requested officially a bailout. Portugal followed in 2011. Apart from rapidly rising levels of public deficit and debt, these countries were – and partially continue to be – plagued by low rates of economic growth and high unemployment. For reasons of simplicity, I refer to this complex economic crisis as “Euro crisis” in the following.
6 This is a common and intuitive assumption made by researchers (e.g. Besley, 2004; Barro, 1973). It also underlies
7
ground incentivizes Italian Parliamentarians in different ways. The present analysis exclusively examines MEPs’ work effort during the seventh term of the EP. A distinction to account for dif-ferences in proportional and majority systems is therefore redundant.7
Gagliarducci et al. (2010) highlight the tradeoff that countries face when allowing their politicians to keep high-profile jobs outside Parliament. Such moonlighting politicians tend on the one hand to be more talented and therefore more preferable to run for office. On the other hand, they devote less time to parliamentary tasks due to their dual burden. This is confirmed by empirical data consisting of Italian Parliamentarians from 1996 to 2006. According to the au-thors, it would be difficult to make these politicians focus exclusively on their work as Parlia-mentarians. However, the prohibition of moonlighting would lead to adverse selection in a sense that the most talented politicians no longer run for office. Arnold (2013) studies the effect of German Parliamentarians’ outside earnings on their activities in Parliament. He finds strong evi-dence that higher outside earnings cause Parliamentarians to exert less work effort. The author makes use of the fact that German politicians are obliged to reveal their paid activities outside of Parliament.
Not too much work has been done on MEPs so far. The harmonization of MEPs’ salary in July 2009 prompted Mocan and Altindag (2013) to investigate the subsequent change in MEPs’ work effort. MEPs who experienced an increase in salary due to the reform attended slightly less sessions. The implied labor supply elasticity of -0.04 is comparable to those of conventional jobs. The authors apply several other indicators (e.g. speeches, reports and inquiries) that define a MEP’s parliamentary work effort. This approach delivers a holistic evaluation of a MEP’s work effort, whereas the attendance rate by itself is of little informative value. Fisman et al. (2013) investigate the same event but find no significant correlation.
The literature reviewed so far mainly analyzes the effects of variables immediately linked to a politician’s work effort such as salaries, (re-)election issues or the stage of a Parliamentari-an’s political career. The focus of this analysis is however to examine the interaction of a MEP’s work effort and the socio-political background prevailing in his home country. This background is referring to a country’s level of trust towards the EP and the quality of its political institutions. In this field of research, Guiso et al. (2013) highlight that cultural contrasts among coun-tries in economic unions like the EU are a major obstacle in coming to agreements on efficient policy reforms. They argue that such dilemmas can only be overcome by a fiscal union that
7
As a matter of fact, the involved European countries are restricted to the use of a proportional voting system. The individual national designs of this voting system can still differ. In the 2009 EP elections countries typically chose to vote on national parties (closed-list system) instead of single candidates (open-list system) [Fisman et al. 2013, p.7]. Hence, MEPs are approximately homogeneous in this regard.
8
eliminates the “clashes” countries like Germany and Greece are facing right now due to their heterogeneity in culture.
Closest to the context I explore is the study by Nannicini et al. (2013). The authors take a closer look at the cultural attitude of Italian electoral districts and the corresponding politicians’ behavior. They provide a theoretical model that implies an increase in a politician’s misbehavior as the amount of “social capital” in a given district decreases. A politician’s misbehavior is as-sessed according to his absenteeism rates in Parliament and judicial requests for criminal inves-tigations against him. The amount of a district’s social capital is determined by blood donations. The intuition is plausible: high numbers of blood donations imply more civic voters. Civic vot-ers are more altruistic and base their vote on aggregate welfare instead of individual benefits. The act of blood donation is therefore used as an indicator of how likely people are to hold poli-ticians accountable for their actions. The model predicts that the more social capital a district has, the less likely a politician is to misbehave. The politician is intrinsically disciplined by the threat of being punished by civic voters in case of misbehavior.
Applying data from the Italian House of Parliament, the model’s predictions are con-firmed. The authors find that politicians who were elected in a district with lower fractions of civic voters (i.e. a district with relatively little social capital) are more often absent from Parlia-ment and are more often object of requests for criminal investigations. In other words, this ob-servation suggests that the more civic a district is, the more likely a politician is to be held ac-countable in case of misbehavior.
The use of social capital is in this context a measure of trust within districts, i.e. a form of interpersonal trust. Instead of examining trust among one another, I look at the level of trust to-wards the political institution itself. To specify, the survey employed in the present study indi-cates European citizens’ acceptance and support towards the EP. In order to account for this modification, I redefine the term social capital as “cultural acceptance”. In the style of Nannicini et al.’s theoretical concept (2013), the degree of cultural acceptance is then interpretable as a proxy for voters’ capability of holding MEPs accountable to high work standards. Accordingly, an increase in cultural acceptance disciplines a MEP and induces him to exert more work effort. The correspondent hypothesis reads as follows:
The higher a country’s level of trust towards the EP is the higher is the work effort of
a MEP from this country.
In general, trust has many times been proven to be a potent sign of populations’ social norms and its explanatory power has been discovered in various circumstances. Arrow (1972)
9
emphasizes the far-reaching importance of trust in economic transactions: “Virtually every commercial transaction has within itself an element of trust, certainly any transaction conducted over a period of time. It can be plausibly argued that much of the economic backwardness in the world can be explained by the lack of mutual confidence” [p.357].
High levels of mutual trust therefore reduce the need for business partners to costly moni-tor each others’ contractual compliance (Knack and Keefer, 1997). Tabellini (2010) studies the effect of culture on economic development in Europe. Measures of trust and mutual respect are used to define a region’s level of culture. The threat of reverse causality complicates this analy-sis. The author successfully isolates the exogenous component of culture by using historical data on European regions’ literacy rate and political institutions. The data set on 69 regions in Europe suggests a positive correlation between this exogenous part of culture and current economic de-velopment measured in per capita output. Knack and Keefer (1997) report a significant influence of trust on a country’s economic performance measured in GDP growth rate. The authors make use of survey indicators that are similar to those employed in the present analysis after having carefully discussed the basic problems associated with the measurement of trust and social norms (see section 3.1 for an extended review of this discussion).
