Printer Friendly



In party-dominated political systems with proportional representation (PR) with closed list electoral systems, individual members of parliament (MPs) have incentives to stay loyal to the organizations that brought them to parliament. Party discipline is typically high in such systems, and defections are relatively rare (Grose and Yoshinaka, 2003; Shabad and Slomczynski, 2004; Mershon and Shvetsova, 2008; Janda, 2009; Stefan, Gherghina and Chiru, 2012; O'Brien and Shomer, 2013; Martin, Saalfeld and Strom, 2014). However, occasionally, there are MPs who, for various reasons, fall out with their political organizations and seek better matching parties to secure their career development, or achieve ideological or policy goals. In Eastern and Southern Europe, these movements have been shown to be more frequent than in typical Western contexts (Heller and Mershon, 2009; Volpi, 2018).

Sometimes, MPs' decision to defect comes as a direct response to crises faced by their political parties (Kreuzer and Pettai, 2003; Thames, 2007). Other times, they come incrementally with their development as seasoned political representatives (Gherghina, 2016; Mershon and Shvetsova, 2009; Heller and Mershon, 2005). Yet other times, they are slowly marginalized by their groups for not fitting the ideological or policy profile of the organization (Desposato, 2006; Owens, 2003; Bozoki and Ishiyama, 2002). Either way, defector MPs are support votes lost by one party and won by the receiving party on the roll call voting floor.

In general, there is little attention given to party switchers. The phenomenon is rare enough and the populations small enough to barely be relevant for traditional statistical analysis, yet large enough to deter systematic qualitative approaches (Heller and Mershon, 2009; Volpi, 2018). We argue that the network approach to studying party switchers brings both the quantitative quality to the analysis that is not affected by the size of the population, while also providing some micro-and macrolevel qualitative context, previously overlooked with more traditional approaches.

The science of complex networks offers a new and fresh perspective on human interactions across a diverse range of scientific fields. The focus is put on areas in which social network analysis (Carrington, Scott and Wasserman, 2005; Borgatti et al., 2009) and political science overlap (Lazer et al., 2009; Gerber, Henry and Lubell, 2013; Dal Maso et al., 2014). Thus, it is made possible to analyze the dynamics of local-level interactions (e.g., strategic decisions of MPs) and emergent phenomena at the global level (e.g., the cohesiveness of the parliamentary party system). A powerful tool employed by network science is centrality analysis (Wasserman and Faust, 1994), through which nodes are assigned a meaningful importance in the network configuration. To this end, we propose the usage of PageRank (Page, 1999) to measure the importance of a party switcher in the context of the parliamentary network of party members.

Network analysis has been used in legislative studies in the past 10 years, significantly contributing to the collective knowledge and understanding of MPs' structural, organizational and inter-personal interdependency and role within dynamic networks of policy and politics. However, most of them are focused on other types of legislative collaboration, such as co-sponsorship of legislation (Fowler, 2006; Bratton and Rouse, 2011; Kirkland and Gross, 2014), information exchange (Ringe, Victor, and Gross, 2013), policy innovation (Mintrom and Vergari, 1998), or friendships (Kirkland, 2011), mostly in the US context, and we are not currently aware of any study examining the problem of party switching from a network perspective in the European context (i.e., exploring or testing network dynamics of politicians' shared party memberships).

Literature on the effects of defector MPs on party discipline and support in various contexts has previously shown that these individuals bring a series of benefits to the receiving parties, from social and political networks, to funding opportunities, and a personalized vote from constituencies, based on their experience in office, time in politics and their personal and political popularity with the electorate (Olson, 1998; Tavits, 2005; Tavits, 2009; Gherghina and Jiglau, 2011). On the other hand, they were also found to bring some negative aspects with them--lack of credibility in the eyes of the electorate, depending on the timing of the switch, and lack of trustworthiness, depending on the frequency of their switching between parties (Gherghina and Chiru, 2014; Desposato, 2004; Tavits, 2008).

Thus, the receiving parties of defector MPs have to balance between the benefits and the burdens they take in together with a particular MP. In theory, MPs can opt to join any political party at any time, without much justification, except for those appropriate for their electors. In practice however, there are strong incentives of MPs to join parties strategically, in order to get support from the party leadership for re-election (Klein, 2016a; Klein, 2016b; Stefan, Gherghina and Chiru, 2012; Heller and Mershon, 2008), to further their political career in public office (Pinto, 2015; Owens, 2003), or to gain more credibility on a personal vote (Tavits, 2005; McMenamin and Gwiazda, 2011).

There are thus two overlapping processes that happen with respect to party switchers: on the one hand, there is the strategic decision of MPs to join certain parties, to further their personal agendas; on the other hand, there is the strategic decision of receiving parties to support defector MPs for re-election on their lists, to further their political agendas. We thus ask the following questions:

(a) Are the strategic decisions of defector MPs motivated by party performance? In other words, are defector MPs motivated by the performance of the receiving parties or by the prospects of the parties they intend to leave behind?

(b) What are the micro-level effects of individual MPs' strategic decisions to move from one party to another at the macro-level--the parliamentary party scene?

To answer these questions, in this paper, we pursue two aims: 1) to explore the temporal dynamics of strategic choices of MPs who defect from one political party and join other parties over the course of their careers in the Romanian Parliament. And 2) to explore the temporal dynamics of strategic choices of parliamentary parties that receive defector MPs and support them for re-election on their lists. This is an initial exploratory study of the phenomenon of party switching using network analytical concepts and tools. More traditional approaches to determinants of party switching at the individual and party levels are not our concern in this study.