There is no predetermined theoretical guideline that directly addresses the relation investi-gated in the present study. Apart from the above described vague analogy to the concept of Nannicini et al. (2013), one could frame the situation at hand also as a Principal-Agent problem. Even though this theory has been expanded to the relation of politicians (agents) and voters (principals), it is used in contexts deviating from what I aim to explore.
Ferejohn (1986) models the clash of interests between a politician (agent) and a voter (principal) in order to find an optimal performance-oriented voting rule. This model is based on the assumptions that a politician is self-interested and his degree of work effort can only be ob-served by himself. Accordingly, a politician decides on the minimum amount of work effort to invest in order to stay in office. For the purpose of the present study it is mainly the theoretical foundation, i.e. the assumption of opportunistic politicians, which is relevant.8
The second strong and common assumption underlying Ferejohn’s (1986) Principal-Agent framework refers to the degree of information asymmetry between an agent and his principal. The assumption that a politician’s degree of work effort can only be observed by himself is de-batable for the present analysis. The degree of transparency regarding a MEP’s work – at least in
8 Subsequent to this, the author argues that the principal would naturally prefer an agent who exerts as much work
effort as possible. In order to reduce the information asymmetry, the principal would have to bear agency costs. Therefore, a voter prefers an election system that mitigates this dilemma in the first place and “maximizes her well-being subject to the constraint that politicians are pursuing their self-interest” [Ferejohn 1986, p.8].
10
Parliament – has clearly increased.9 Still, the Principal-Agent concept based on the assumption of self-interested politicians can serve as a solid framework once slight modifications are im-plemented. Most notably, an extension to account for the level of trust between the involved parties is required.
Gamson (1968) established a theoretical framework to investigate the impact of public trust on governmental success. One of the key implications of this work suggests that trust is the basis for a government to launch effective programs which in turn guarantees successful govern-ance. Since then many authors have confirmed this view (e.g. Putnam, 1993; Dalton, 2004).
Batto (2005), on the other hand, derives the opposite hypothesis by embedding a country’s level of trust towards its government into the Principal-Agent framework. He conjectures that a high level of trust in governments is eventually abused by incumbent governments because trust “indicates a greater willingness to delegate power without imposing controls on that power” [p.1]. This intuition is then confirmed by examining data from Asian countries.
Substituting government for a single politician, the following analogy is implied: MEPs from relatively trustful countries are more likely to exert less work effort since they abuse the people’s trust in their work. A trusting voter could be interpreted as someone who doesn’t be-lieve that a MEP would act opportunistically at her expenses (Knacker and Keefer, 1997). These voters’ trust in a MEP’s work is relatively high and therefore they monitor him less strictly. Also their willingness to bear agency costs and punish misbehaving politicians in elections is lower. The MEP is aware of this fact and as a consequence exerts less work effort. In summary, the following hypothesis can be derived under this modified Principal-Agent framework:
The higher a country’s level of trust towards the EP is the lower is the work effort of a
MEP from this country.
Bearing in mind the mixed evidence in favor of the opportunism assumption (Lott, 1993), a second explanation, that argues the same direction yet treats the problem from a different an-gle, would be the following: MEPs from countries where the level of trust towards the EP is rel-atively low work harder in order to convince the voters they are responsible to and regain their trust. MEPs from trustful countries, on the other hand, have a lower incentive to exert high work effort. This reasoning is based upon the concept of “political capital” introduced by Bourdieu (1981). To elaborate, political capital symbolizes a politician’s political power that is “derived from the trust that a group of followers places in him” [Schugurensky 2000, p.420]. The amount of political capital is highly responsive to changes in the followers’ level of trust. A politician,
9
11
who experienced a drop in voters’ trust, increases his work effort to reestablish political capital and thereby political power.
In an interesting paper, Arnold et al. (2012) point to the determinants of trust towards EU institutions. Using data contained in the “Eurobarometer” (see section 3.1) for the years 2005-2010, their main estimates suggest that a high level of trust in national institutions stimulates trust towards the EU institutions. At the individual level, trust is positively affected by the per-sonal socio-economic situation as well as the perceived benefits gained through the EP. This last insight addresses a key concern of the present analysis. The fact that the level of trust towards the EP is partially shaped by personal benefits gained through the EP indicates that the relation between a MEP’s work effort and trust might be plagued by reverse causality. In sections 3.1 and 3.4 I discuss in further detail how this type of endogeneity is linked to the present analysis and how to deal with it.
A second study by Roth et al. (2011) explicitly deals with the impact of the Euro crisis on citizens’ level of trust towards European institutions. Employing a fixed effects regression mod-el, they correlate four macroeconomic indicators (inflation, unemployment rate, GDP growth and public debt) with trust indicators towards European institutions. The trust indicators (e.g.
Trust, see section 3.1) are again obtained from the Eurobarometer for pre-and post-crisis years
(1999-2010). The results indicate that only high rates of unemployment are significantly associ-ated with a decline in European citizens’ level of trust towards the EP during the years of crisis. MEPs’political background is a further potential explanation for his work effort. Previous research mainly focuses on the macro-economic effects of political institutions. Again many researchers have linked political institutions to a country’s economic development. The evidence for the relevance of democratic institutions in this regard is mixed (for an overview see Helpman 2004, chapter 7). Some studies conclude that solid democratic structures and the expectation of future institutional stability have at least favored faster economic growth from 1850 onwards (Persson and Tabellini, 2010). Others report that institutions have had no significant historical impact on economic development (e.g. Guerriero, 2014; Glaeser et al., 2004).
According to Glaeser et al. (2004), it is rather the amount of human and social capital that drives a country’s economic performance. The authors elaborate also on the methodological concerns that complicate the detection of causal relationship between institutions and economic development. Apart from the challenge of finding valid instrumental variables to control for po-tential reverse causality, they point to the deficiencies of commonly used indicators measuring institutional quality. The critique primarily exhausts the historical flaws of aggregated indicators that are similar to the one employed in the present study. For example, the authors take issue
12
with the fact that former dictatorial governments were able to score as high as democratic gov-ernments for choosing good policies.10 Since the present investigation examines a recent time period consisting of only four and a half years, this controversy is irrelevant.
The relation between a politician’s work effort and his institutional background has hardly been explored thus far. The papers closest to mine are by Nannicini et al. (2013) and Fisman et al. (2013). While the mentioned study by Nannicini et al. (2013) focuses on the assessment of districts’ likeliness to hold politicians accountable, I aim to take a look at a country’s general institutional conditions and its capability to hold MEPs accountable. As the authors emphasize, for social capital to be existent, political institutions are required which “allow politicians to be held accountable” [Nannicini et al. 2013, p.241]. Fisman et al. (2013) report a strong correlation between a MEP’s work effort and the level of corruption in his home country using data from the fifth, sixth and seventh EP. They find that a MEP is increasingly susceptible to shirking when faced with a higher level of corruption at home.
Due to this direct evidence on MEPs and the beneficial impact democratic institutions tend to have in general I expect a MEP’s work effort to be positively associated with the institutional background in his home country.
3. Data and Empirical Strategy
The sample consists of 745 MEPs from 27 countries. The number of MEPs is not constant over the investigated period of time. Initially, there is data on 656 MEPs available. Due to the enlargement of the EP in December 2011, this number increases to 745 MEPs by the end of 2013.11 Operationally, there is no need to drop any of these observations. The late entrants don’t differ from their parliamentarian colleagues except for the commencement of work.
The analysis spans 54 months between July 2009 and December 2013, all of them belong-ing to the seventh legislative period of the EP. The study sets in directly after the salary harmo-nization in July 2009. This enables me to focus exclusively on non-pecuniary incentives that potentially drive a MEP’s work effort. The 54 months are then divided into nine intervals of six months each. I deliberately choose the period following the beginning of the Euro crisis. As sec-tion 3.1 reveals, the analysis covers a period characterized by considerable turbulences affecting a MEP’s socio-political background. Intervals of six months are therefore adequate to guarantee
10
According to the authors, the conditions under which good policies were implemented, i.e. whether the politicians in charge are constrained by democratic elections or not, must be taken into account by an indicator measuring institutional quality.
11 Officially, data on 757 MEPs from 28 countries are available by the end of 2013. Croatia became part of the EU
13
that these turbulences result in sufficient fluctuation in the independent variables, which in turn facilitates the analysis (see section 3.4).
Moreover, both the data on MEPs’ work effort (see section 3.3) and the surveys among European citizens (see section 3.1) are published every six months.
3.1 Measuring the Public Attitude towards the European Parliament
The explanatory variable Trust is based on fieldwork carried out by “TNS Opinion & So-cial” on behalf of the European Commission. The surveys are published in the “Eurobarometer”.12
Approximately 100013 face-to-face interviews per country are conducted in member states of the EU member states every six months. A “multi-stage random sampling de-sign”14
is used in every country to avoid sampling errors and guarantee a proportional coverage of its population density. The application of population size weighting ensures the proper repre-sentation with respect to sex and age.
Knack and Keefer (1997) highlight language difficulties as a further potential threat that might lead to an inconsistent measurement of trust. TNS Opinion & Social obviates the threat of translation errors with a “multistage process” characterized by detailed and uniform instructions. Moreover, proof reading and back-translation control are applied.
The reliability of various surveys (e.g. by “General Social Service” (GSS) and “World Values Survey” (WVS)) which try to measure interpersonal trust has been challenged by several authors. A study by Glaeser et al. (2000) suggests that the answers given by participants in GSS trust surveys are not characteristic for their behavior in trust games. Others point out that these surveys are typically conducted among students and therefore lack the necessary heterogeneity and representativeness given in the sample of the present study (Naef and Schupp, 2009).
Moreover, response bias is unlikely to be a problem for the indicator employed in the pre-sent analysis. While surveys asking about interpersonal trust might suffer from such phenome-non, there is no obvious reason for participants to deny their attitude towards a political institu-tion. Hence, the employed survey meets the most relevant technical requirements. Since it might still be considered one-dimensional, I assess the robustness of the basic findings by applying two further indicators of the Eurobarometer in section 5.
12 The surveys employed in the present study are part of the Eurobarometer editions 72 to 80:
http://ec.europa.eu/public_opinion/archives/eb_arch_en.htm, accessed on June 16th, 2014.
13 The number of interviews conducted deviates from this norm in five cases due to the countries’ size: Cyprus,
Malta and Luxembourg (500 interviews each), UK (1300) and Germany (1500).
14 For a detailed overview see:
14
In the course of interviews conducted by TNS Opinion & Social, European citizens are asked about their attitude towards the European Parliament.15 Operationally, I use the share of people who tend to trust the EP in order to assess a country’s level of trust. Based on the discus-sion in section 2, I rely on this survey to illustrate the relationship between a MEP and the peo-ple he is accountable to. Assuming that a MEP is regularly in contact with the peopeo-ple of his con-stituency16, he learns about their concerns and sorrows regarding the European Union. Presuma-bly, he also gets to know about their faith in the EP to solve issues. Drawing from the modified definition introduced in section 2, it is also legitimate to think of Trust as a country’s amount of cultural acceptance. I conjecture that a MEP adapts his work effort subsequent to such exchang-es. As the discussion in section 2 further illustrates, a decline in Trust can theoretically either result in a higher or lower work effort, depending on the underlying framework.