Although the populations for such a study are generally small, they are significant in political systems characterized by: (i) issue polarization, where every support vote counts both for passing legislation, as well as for getting parties the electoral and public credibility backed by political elites (Olson, 1998; Tavits, 2005; Tavits, 2009; Gherghina and Jiglau, 2011); (ii) cartel parties, where parties depend and actively seek state resources--professional politicians being a key part of the resource pools of cartel parties (Chiru, and Ciobanu, 2009; Bozoki and Ishiyama, 2002); (iii) parliamentary and government coalitions (especially minority coalitions), where every resource at hand is valuable for maintaining a working majority and avoiding decision blockages (Gherghina and Jiglau, 2011; Stefan, Gherghina and Chiru, 2012; Tavits, 2005).

Since the fall of communism in Romania, in December 1989, the country has taken decisive steps into developing a competitive multi-party system and an active civil society. For the last 28 years a lot of changes had taken place on the political scene of Romania, such as the creation of new parties and dissolution of others, people leaving certain political parties to create new ones, or even people switching to a party representing the opposing ideology of his/her previous party. In this paper, we explore the landscape of post-1989 Romanian politics that reflects such dynamics.

Romania is an appropriate case to study for several reasons. First, the multi-party scene in this post-communist country has been relatively small and stable, with about five political actors who have secured leadership of the country from 1990 until today. There were some mergers, splits and fusions on the political scene, but the five large actors have been the ones dominating the political game for the past 27 years (Gherghina and Chiru, 2014) (5).

Second, these actors have more or less reliably alternated in office from one election to the other, even though the electoral turnover in Romania was not as clear-cut as in other post-communist countries, such as Hungary in the first 20 years, due to the many electoral and cabinet coalitions formed, in order to secure either parliamentary majority or some forms of governing consensus. Sometimes, these coalitions would be among political parties with ideologically opposing views. Given however the migration of political party attitudes increasingly towards to center of the ideological spectrum on policy issues, these coalitions were possible in the first place (Gherghina and Jiglau, 2011; Tavits, 2008; Dal Maso et al., 2014).

Third, precisely because of this dynamic game of coalitions, defector MPs play an important role, since they can destabilize or de facto stabilize fragile political arrangements, and they can act strategically using temporal momentum in political crises and on key issues.

Strategic decisions and unintended consequences

The literature in political science is mostly concerned with the strategic choices of defector MPs or those of the receiving parties to justify their acceptance. We are however interested in exploring the unintended consequences of these strategic movements at both the micro- and the macro-levels of the political system. Do party switchers' over time moves contribute to the fragmentation or the cohesion of the party system? Are individual party switchers penalized for their behavior or do their political careers thrive in new, supporting organizations? The literature in network theory allows us to operationalize the complexity of a dynamic system, making explicit both the local level interactions, as well as the intended and unintended consequences of these interactions at the level of the entire system.

Typically, the literature on party switchers defines the choices of MPs to leave one party and join a new party (a) by simply assuming their behavior is strategic; (b) by monitoring only the start and end party, without taking into account intermediary parties along the career path of a politician; and (c) without quantifying the amount of time spent by a politician with a given party (Gherghina and Chiru, 2014). Thus, the literature fails to properly account for the career contexts of party switchers and their level of opportunism around seeking re-election.

Our aim to fill in this gap in the literature is to propose a better quantification of party switching, a party loyalty penalty score that is sensitive to the complexity of individual strategic behavior, by providing a measure that makes explicit the extent of strategic behavior for each MP. The more opportunistic the choices of individual MPs, the larger their switching score.


Our data are extracted from the Romanian Parliament official website ( using Beautiful Soup, a Python library for parsing HTML documents (Richardson, 2017). The result is a plain text file containing a list of politicians in both the Chamber of Deputies, as well as in the Senate. Each Member of Parliament (MP) has assigned a sub-list of parliamentary parties in which he or she has activated, as well as the starting and end year for their mandate with the respective parties.

From this list, we construct temporal networks of MPs' party co-affiliation from 1990 until 2018, where two MPs (nodes) are connected if they share affiliation to the same party (links) while in a specific electoral mandate (cycle). The network is dynamic, because the co-affiliation connections change from one electoral cycle to the other, based on the composition of the two houses of the Parliament.

In terms of the graph modelling, the ROmanian POliticians NETwork (RoPoNet) is represented as an undirected graph G=(N, E), consisting of the set N of nodes and the set of edges E (i.e. unordered pairs of elements from N). In RoPoNet each node represents a politician; an edge (relationship) between two politicians is set if at some point in time, between 1990 and 2018, the politicians were members of the same political party.


We analyze three sets of measures: (1) the performance of political parties in parliament, in order to assess their level of attractiveness to individual MPs; (2) individual and party switcher scores, in order to assert the levels of parliamentary defections across time; and (3) the evolution of the network structure over time, to better understand the context for the co-evolution of party structure and politicians' behavior in the Romanian Parliament. In this paper, we resume our analysis to describing these measures and how they are related. We end the analysis with generating a few testable hypotheses that can be explored in a future study using inferential network analysis.

Using the data at hand, we calculate three novel measures, one at the individual level--Politician Switcher Score (SS), and two at the organizational level--Overall Party Performance (OPP) score and Party Switcher Score (PSS).

A Politician's Switching Score (SS) is thus composed of three elements:

[SS.sub.x] = [n.summation over (i = 1)] [st.sub.i] x [sw.sub.i] x [sp.sub.i] (1)

Where [SS.sub.x] is the individual politician switching score; i is the switch index for politician x. At switch i, politician x goes from party [pp.sub.i] (previous party) to [np.sub.i] (new party).