Figure 1 shows exemplarily for Spain the recent movement in the variable over the ob-served nine periods. The level of trust drops by 30% in total. The crucial question is what ulti-mately caused this drop. Was Trust manipulated by an exogenous factor such as the Euro crisis? Or can its development be driven by the work of the EP? At first glance, both scenarios are in-tuitive. If the latter was the case though, the inference of causal claims would be invalid due to simultaneity. As reported in section 2, the individual level of trust seems to be shaped by the perceived benefits gained through the EP as well as the personal socio-economic situation (Ar-nold et al., 2012). While the former factor of perceived benefits might be an important long-run determinant of citizens’ trust towards European institutions, I argue that it is of secondary im-portance regarding the investigated period of time. Moreover, by controlling for macroeconomic outcomes through the countries’ unemployment rates (see section 3.4) I cover potential short-term fluctuations. From a voters’ point of view, it seems unlikely that there are any further ac-tions taken by MEPs that are pivotal when she assesses her level of trust towards the EP.
On the other hand, the Euro crisis had a powerful effect on European citizens’ level of trust towards the EP (Roth et al., 2011). Indeed, the Euro crisis, which directly burdens millions of Europeans, is most likely to be the driving force in explaining the development of Trust. A second glance at Figure 1 is in accordance with this reasoning. The downward trend in Trust is most pronounced in the first three periods, i.e. in the immediate aftermath of the crisis. This af-termath was dominated by the attempt of the European crisis management to consolidate public
15 The exact question is: “For each of the following European bodies, please tell me if you tend to trust it or tend not
to trust it: The European Parliament.” The possible answers are either “Tend to trust”, “Tend not to trust” or “Don’t know”.
16 Mocan and Altindag (2013) point out that MEPs are likely to spend “most of their time with their constituencies
15
budgets. The most prominent feature has been and continues to be the implementation of rigid austerity programs. In general these programs have been considered an unfair punishment among concerned citizens. Even though the EP is not responsible17 for the imposed austerity programs, general anger and disappointment due to the EU’s approach of how to handle the cri-sis have most likely resulted in lower reputation values for the EP.18 The small upward shift in
Trust going from period eight to nine can be interpreted as a first signal of economic recovery
and regain of support by the EP and its members.
Figure 1: Development of the Variable Trust
Data for the level of trust towards the EP is only available on an aggregated national level which is in part a weak point of the analysis. Due to the lack of the data’s further subdivision, I cannot control for single European regions’ peculiar attitudes towards the EP. The social and cultural variety within certain countries partially invalidates the assumption of a nationally ho-mogenous public attitude. Six out the 27 countries in the EU have so far established constituen-cies for the EP’s elections in order to account for such inner-country differences.19
These con-stituencies are likely to differ substantially in their attitude towards the EP. Technically, the del-egated MEPs from countries such as Belgium are therefore not endowed with the equal amount of cultural acceptance. In 2012, a pilot edition of the Eurobarometer20 was published which con-tains surveys conducted according to the NUTS-1 and NUTS-2 regional division. The trust
17
The European Commission negotiated on behalf of the EU the conditions of how to deal with countries seeking financial aid. The subsequent adoption of austerity programs was within the sphere of national Parliaments’ compe-tences.
18 There are a number of exceptions. Sweden’s level of trust towards the EP, for example, has remained fairly stable
over the observed period of time.
19 In Belgium, for example, 13 of the 22 available seats are reserved for the Flemish constituency. The Wallon
con-stituency is entitled to eight seats, while one seat belongs to the German-speaking concon-stituency.
20 A so-called “Flash Eurobarometer Report” (No.356), see:
http://ec.europa.eu/public_opinion/archives/flash_arch_360_345_en.htm#356, accessed on July 7th, 2014. 0% 10% 20% 30% 40% 50% 60% 1 2 3 4 5 6 7 8 9
16
cator21 that comes closest to the one employed in the present analysis confirms the regional dif-ferences. For the three Belgian constituencies, the level of trust ranges between 50% (Wallon) to 58% (Flemings). So far this edition marks a nonrecurring exception.
3.2 Measuring the Quality of Political Institutions
To provide a holistic assessment of a MEP’s institutional background, I primarily rely on an aggregated index covering various institutional and political characteristics (see below). In contrast, the robustness section exploits two alternative indicators obtained from the World Bank’s series of governance indicators. Based on the powerful impact political accountability has on politicians’ behavior, I consult the index “Voice and Accountability”. This indicator cap-tures the “extent to which a country’s citizens are able to participate in selecting their govern-ment”.22
An in-depth study of the governance indicators’ composition praises the facilitation of “meaningful cross-country and over-time comparisons” [Kaufmann et al. 2011, p.243]. The lit-erature offers various definitions for a country’s institutional quality. Most suitable for the pre-sent index is the one by Guerriero (2014), who defines it “as the strength of the rules enabling voters to select more public-spirited representatives and check more closely their decisions” [p.9]. The author introduces the term “inclusiveness” to assess this aspect of political institu-tions. I therefore label the variable as Inclusiveness.
Furthermore, I explicitly examine the relation between a country’s degree of corruption and its MEPs’ work effort. According to Fisman and Miguel (2007), corruption can be seen as an identifier for a country’s social norms and work ethics. As the results by Fisman et al. (2013) indicate, corruption is strongly correlated with a MEP’s work effort and therefore is worth ana-lyzing as a single variable. The World Bank describes the “Control of Corruption Index” as a country’s ability to control corruption (labeled as Corruption-Control in section 5). It is a com-ponent of several indices constructed by experts with an exclusive focus on corruption.23
For the most part of the analysis I exploit data provided by the “Democracy Ranking As-sociation”. The “Global Democracy Ranking” is an aggregated indicator that evaluates a coun-try’s democratic quality according to six dimensions. Among these dimensions, a councoun-try’s de-gree of political rights is integrated and accounts for 25% of the whole Democracy Ranking
21 The exact question is: “I would like to ask you a question about how much trust you have in the European Union.
Could you please tell me if you tend to trust it or tend not to trust it?”
22 It also assesses a country’s degree of medial freedom. The World Bank rescales the data obtained from different
sources into common units. Thereby the index ranges from -2.5 (worst possible score) to 2.5 (best possible score). For a detailed description of the applied methodology see:
http://info.worldbank.org/governance/wgi/index.aspx#doc, accessed on June 16th, 2014.
23 Again the different values are standardized, i.e. the index ranges from -2.5 (worst possible score) to 2.5 (best
17
score. This indicator measuring political rights is itself a weighted indicator consisting of seven sub-scores collected by the independent watchdog organization “Freedom House”. The variable accounts for a country’s capability to “have representatives that are accountable them”.24
Still, the degree of a country’s institutional inclusiveness plays only a minor role in the final Democ-racy Ranking. The same problem arises regarding a country’s level of corruption. This indicator is based on the “Corruption Perceptions Index” obtained by Transparency International and con-tributes 10% to the final score on the Global Democracy Ranking.
At the same time, the Global Democracy Ranking incorporates several other institutional characteristics25 which assure a coherent picture of the political and institutional environment within a MEP’s home country. I label the variable as Democracy in order to account for the broad coverage provided by this aggregated indicator. At the same time, I am not able to disen-tangle the particular effects of the single components included in Democracy. For example, even though the level of corruption accounts for 10% of Democracy, it would be incorrect to infer that it accounts for 10% of the total effect Democracy has on a MEP’s work effort. Even regard-ing the sign of sregard-ingle dimensions’ coefficients I cannot draw any conclusions. While Democracy might overall turn out to positively affect a MEP’s work effort, single dimensions could theoret-ically have a negative impact. Section 5 therefore analyzes the respective impact of a country’s institutional degree of inclusiveness and control of corruption on its MEPs’ work effort.
As for Democracy, a country’s score ranges theoretically from 1 point (lowest possible score) to 100 points (highest possible score). For the present sample consisting of 27 EU-countries, the Global Democracy Index varies from 61.80 to 88.20. Table 1 reports summary statistics for all employed variables.
The identification of significant estimates for both Democracy and the two indicators ob-tained by the World Bank proves to be more complicated within the present empirical frame-work. The fixed effects regression model introduced in section 3.4 focuses on changes in the dependent variable caused by changes in the explanatory variables both expressed in deviations from their country mean (Stock and Watson, 2012). Unlike Trust, the variables measuring insti-tutional quality experienced only little variation during the investigated period across countries. Figure 2 illustrates this phenomenon again for Spain. A drop in Democracy from 78 to 76.9 im-plies an overall change of barely 1.5%. While the Euro crisis caused societies’ trust towards the
24
For a detailed overview see:
http://www.freedomhouse.org/report/freedom-world-aggregate-and-subcategory-scores#.U66CSvl_vh5, accessed on June 18th, 2014.
25 Apart from political inclusiveness and control of corruption, the Global Democracy Ranking accounts for a
coun-try’s civil liberties, gender gaps, freedom of press and governmental composition. For an overview of the single variables’ composition see: http://democracyranking.org/wordpress/?page_id=747, accessed on June 16th
18
EP to fluctuate considerably, the quality of political institutions remained fairly stable and ro-bust. All EU-countries can be considered established democracies over the sample. Under the pooled OLS regression (see equation (3) in section 3.4), which captures the impact of the institu-tional background from a different angle, significant estimates are more likely.
Figure 2: Development of the Variable Democracy
Both the Global Democracy Ranking and the two World Bank indicators are released only once a year.26 Neither source provides data for the year 2013. I therefore use the respective val-ues of 2012 as a proxy for 2013. The construction of six-month time intervals implies that these explanatory variables are invariant for two consecutive periods, whereas their values remain the same for the last four periods, i.e. from the sixth period onwards.
3.3 Measuring Work Effort
Depending on the degree of transparency, a variety of activities have been used by previ-ous authors to define a politician’s work effort. Gagliarducci et al. (2011), Bernecker (2014) and Nannicini et al. (2013) are examples where the attendance rate is the only employed parliamen-tary work measure. Bernecker (2014) argues that an absent German Parliamentarian is rent-seeking [p.7]. Nannicini et al. (2013) consider the Italian Parliamentarians’ absence from Par-liament as a “breach of the implicit contract between the representative and his voters” [p.234]. While the exclusive employment of the attendance rate to operationalize shirking might be ap-propriate for national Parliamentarians, this measure has a lower external validity in the case of MEPs.
As mentioned earlier, the official workplaces of a MEP are in Strasbourg and Brussels. The EP is located in Strasbourg, where the MEPs hold in general their plenary sessions. Brussels hosts the EP’s committee meetings as well as a few plenary sessions. Moreover, the political
26 The Global Democracy Ranking offers no data on Malta and Luxembourg. These countries are therefore dropped
from the sample when this variable is consulted. 76 76,5 77 77,5 78 78,5 1 2 3 4 5 6 7 8 9
19
groups formed in the EP usually meet in Brussels. During the investigated period of time, 51 plenary sessions – each consisting of four consecutive days – were held in Strasbourg. In addi-tion, 15 sessions took place in Brussels. The sessions are primarily used to decide on draft laws proposed by the European Commission, “the only institution empowered to initiate legislation” in the EU.27
Apart from his duties in the EP, a MEP is supposed to devote some time to his home con-stituency and attend external events. As mentioned before, during a legislative period MEPs are likely to spend a substantial amount of time at home. Even though it would be appropriate to include these activities to assess a MEP’s work effort, I can neither account for the effort a MEP invests in constituency work nor the external events he attends.
A precise depiction of a MEP’s work effort in Parliament is on the other hand perfectly feasible. I closely follow the approach of Mocan and Altindag (2013), who consult five work indicators besides the attendance rates. An alternative would have been to bundle a MEP’s activ-ities to generate a “Legislative Production Function” as done by Fisman et al. (2013). This strat-egy is based on a concept introduced by Dal Bó and Rossi (2011). Apart from the technical chal-lenges of implementation, the weighting of individual work activities turns out to be problematic and inevitably subjective. For example, should the attendance rate be valued higher than the amount of questions asked?