(i) The switching type of the MP ([st.sub.i]) (6)--based on whether the party a politician leaves behind has merged, fused or otherwise split the year the MP has made the decision to leave. A politician who leaves the party the year the party decides to change their structure, either to merge with another party or split from a larger structure, receives a smaller score than a politician who leaves the party before the political organization has made a clear decision to change their status. The first is considered to have switched to a new party due to organizational reasons, while the latter is more likely to have done it for career, ideological, or policy purposes.

[mathematical expression not reproducible] (2)

(ii) The switching weight of the MP ([sw.sub.i])--the aggregate experience of a politician with different parliamentary parties, based on the ratio of the number of years spent in a party they leave and the number of years they spend in a new party. The ratio is adjusted to the number of moves a politician makes between parties throughout their entire career in parliament. A politician who makes more moves and spends less time in a new party receives a higher score than one who makes less moves and spends more time with the new party. This element accounts for the lack of loyalty based on a politician's political acumen.

[sw.sub.i] = Number of years politician x has spent in [pp.sub.i]/ Number of years politician x has spent in [np.sub.i] (3)

(iii) The switching power of the parties ([sp.sub.i])--the ratio of political performance between the new party and the party from which an MP leaves. Political performance is calculated as the sum of political experience a party has in Parliament, government or with providing a Prime Minister. The more time a party spends in parliament, in government and by providing a PM from the total lifetime of the party, the higher their political power or performance score. This element accounts for the decision of an MP to switch to a high performing party that can ensure progress in their political careers.

[sp.sub.i] = Overall party performance for [np.sub.i] when switch i occurs/ Overall party performance for [pp.sub.i] when switch i occurs (4)

In order to calculate the politicians' switcher scores, we also need to calculate the overall performance of the parliamentary parties. This is a parsimonious additive measure, indicative of the attractiveness of political parties for individual MPs, based on the party's parliamentary power, government power and the extent of giving the country its Prime Minister.

In a political system characterized by strong political parties and in an electoral system conducive to party discipline, such as Romania under PR with closed electoral lists, the incentives for MPs to defect to political parties that can ensure their political careers is high (7). Thus, the penalty score for an MP who switched from a weaker party to a stronger one is higher, because a switch to a stronger party means better career opportunities. The Overall Party Performance score reflects the capability of the political organization to support its members' political careers.

A party's Overall Party Performance (OPP) is also composed of three elements that summarize a party's experience in high level office:

[OPP.sub.p] = [pp.sub.p] + [gp.sub.p] + [pm.sub.p] (5)

Where [OPP.sub.p] is the overall party performance score for party p.

(i) Parliamentary power of the party ([pp.sub.p])--the number of years the party has been in parliament, over the lifetime of the party (8), measured in years. The more a party was present in parliament, relative to the number of years it has existed on the political scene, the higher its parliamentary power.

[PP.sub.p] = Years in Parliament/Lifetime of party (6)

(ii) Government power of the party ([gp.sub.p])--the number of years the party has been in government, over the lifetime of the party, measured in years. The more a party was present in government, either on its own or in a coalition, relative to the lifetime of the party on the political scene, the higher its government power.

[gp.sub.p] = Years in government/Lifetime of party (7)

(iii) Prime minister power of the party ([pm.sub.p])--the number of years a party has given the country its Prime Minister, over the number of years the party has been in government. The more a party was present in government and the more years it gave a Prime Minister, the higher the performance score.

[pm.sub.p] = Years with Prime Minister/Years in government (8)

Finally, we aggregate the individual scores to the party level and across parliamentary cycles. This perspective highlights the organizations most open to receiving party switchers. The higher the party score, the more party switchers from all over the political spectrum the party accepts.

The Party Switcher Score (PSS) is the individuals' scores, aggregated at the level of the party and is calculated as the geometric mean of the individual politicians' switcher score, weighted by the size of the party. We calculate the measure only for parties that have switchers.

[PSS.sub.p] = [m.square root of [[PI].sup.m.sub.i=1][SS.sub.i]] (9)

Where i is the individual politician index, m the total number of switchers in party p; we further normalize this score as the relative overall party switcher score for party p:

[RPSS.sub.p] = [PSS.sub.p]/[PSS.sub.r] (10)

Where r is a reference party. We can choose r to be the party with the highest PSS score; in this case, we assure that 0 < [RPSS.sub.p] [less than or equal to] 1 for any party p.

Main Findings

Strategic moves of individual party switchers

We start by analyzing the Party Switcher Score (PSS) over the period 1990 until 2018 (present). The evolution of the average PSS shows an overall increasing trend of the score until 2018. Not surprisingly, at the party level, individual MPs migrate to new parliamentary parties in election years or soon before those, highlighting the opportunistic behavior of both MPs and receiving parties enabled by the electoral cycles. In Figure 5, we illustrate significant changes in the PSS in 1992, 1996, 1998/1999, 2003/2004, 2007/2008, 2011/2012, 2013, 2015, 2016, and 2018.

In Figure 6 we convert the data presented in Figure 5 to emphasize with red lines the variation of PSS right before, or during election years.

Plotting the Party Switcher Score change rate over time in Figure 2 reveals interesting dynamics around the goals of the receiving parties of defecting MPs. Overall, the patterns are similar across years--there are fluctuations of increasing and decreasing trends around elections. However, there are differences in whether the changes happen during the electoral year or before that time. These differences indicate possible roles party switchers play in the game and their value to receiving parties--e.g., strengthening parliamentary support coalitions or even manufacturing new parliamentary majorities.

The elections of 1992, 1996, and 2016 show similar patterns of increase in the overall PSS score, and a sharp decrease in the following year. We argue that these are indicative of turbulent times, where majority coalitions are hard to reach, and parties scout for defectors, as well as potential defectors scout for winning parties in hope to achieve new parliamentary majorities.