The employed work indicators for this study are obtained from the nongovernmental or-ganization “VoteWatch Europe”. This independent oror-ganization aims at promoting “better de-bates and greater transparency in EU decision-making”.28 VoteWatch Europe gathers the rele-vant data from the EP’s website and additionally receives authentic files from the EP’s Secretar-iat to guarantee accuracy.
The data on MEPs’ attendance rates (variable labeled as Attendance in the following) are restricted to voting decisions where an electronic record of the decision took place (so-called “roll-call vote”). A roll-call vote is necessary every time the simple “show of hands” approach is not enough to clearly identify a majority. This method was required in 1.454 voting decisions during the investigated period of time. With a total of 12.646 voting decisions, the analysis is based upon the attendance rates for slightly more than 10% of all decisions. This phenomenon is very common for researchers referring to a Parliamentarian’s attendance rate. Apart from Mocan and Altindag (2013), who rely on Votewatch Europe as well, Bernecker (2014) has only data on
27 See: http://www.europarl.europa.eu/aboutparliament/en/0081f4b3c7/Law-making-procedures-in-detail.html,
accessed on June 16th, 2014.
28 For more information on the purposes of the organization see: http://www.votewatch.eu/en/about-us.html,
20
German Parliamentarians’ attendance in mandatory voting sessions at his disposal. These ses-sions amount to only a fraction of the total amount of decises-sions taken.
Fisman et al. (2013) point out that the measurement of a MEP’s attendance based only on his registration at the beginning of a plenary session could turn out to be misleading.29 With the data at hand I obviate such misinterpretations. A MEP who attends a roll-call vote is unambigu-ously exerting work effort in that moment.30 Moreover, it seems unlikely that a MEP who at-tends a roll-call vote of a session does not stay for subsequent votes, irrespective of whether they are further roll-call votes or not. Still, for the sake of transparency and to assure an overall as-sessment of a MEP’s work effort, Votewatch Europe demands an extended use of roll-call votes. Due to the lack of data, I’m not able to control for excused MEPs (e.g. because of health issues).
Apart from the variable Attendance, I introduce six further indicators to measure a MEP’s work effort. The following definitions build on the information provided by the EP.31
MEPs have the opportunity to address other EU institutions by asking parliamentary ques-tions. It is primarily an instrument to scrutinize other EU institutions and bodies. This effort is reflected in the variable Questions.
Opinions (Opinions) are processed by MEPs in case they have suggestions for improve-ment concerning a report drafted by committees.
Drafts and amendments of reports (Reports-Drafted and Reports-Amended) are reserved for MEPs who participate in a parliamentary committee that deals with a legislation text submit-ted by the European Commission. At the end of 2013, 22 of these committees exissubmit-ted, each cov-ering a special topic. The number of committee members varies “between 24 and 76 full mem-bers and an equivalent amount of substitutes”.32
The vast majority of MEPs is either a direct member of a committee or one of its substitutes.
Declarations (Declarations) have a maximum length of 200 words and present only the au-thors’ and signers’ opinion on a respective topic.
Resolutions (Resolutions) are subject to the same formal restrictions as declarations. De-pending on the issue, a resolution is handed on to the responsible committee. The committee
29
Austrian MEP Hans-Peter Martin’s filmed evidence of some of his fellow MEPs’ abuse of expense allowances provoked popular outrage. The material shows several MEPs signing the register in the morning and leaving the EP subsequently. See: http://www.youtube.com/watch?v=uBmwzAUHRLw, accessed on June 17th, 2014.
30 Votewatch Europe therefore uses the term “participation rate” to emphasize the act of participating in a voting
decision instead of simply being “in the building”.
31 See: http://www.europarl.europa.eu/aboutparliament/en/00623fe732/How-plenary-works.html, accessed on June
9th, 2014.
32 For a detailed description of the committees’ work see: http://www.europarl.europa.eu/committees/en/home.html,
21
then decides on how to proceed with the resolution. At best, the resolution is used as a template for a draft law.
I decided against the inclusion of a MEP’s number of speeches given in the EP. The re-spective data available on Votewatch Europe make no distinction between oral speeches in front of the plenary and simple statements a MEP is allowed to give in order to explain his voting de-cision. While the former component certainly contributes to a MEP’s work effort, the latter is only a voluntary explanation of an activity which actually symbolizes the work effort that was exerted beforehand. Due to this ambiguity the indicator is dropped.
3.4 Empirical Strategy
The main empirical specification is as follows:
(1)
Work Effort stands for one of the seven inner-parliamentary activities carried out by the ith
MEP from country c in period t. Trust indicates the level of trust country c has towards the EP in period t, measured in percentage points. Democracy is the absolute score obtained by country c in the Global Democracy Ranking in period t. As explained in the introduction of this section, a period t consists of six month. While data on Work Effort and Trust are available on a biannual basis, new information on countries’ democratic quality (Democracy) is provided only once a year.
As a control variable I include the country-specific and biannual rate of unemployment (Unemployment). Over the investigated period of time, most member states of the European Un-ion have been affected by severe economic crises. The severity varied among countries as well as over time. The unemployment rate of country c in period t is measured in percentage points.33 As mentioned before, Roth et al. (2011) report a significant negative effect of EU-countries’ unemployment rate on Trust during the years of crisis, i.e. from 2008 until 2010. The multiple regression model introduced by equation (1) is based on the assumption that the independent variables are not perfectly multicolinear, i.e. no regressor is a linear function of another. For the present sample, a negative correlation of -0.30 is implied between Unemployment and Trust. The independent variables are therefore far from being perfectly multicolinear. No control for MEPs’ parliamentary salary is required since it had been adjusted to a uniform level before the study begins.
33 Using alternative measures for the Euro crisis (e.g. GDP growth or per capita income) imply very similar results
22
While some of the authors introduced in the literature review examined issues that are sim-ilar to the one at hand, none of them relied on a fixed effects regression model. The present study exploits the specific advantages of this model.34 Most importantly, the model enables me to control for the impact of time-invariant and unobserved factors through the MEP-specific intercept . For example, a MEP’s inherent political talent and his general work ethic are likely to influence his work effort. The direct measurement of such individual characteristics is practi-cally not feasible. The identification of reliable variables in order to proxy them is challenging. The fixed effects regression model eliminates the impact of these characteristics and thereby prevents a common source of omitted-variable bias.
In order to control for potential time effects, the main specification is supplemented by time dummies (equation (2)). This second specification captures through the effect of unob-served factors that are constant across MEPs but evolve over time. For example, reforms like the Treaty of Lisbon could be such a factor. The treaty has changed the EP’s political compe-tences in general and therefore affects all MEPs likewise. Policy measures taken by the Europe-an Central BEurope-ank are a second potentially relevEurope-ant time effect. The EuropeEurope-an interest level, for example, has been adapted several times within the investigated period of time. In case it had an impact on a MEP’s work effort, the time dummies are going to absorb it.35
(2) To some extent, both the individual-fixed effects and the time-fixed effects shield the main specifications from potential reverse causality. For example, a country’s level of trust towards the EP might be downward biased due to the fact that the work effort of MEPs from this country is in general low. Such long-lasting behavior of MEPs, which affects their reputation at home, would again be captured by . As I argued above, I do not consider reverse causality to be a relevant issue under the prevailing circumstances. Apart from the included rates of unemploy-ment there are no further salient factors that vary across countries and over time that potentially influence people’s attitude towards European institutions. Throughout I do not report the coeffi-cients attached to either the individual-fixed effects or the time-fixed effects because they are not of primary interest. To sum up, endogeneity – especially in terms of omitted-variable bias – is a serious threat to this analysis. I therefore consider the fixed effects model with its unique
34 The preference of a fixed effects model over a random effects model is in accordance with the results delivered
by the “Hausman Test”. The random effects model would only be adequate for the (rare) cases in which individual-fixed effects (captured in ) are not correlated with the explanatory variables (see e.g. Allison, 2009). This hypoth-esis is clearly rejected under the given circumstances.
35 For example, the increase of 0.25 base points in July 2011 was widely interpreted as a signal of economic
23
teristics to be most adequate. The described features allow me to explore how the change in the MEPs’ socio-political background affected their work effort over the years of crisis.
Unless otherwise noted, I make continuously use of standard errors clustered at the indi-vidual – i.e. MEP – level following Mocan and Altindag (2013). Technically, clustered standard errors are valid if an error term within a cluster is potentially heteroskedastic and autocorrelated [Stock and Watson 2012, p.366-367].36 In other words, this approach allows the error terms within a cluster to exhibit either heteroskedasticity or autocorrelation – or even both phenomena – over time. However, the regression errors across clusters are assumed to be uncorrelated. Some researchers considered this assumption as too strong (e.g. Nannicini et al., 2013; Fisman et al., 2013).
Fisman et al. (2013), for example, choose for their investigation of MEPs’ work effort a more conservative strategy by clustering the standard errors at the country level. Against the background of the present analysis, according to which all MEPs from a single country are as-sumed to be homogeneous in their socio-political background, this approach appears legitimate as well. Section 5 therefore provides estimates based on country-clustered standard errors as a robustness check.
For comparison, a pooled OLS regression model is employed (equation (3)). This third specification ignores the panel structure of the data, i.e. the model stacks the panel data over individual MEPs and over time into one single regression. The purpose of this baseline model is to see whether it suits the main estimates obtained through equations (1) and (2).
(3) As I suspect fixed effects to be present and moreover correlated with the independent vari-ables, the estimators in equation (3) would be inconsistent if these fixed effects remained ex-cluded from the regression.
The variables of interest remain Trust and Democracy. In addition, five control variables are introduced (vector ). Table 1 provides an overview of all variables’ definitions, data sources, and statistical summaries. Equation (3) includes Turnout, i.e. the turnout in the 2009 EP elections for every country c measured in percentage points. Based on the results of Bernecker (2014), there is good reason to act on the assumption that Parliamentarians, who experienced a fierce and close election campaign, tend to exert more work effort. This issue is only vaguely reflected in a country’s turnout. Finally, the regression accounts for individual-specific features (age and gender) and covers a MEP’s tenure and group affiliation in Parliament.
36 Clustered standard errors represent therefore one type of “heteroskedasticity-and autocorrelation-consistent
24
Table 1: Summary of Variables
Variable Definition Source Mean Min Max Abs.
Work Effort: Attendance: Questions: Opinions: Reports-Drafted: Reports-Amended: Declarations: Resolutions: See text. See text. See text. See text. See text. See text. See text. VoteWatch Europe (for each case), avail-able at: http://www.votewatch .eu/ 84.75% (14.40%) 9.948 (24.471) 0.271 (0.666) 0.280 (1.123) 8.233 (9.065) 0.158 (0.454) 3.629 (8.039) 0% 0 0 0 0 0 0 100% 626 10 38 156 7 71 6298 6298 6298 6298 6298 6298 6298 Social Back-ground: Trust: EU-Inclusion*: EU-Perception*: See text. See text. See text. Standard Eurobarometer (for each case), available at: http://ec.europa.eu/ 45.48% (12.22%) 31.35% (11.50%) 35.54% (10.73%) 18% 9% 13% 76% 243 65% 243 66% 243 Political Back-ground: Democracy: Inclusiveness*: Corruption-Control*: See text. See text. See text. Global Democracy Ranking, available at: http://democracyranki ng.org/ 75.90 (6.04) 1.131 (0.306) 1.042 (0.788) 61.80 0.30 -0.27 88.20 108 1.71 108 2.52 108 Worldwide
Govern-ance Indicators, both available at: http://worldbank.org/ Control (FE-Model & Pooled OLS)
Unemployment: See text. Eurostat, available at:
http://epp.eurostat.ec. europa.eu/ 10.10% (4.95%) 4.05% 27.5% 243 Other controls (Pooled OLS)
Turnout: See text. European Parliament,
available at: http://www.europarl.e uropa.eu/ 43.38% (15.01%) 0.652 (0.476) 50.646 (10.575) 0.537 (0.499) 0.836 (0.369) 19.64% 0 21 0 0 90.75% 1 81 1 1 27 7 746 746 746 746
Male: Dummy equal to 1 if the
MEP is male, 0 if MEP is female.