The elections of 2004, 2008, and 2012 also show similar patterns of PSS change, different from the other elections. The increase in the PSS score happens one year prior to the elections, suggesting that receiving parties aim at strengthening their parliamentary support coalitions. For the 2008 and 2012 elections, however, these effects are barely statistically significant. The patterns are significant however for the 2008 election.

In Table 9, we present the measurements of OPP and average PSS for the list of political parties that are or were active in Romania during the period 1990-2018. The parties are sorted by descending value of their average PSS from 1.654 (current PSD) down to 0.0 (UDMR, USR). We find a relative correlation between parties with high performance (OPP) and high PSS. We obtain a significant Pearson correlation coefficient of 0.653 between the two measures on all 15 parties, and a correlation of 0.747 for the top three parties in Table 9.

We associate a high PSS score with a heterogeneous party (i.e., a party accepting members with former memberships in other political parties), and, conversely, a high OPP score with an expected homogeneity of members within the same party. Thus, we could state that heterogeneity at the individual level actually leads to homogeneity at the party level.

Party performance in Romania and attractiveness for party switchers

We visualize the relationship between the performance and switcher score of each party, to understand whether the two measures are significantly related. To this end, Figure 7 plots in a two-dimensional space the considered parties from the Romanian political landscape. We suggest with a red line a linear correlation between PSS and OPP. Along the main red axis we find major players, such as FSN/PD/PD-L/PMP and ALDE, and smaller, coalition players, such as PER, PAC, PP-DD, PUNR. For these parties, whether individual MPs join them is a matter of the party's historical performance in office.

However, we find notable exceptions from this correlation. For example, UDMR and USR have PSS=0, but performances > 0. Conversely, PRM and PUR/PC have small OPP ([approximately equal to]0.5) but relative high PSS ([approximately equal to]1). These observations consolidate the public opinion regarding these parties. USR claims not to accept former party member, and was, thus, founded by new political actors. UDMR represents the Hungarian minority in Romania and is thus encapsulated. PRM is and was often a party of transition for many politicians, but never gained high performance.

Political organizations above the correlation axis seem to be parties where switching is more opportunistic than the rest of the players on the political scene, for both the individual MPs, as well as for the parliamentary parties. PUR/PC and PRM are transition parties for many MPs despite the parties' lack of high performance in office, while the larger and more visible PNL and PSD display better performance in office, but still welcoming many party switchers. Parties below the correlation axis, such as the small historical party PNTCD/PNT and UNPR, seem to be political organizations with more loyal members, less attractive to party switchers, despite their past performance in office.

From micro-level strategic moves to macro-level unintended consequences

Figure 8 displays the evolutionary dynamics of influential switcher MPs. Two MPs are connected if they share membership in the same party within the same electoral cycle. The size of the nodes and the color reflect their PageRank scores (influence scores) (Page et al., 1999). Larger and redder nodes are MPs who switched between parties within a given parliamentary cycle, by bridging between parliamentary parties. The graphs are roughly cut to reflect the left-right divide of the parliamentary party scene. L stands for parties on the left ideological spectrum (e.g., PSD), R stands for parties on the right ideological spectrum (e.g., PNL).

The first observation in Figure 8 is the temporal dynamics of fragmentation of parliamentary discipline in Romania. Up until 2004, the political scene is much more fragmented than after 2004, where, increasingly, the system evolves towards a two dominant party blocks rather than a multi-party system.

The second observation is that party switchers seem to move along with the process of crystallization of the parliamentary party system from fragmented multi-party to a coherent two-party block. The party switches until 2000 are characterized by migrations within the same political families, i.e., from left to left and from right to right parties. In the 2000-2004 period appear the first significant jumps across the political spectrum from left to right or the reverse. These patterns are then strengthened in the following parliamentary cycles. More data would be needed to assert the impact of the size of the party on these moves, but given the initial evidence, party switchers seem to have an integrative function within the party system.

The third observation is that heterogeneity at the individual (micro) level leads to homogeneity at the party system (macro) level. As the number of party switchers increases over the years, so do the density of connections among MPs, and the average number of connections per MP, while the number of network communities decreases steadily, as do the diameter of the network and the average path length (i.e. the average distance on the shortest paths across all pairs of network nodes). The layout algorithm used in Figure 4 is Force Atlas 2 (Jacomy et al. 2014), a force-directed spring embedder that pulls closer together nodes that are more similar to each other and pushes further apart nodes that are more dissimilar from each other. Similarity in this context is meant as a network-analytical term, measured as the Euclidean distance between nodes in a graph, i.e., the distance in terms of number of connections among MPs. The gravity indicator of the algorithm is high, so that all connected and disconnected components remain close together. Also, the nodes are not overlapping, to allow for a better visual interpretation of group sizes.

The left-right separation on the individual quadrants in Figure 4 has been done relatively, using a qualitative assessment of individual party manifestos. A more precise distinction was not our aim for this exercise, although in future research we will be able to apply different strategies to make the assignment replicable and more accurate (e.g., using text analysis of party or election manifestos, using a multidimensional scaling analysis of party policy outputs, or using expert surveys).

The measures at the whole network in Table 2 summarize the graphs shown in Figure 4. In Table 2 we present the summary statistics of the eight network snapshots. Network density measures the ratio of existing connections over all the possible connections in the network and shows whether the frequency of shared membership of MPs in different parties has increased or decreased with time. The range of the network density measure is from 0 to 1. The closer to 0 the density is, the sparser the network. Conversely, the closer to 1 the density is, the more connections there are within the network. Overall, network density decreases slightly until 2000 and then increases constantly until 2018. A significantly below average density is observed during the 1996-2000 period, when party switching as a network phenomenon has been lowest. Above average network density is observed during the last two electoral cycles, 2012-216 and 2016-2018. These are periods marked by significant crossparty movements.

The number of network communities decreases constantly since 1990, reinforcing the idea that the entire party system becomes more coagulated over time. A community in network terms has a very specific meaning: a group of nodes that share more links within the group than outside the group with other parts of the network. This interpretation is different than what one would expect from a more qualitative interpretation of communities, one that defines them in terms of strictly party affiliation, for example. A network community is defined by an algorithm that is agnostic to the qualitative traits of the nodes (e.g., party affiliation). Thus, a resulting community can be formed of MPs from different parties--what brings them together is the pattern of connections they have to different parties in Parliament. The communities are detected using the algorithm of Blondel et al. (2008), which is a heuristic method based on modularity optimization (Newman, 2006). While the community detection and layout algorithms have different purposes, it was demonstrated by Noack (2009) that the clustering produced by the two methods is equivalent.

Having understood this different interpretation of communities, the number of communities in the networks is indicative of the level of integration of MPs within smaller or larger groups. While the absolute numbers are not necessarily comparable across time, they are to be interpreted within each electoral cycle respectively. The more communities in the network, the more fragmented the party scene in Parliament. Conversely, the less communities, the more connected or integrated the party scene in Parliament. While in the first short electoral cycle, 1990-1992, there were ten network communities, by the electoral cycle, 2012-2016, the number of communities halved.

The degree of a node is the number of links it has to other nodes. In this case, the degree of an MP is the number of connections it has to other MPs, based on shared party membership. This score is influenced by two things: one, the size of the party (the more MPs in a party, the higher the number of connections an MP can make to other MPs), and second, the number of party switches an MP does during an electoral cycle (the more party switches, the more connected an MP is to MPs in the party one leaves and the parties one joins). Here too, the average number of connections per MP decreases from 1990 until 2000, and then increases. Again, the electoral cycles 1996 to 2000 shows the lowest average number of connections per MP, in line with the low density, suggesting the phenomenon of party switching was at its lowest during that time. Also, the periods 2012 to 2016 and 2016 to 2018 feature the highest number of connections per MP, reinforcing the trend from the density reports that the party switching phenomenon significantly increased in the last two electoral cycles.

Network diameter and the average path length are two measures of the small worldness of a network. The diameter is the longest path between any two MPs in a party sharing network. The average path length is the typical number of handshakes separating any two MPs in the network. The most notable period is 1992 to 1996, when both the network diameter and the average path length are significantly above average, indicating that during this period, an MP would need an average of three handshakes with other MPs to reach any other MP, and a maximum of six handshakes. After this period, these scores decrease. It takes only a maximum of three handshakes for an MP to be introduced to any other MP in the network, while only two handshakes, on average.

Discussion and Conclusions

In this paper, we explored provisional answers to two main questions: (a) Are the strategic decisions of defector MPs motivated by party performance? (b) What are the micro-level effects of individual MPs' strategic decisions to move from one party to another at the macro-level--the parliamentary party scene?

We acknowledge that the modern political system is a complex system (Bar-Yam, 2002), and thus treat it appropriately with measurement instruments that make explicit rather than assume complexity of behavior. In a complex system we often find properties or actions that are not scale-invariant (Boccara, 2010). As such, we investigated whether party switching, which generally has a negative connotation at the individual level because of the break in party discipline, has as emerging effect at system level a breakdown of the parliamentary party system. One of the main findings of the study is that, indeed, party switching is not scale-invariant. Despite the disruptive effects of individual MPs choices to defect for the party they leave party switchers' moves have an integrative effect at the party system level on the long run. By bridging between different political communities, party switchers fill in structural holes in the network.

We have attempted to answer these questions by proposing theoretical and methodological frameworks that make explicit, rather than take for granted, the strategic moves of parliamentary defectors. From a methodological perspective, we aimed at providing a better quantification of the quality of party switching moves, as well as exploring the mechanisms between the individual level party switching moves and the consequences on the performance of the involved parties. To that end, we have developed three novel measures for quantifying an individual MP' switching score (SS), the overall party switching score (PSS), and the overall party performance (OPP).

The individual switcher score proposed in this paper is a more sophisticated measure of individual opportunistic behavior in office, which considers three elements of their strategic calculations for defection. The SS measure makes some strong assumptions about politicians' behavior and motivations behind party defection while in office: first, that this strategic behavior is opportunistic, i.e., MPs are looking for opportunities to get re-elected, by joining parties that have higher chances of supporting their careers, whether in parliament or in other structures of the state. While we don't explicitly test this assumption, further research could focus specifically on this problem. A study of the re-election chances of party switchers in the Romanian Parliament from 1990 until 2008 (Gherghina, 2016) shows that around one in five politicians do get re-elected on the lists of their new parties, and that the dominant factors that help their re-election are the strategic choice of the new party and their legislative experience in office. Second, the algorithm weighs more the characteristics of the parties than individual characteristics in the process of deciding. Third, politicians are aware of the overall party characteristics and dynamics, allowing them to choose the best options on the political market.

There are several characteristics of the Romanian system that led us to assume that politicians are indeed aware of the performance of political parties on the electoral scene: Romania is a system dominated by strong parties, generally, strong party discipline, and parties reward loyalty in the form of office (clientelism and patronage are two phenomena strongly embedded into the Romanian party politics).