VoteWatch Europe (for Male, Age,
New-comer and Moderate-Group), available at:
http://www.votewatch .eu/
Age: Age of each MEP at the
beginning of the observed time span, i.e. July 2009.
Newcomer: Dummy equal to 1 if the
MEP entered the EP for the first time in 2009 (or later in case of substitute members), 0 if MEP had prior tenure(s) in the EP.
Moderate-Group: Dummy equal to 1 if MEP belongs to a parliamentary group, which as a basic principle, holds a positive view towards the EU.
Note: 1. The column Mean reports the mean value and the standard deviation in parentheses for each variable.
2. All variables defining work effort have been collected on half-yearly bases from 14 July 2009 until 31 December 2013.
25
4. Results
4.1 Results of the Fixed Effects Regression Model
Table 2 reports the estimates for the main empirical strategy (equation (1)). The effect of
Trust on a MEP’s work effort is unambiguous. Regardless of the employed work indicator, the
estimates of imply a negative relation between a country’s level of trust towards the EP and its MEPs’ work effort. With the exception of Declarations, these estimates are significant at 1% (specifications (2) to (5) and (7)) or at 5% (Attendance). This is consistent with the hypothesis derived within the modified Principal-Agent framework in section 2. Accordingly, the higher
Trust is, the lower a MEP’s work effort turns out to be. A trustful voter is less likely to punish
low work effort as he cannot imagine a MEP to opportunistically abuse this trust.
Due to the mixed evidence in support of rent-seeking politicians (Lott and Bronars, 1993), the alternative explanation derived from the concept of political capital is equally valid: MEPs from countries where the level of trust towards the EP is relatively low work harder in order to convince the voters they are responsible to and regain their trust. The impact of Trust on single work indicators is partially substantial. A 1% increase in Trust induces a 9% decline in partici-pated roll-call votes (Attendance). The number of questions asked by a MEP (Questions) drops by 36 due to a 1% increase in Trust. Also the amount of amended reports (Reports-Amended) is remarkably lower (26) for a 1% increase in the level of trust.
In contrast, the coefficient of Democracy is only in part statistically significant within this framework. As discussed in section 3.2, this observation is not unexpected. Columns (3) and (5) of Table 2 indicate nonetheless that the quality of political institutions positively affects a MEP’s number of opinions processed and amended reports. At first glance the magnitude of in these two cases is quite small. A 1 point increase in Democracy induces 0.066 more opinions pro-cessed (Opinions) and accordingly 2.175 more amended reports (Reports-Amended). Expressed in standard deviations, the effect is – at least with respect to Reports-Amended – more persua-sive. A one-standard deviation improvement in Democracy (6.04) implies 13 more amended reports. Moving from the lowest value of Democracy (61.80) to its mean (75.90) would increase the number of amended reports by approximately 30.
As a basic principle one can adhere to the intuition that more democratic political institu-tions in a MEP’s home country induce him to exert more work effort.
26
Table 2: Impact of Socio-Political Background on MEPs’ Work Effort
(1) (2) (3) (4) (5) (6) (7)
The dependent variable is:
Attendance Questions Opinions
Reports-Drafted Reports-Amended Declarations Resolutions Trust -0.091 (0.036)** -36.616 (9.760)*** -0.676 (0.174)*** -0.923 (0.323)*** -26.079 (2.276)*** -0.149 (0.123) -9.917 (1.897)*** Democracy 0.004 (0.003) 0.531 (0.459) 0.066 (0.014)*** 0.034 (0.028) 2.175 (0.221)*** -0.003 (0.011) 0.142 (0.146) Observations R² 6215 0.003 6215 0.011 6215 0.012 6215 0.002 6215 0.079 6215 0.0008 6215 0.012 Notes: 1. Robust standard errors in parentheses clustered by MEP.
2. *** denotes significant at the 1% confidence interval; **, 5%; *, 10%. 3. The control variable Unemployment is included in all specifications.
Table 3 shows the results for the MEP-fixed effects model supplemented by time effects as introduced in equation (2). The inclusion of time effects has to some extent an impact on the magnitude of Trust and Democracy (mainly with respect to Attendance and Reports-Amended), yet they do not contradict the previously obtained results. The effect on Attendance is even more pronounced when time effects are accounted for, i.e. the participation in roll-call votes is 13.5% lower for a 1% increase in Trust. The reduction in Reports-Amended is on the other hand clearly less distinctive (3 amendments less instead of 26) and no longer significant.
No clear and reliable pattern emerges for Democracy. Within this framework, the esti-mates of become significant regarding Attendance (at 1%) and Questions (at 5%), while they are no longer significant with respect to Opinions and Reports-Amended.
Table 3: Impact of Socio-Political Background on MEPs’ Work Effort (Time Effects Included)
(1) (2) (3) (4) (5) (6) (7)
The dependent variable is:
Attendance Questions Opinions
Reports-Drafted Reports-Amended Declarations Resolutions Trust -0.135 (0.049)*** -40.849 (13.644)*** 0.148 (0.232) -0.582 (0.462) -3.099 (2.645) -0.143 (0.153) -5.637 (2.257)** Democracy 0.011 (0.004)*** 1.187 (0.520)** 0.013 (0.015) 0.010 (0.029) -0.183 (0.253) -0.0006 (0.013) 0.229 (0.168) Observations R² 6215 0.011 6215 0.031 6215 0.035 6215 0.006 6215 0.222 6215 0.028 6215 0.076 Notes: 1. Robust standard errors in parentheses clustered by MEP.
2. *** denotes significant at the 1% confidence interval; **, 5%; *, 10%. 3. The control variable Unemployment is included in all specifications.