The politicians' Switcher Score (SS) is one of the most important propositions we make in this study, even though our attention is primarily directed at exploring the aggregated party scores. It is the stepping stone of the Party Switcher Score. For the purposes of this study, we do not challenge the assumptions on which the score is built, assumptions informed by previous relevant literature (.e.g., Gherghina, 2016; Gherghina and Chiru, 2014; Klein, 2016a, Coman, 2012).

In a follow up study, however, we will explicitly test the validity of this measure, both in terms of its accuracy, as well as in terms of its relevance to the existing literature. The main argument there is that the sophistication of the algorithm makes a valid and significant contribution to the understanding of party switching as a phenomenon, both at the level of individual politicians, as well as for the parties which they leave and those that receive them: are parties more likely to receive more opportunistic or typically loyal politicians? Do governing or parliamentary parties have different strategies when it comes to accepting defector MPs? Is there any difference between the decisions to accept defector MPs between small and large parties? Indications about the relevance of these questions have riddled the results of this exploratory exercise. In this study, however, we stop at showing that defector MPs occupy important bridging positions in an evolving parliamentary party scene, from a multiparty system to a polarized two-block party system.

We can also look at how politicians themselves are affected by their switcher scores are more opportunistic MPs less likely to get re-elected on the ticket of a new party? Does their switcher score affect the type of office they occupy after the win of the party? Also, more data on MPs' co-sponsorship decisions and voting records (roll call data) can bring additional information on the relevance of the politicians' switcher scores for receiving parties, as well as the political acumen and policy orientations of switcher MPs.

The Party Switcher Score (PSS) measure is not meant to be normative in nature, but rather pragmatic. To suggest that the measure is a summary score for parties attracting disloyal politicians would be reductionist, since the measure is complex enough to incorporate several strategic calculations an MP does, relying also on their experience in office and their awareness of the political market of opportunities. While at the individual level the switcher score reflects the opportunism of politicians, at the party level, this score reflects the adaptability of the organization to political competition.

The Overall Party Performance (OPP) index we construct is still a raw measure of performance that is meant to help understand the attractiveness of certain parties for individual MPs. Currently, the measure captures the number of years a party is in Parliament and government and how long it lasted on the political scene altogether. We plan to extend it to also consider the size of the parties (as shares of seats in Parliament and the share of offices in government). The evolution of the network structure and the distribution of influential MPs within these yearly affiliation networks is a strong indication of the process of local-level moves leading to unintended macro-level phenomena. One of the main findings of this paper is that the small-scale and incremental moves of MPs switching parties along their political careers in office have correlated with evidence of the simplification of the parliamentary party groups from a multi-party, fragmented system, to a two-block, cohesive system.

Taking a novel perspective to understanding the moves of political party switchers revealed not only aspects of the strategic decisions behind individual and party choices of joining forces, but, most importantly, the unintended consequences, on a larger scale, of their affiliation. On the one hand, individually, party switchers are seen with skepticism and criticism because of their lack of loyalty to the organizations that promoted them to office. On the other hand, they are received and supported by new parties on the account that they provide support for the party, both in parliament, and in their constituencies. Also, the decisions of MPs to join new parties has little to do with the overarching political environment and more to do with prospects of getting re-elected, while the decisions of parties to receive defector MPs has little to do with the overarching political environment, and more to do with their immediate function within the party's support system.

Yet, a clear finding of our initial, exploratory study, has shown that small, local level, strategic decisions of defector MPs to leave behind some parties and join different parties have consequences at the macro level, for the entire parliamentary party system. Heterogeneity at the individual level actually leads to homogeneity at the system level. A testable hypothesis for further research provided more data are collected is that defector MPs bring along resource mobilization capabilities (effective network size) occupying structural holes in the network.

As hypothesized by network theory, the complex system of multi-party politics in Romania seems to self-organize into a coherent two-party block, where party switchers play the role of integrators, proxies or facilitators of cross-party collaboration. Party switchers seem to play a distinctive role in the process of crystallization of the parliamentary party system from fragmented multi-party to a coherent two-party block. The extent to which they drive the process, or are themselves driven by the process, remains to be answered in a follow-up study where endogenous and exogenous factors involved in the process are taken into consideration.


We are thankful to the reviewers of the article, whose comments, questions and feedback have helped improve the quality of this work. Also, many thanks to the editor of the Special Issue, whose patience and comments have helped us move this paper further.

Author A.T. was partly supported by the Romanian National Authority for Scientific Research and Innovation (Unitatea Executiva pentru Finantarea Invatamantului Superior, a Cercetarii, Dezvoltarii si Inovarii), Project PN-III-P1-1.1-PD-2016-0193.


Bar-Yam, Y. (2002). Complexity Rising: From Human Beings to Human Civilization, A Complexity Profile, Encyclopedia of Life Support Systems, EOLSS UNESCO Publishers: Oxford, UK.

Blondel, V. D., Guillaume, J. L., Lambiotte, R. and Lefebvre, E. (2008). Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008(10), p. P10008.

Bratton, K. A., and Rouse, S. M. (2011). Networks in the legislative arena: How group dynamics affect cosponsorship. Legislative Studies Quarterly, 36(3), pp. 423-460.

Boccara, N. (2010). Modeling Complex Systems. Springer Science & Business Media.

Borgatti, S. P., Mehra, A., Brass, D. J. and LaBianca, G. (2009). Network analysis in the social sciences. Science, 323(5916), pp. 892-895.

Bozoki, A. and Ishiyama, J. T. (2002). The Communist Successor Parties of Central and Eastern Europe. ME Sharpe.

Carrington, P. J., Scott, J. and Wasserman, S. (eds.). (2005). Models and Methods in Social Network Analysis. Cambridge University Press.

Chiru, M. and Ciobanu, I. (2009). Legislative recruitment and electoral system change: the case of Romania. CEU Political Science Journal, 4(2), pp. 192-232.

Chiru, M. and Neamtu, S. I. (2012). Parliamentary representation under changing electoral rules: Co-sponsorship in the Romanian parliament. In Inaugural General Conference of the ECPR Standing Group on Parliaments, Parliaments in Changing Times, Dublin 2012, pp. 24-27.

Dal Maso, C., Pompa, G., Puliga, M., Riotta, G. and Chessa, A. (2014). Voting behavior, coalitions and government strength through a complex network analysis. PloS One, 9(12), p. 0116046.

Desposato, S. W. (2006). Parties for rent? Ambition, ideology, and party switching in Brazil's chamber of deputies. American Journal of Political Science, 50(1), pp. 62-80.

Desposato, S. W. (2004). The impact of party-switching on legislative behavior in Brazil. Unpublished manuscript, University of Arizona.

Fowler, J. H. (2006). Legislative cosponsorship networks in the US House and Senate. Social Networks, 28(4), pp. 454-465.

Gerber, E. R., Henry, A. D. and Lubell, M. (2013). Political homophily and collaboration in regional planning networks. American Journal of Political Science, 57(3), pp. 598-610.

Gherghina, S. and Jiglau, G. (2011). The ideological institutionalization of the Romanian party system. Romanian Journal of Political Science, 11(1), pp. 71 -90.

Gherghina, S. and Chiru, M. (2014). Determinants of legislative voting loyalty under different electoral systems: Evidence from Romania. International Political Science Review, 35(5), pp. 523-541.

Gherghina, S. (2016). Rewarding the 'traitors'? Legislative defection and re-election in Romania. Party Politics, 22(4), pp. 490-500.

Grose, C. R. and Yoshinaka, A. (2003). The electoral consequences of party switching by incumbent Members of Congress, 1947-2000. Legislative Studies Quarterly, 28(1), pp. 55-75.

Heller, W. B. and Mershon, C. (2005). Party switching in the Italian Chamber of Deputies, 1996-2001. Journal of Politics, 67(2), pp. 536-559.

Heller, W. B. and Mershon, C. (2008). Dealing in discipline: party switching and legislative voting in the Italian Chamber of Deputies, 1988-2000. American Journal of Political Science, 52(4), pp. 910-925.

Heller, W. B. and Mershon, C. (2009). Political Parties and Legislative Party Switching. Palgrave Macmillan.

Jacomy, M., Venturini, T., Heymann, S. and Bastian, M. (2014). ForceAtlas2, a continuous graph layout algorithm for handy network visualization designed for the Gephi software. PloS One, 9(6), p. e98679.

Janda, K. (2009). Laws against party switching, defecting, or floor-crossing in national parliaments. In World Congress of the International Political Science Association, Santiago, Chile.

Kirkland, J. H. (2011). The relational determinants of legislative outcomes: Strong and weak ties between legislators. The Journal of Politics, 73(3), pp. 887-898.

Kirkland, J. H. and Gross, J. H. (2014). Measurement and theory in legislative networks: The evolving topology of Congressional collaboration. Social Networks, 36, pp. 97-109.

Klein, E. (2016a). Electoral rules and party switching: how legislators prioritize their goals. Legislative Studies Quarterly, 41(3), pp. 715-738.

Klein, E. (2016b). "The personal vote and legislative party switching." Party Politics, 24(5), pp. 501-510.

Kreuzer, M. and Pettai, V. (2003). Patterns of political instability: Affiliation patterns of politicians and voters in post-communist Estonia, Latvia, and Lithuania. Studies in Comparative International Development, 38(2), pp. 76-98.

Lazer, D., Pentland, A., Adamic, L., Aral, S., Barabasi, A. L., Brewer, D., Christakis, N., Contractor, N., Fowler, J., Gutmann, M., Jebara, T., King, G., Macy, M., Roy, D. and Alstyne, M. (2009). Life in the network: the coming age of computational social science. Science, 323(5915), pp. 721-723.

Martin, S., Saalfeld, T. and Str0m, K. (eds.). (2014). The Oxford handbook of legislative studies. Oxford University Press.

McMenamin, I. and Gwiazda, A. (2011). Three roads to institutionalisation: Vote-, office-and policy-seeking explanations of party switching in Poland. European Journal of Political Research, 50(6), pp. 838-866.

Mershon, C. and Shvetsova, O. (2008). Parliamentary cycles and party switching in legislatures. Comparative Political Studies, 41(1), pp. 99-127.

Mershon, C. and Shvetsova, O. (2009). Timing matters: incentives for party switching and stages of parliamentary cycles. In Political Parties and Legislative Party Switching, New York, NY: Palgrave Macmillan, pp. 201-229.

Neamtu, S. I. (2011). The Effects of the Electoral Reform on Members of the Romanian Parliament: A Network Analysis. MA Thesis, Central European University.

Newman, M. E. J. (2006). Modularity and community structure in networks. Proceedings of the National Academy of Sciences, 103(23), pp. 8577-8582.

Noack, A. (2009). Modularity clustering is force-directed layout. Physical Review E, 79(2), p. 026102.

O'Brien, D. Z. and Shomer, Y. (2013). A Cross-National Analysis of Party Switching. Legislative Studies Quarterly, 38(1), pp. 111 -141.

Olson, D. M. (1998). Party formation and party system consolidation in the new democracies of Central Europe. Political Studies, 46(3), pp. 432-464.

Owens, J. E. (2003). Part 1: Cohesion: Explaining party cohesion and discipline in democratic legislatures: Purposiveness and contexts. The Journal of Legislative Studies, 9(4), pp. 12-40.

Page, L., Brin, S., Motwani, R. and Winograd, T. (1999). The PageRank citation ranking: Bringing order to the web. Technical Report, Stanford InfoLab.

Pinto, L. (2015). The time path of legislative party switching and the dynamics of political competition: The Italian case (1996-2011). The Journal of Legislative Studies, 21(3), pp. 323-341.

Richardson, L. (2017). Beautiful Soup Documentation.

Ringe, N., Victor, J. N. and J. H. Gross. (2013). Keeping your friends close and your enemies closer? Information networks in legislative politics. British Journal of Political Science, 43(3), pp. 601-628.

Shabad, G. and Slomczynski, K. M. (2004). Inter-party mobility among parliamentary candidates in post-communist East Central Europe. Party Politics, 10(2), pp. 151-176.

Stefan, L., Gherghina, S. and Chiru, M. (2012). We all agree that we disagree too much: attitudes of Romanian MPs towards party discipline. East European Politics, 28(2), pp. 180-192.

Tavits, M. (2005). The Development of Stable Party Support: Electoral Dynamics in PostCommunist Europe. American Journal of Political Science, 49(2), pp. 283-298.

Tavits, M. (2008). On the linkage between electoral volatility and party system instability in Central and Eastern Europe. European Journal of Political Research, 47(5), pp. 537-555.

Tavits, M. (2009). The making of mavericks: Local loyalties and party defection. Comparative Political Studies, 42(6), pp. 793-815.

Thames, F. C. (2007). Searching for the electoral connection: parliamentary party switching in the Ukrainian Rada, 1998-2002. Legislative Studies Quarterly, 32(2), pp. 223-256.

Volpi, E. (2018). Ideology and Party Switching: A Comparison of 12 West European Countries. Parliamentary Affairs, gsx079.

Wasserman, S. and K. Faust. (1994). Social Network Analysis: Methods and Applications. Cambridge University Press.

Appendix--Brief visual history of Romanian political parties, 1990-2020

Silvia I. Fierascu (1), Mihai Parvu (2), Alexandru Topfrceanu (3), Mihai Udrescu (4)

(1) Silvia Fierascu (Corresponding author) is a Research Fellow at the Department of Network and Data Science, Central European University, Hungary. She specializes in network science, political and policy networks, organizational development, and corruption. She can be contacted at

(2) Mihai Parvu is a Researcher at the Department of Computer and Information Technology, Politehnica University of Timisoara, Romania. He specializes in social network analysis, test automation, machine learning.

(3) Alexandru Toptrceanu is Assistant Professor at the Department of Computer and Information Technology, Politehnica University of Timisoara, Romania. He specializes in social networks analysis, complex networks, machine learning, network medicine, social psychology.

(4) Mihai Udrescu is Professor at the Department of Computer and Information Technology, Politehnica University of Timisoara and co-founder of Timisoara Institute of Complex Systems, Romania. He specializes in physics of computation, computer reliability, computer architecture, cyber-physical systems, complex networks, machine learning.

(5) For a schematic overview please refer to the Appendix.

(6) We do not code independents.

(7) Between 1990 and 2018, Romania had only one significant electoral system change in 2008, from PR with closed lists to a Mixed Member Proportional system with Single Member Districts. However, the literature covering the effects of this electoral reform in Romania on legislative behavior and party discipline have not found significant changes (Chiru and Neamtu, 2012; Neamtu, 2011).

(8) When calculated in the politicians' switcher scores, the formula for the party's lifetime is adjusted to account for the timing of each MPs' switch, not in the aggregate.

Caption: Figure 5. Average PSS evolution during the period 1990-2018. Election years start from 1992 and repeat every 4 years. We highlight with error bars the standard deviation of the average PSS

Caption: Figure 6. The PSS difference between two consecutive years, highlighting the dynamics of the PSS around government formation

Caption: Figure 7. Performance versus average switcher score of political parties

Caption: Figure 8. Overview of politicians' co-membership in the same party

Caption: Figure 9. Brief visual history of Romanian political parties, 1990-2020
Table 9. Calculated OPP (party performance) and average PSS (party
switcher score) for the relevant political parties in Romania

Rank   Political party      OPP     avgPSS

1      FDSN/PSDR/PDSR/PSD   2.295   1.654
2      PNL                  1.571   1.267
3      FSN/PD/PD-L/PMP      1.745   0.917
4      PUR/PC               0.583   0.894
5      ALDE                 1.750   0.705
6      PRM                  0.552   0.687
7      PUNR                 0.875   0.415
8      UNPR                 1.667   0.303
9      PNTCD/PNT            1.233   0.303
10     PP-DD                0.800   0.277
11     PAC                  0.571   0.147
12     PER                  0.241   0.081
13     PDAR                 0.375   0.012
14     UDMR                 1.241   0.000
15     USR                  0.750   0.000

Note: Data is sorted by highest avgPSS.

Source: Authors' own calculations

Table 10. Network measures over time

                     1990-1992   1992-1996   1996-2000   2000-2004

Network density      0.255       0.178       0.147       0.199
Avg. MP degree       217         204         153         203
No. of communities   10          9           9           7
Network diameter     4           6           5           4
Avg. path length     2           3           2           2

                     2004-2008   2008-2012   2012-2016   2016-2018

Network density      0.230       0.233       0.296       0.344
Avg. MP degree       251         239         307         276
No. of communities   6           7           5           6
Network diameter     3           3           3           3
Avg. path length     2           2           2           2

                     avg     stdev

Network density      0.235   0.059
Avg. MP degree       231     45
No. of communities   7       2
Network diameter     4       1
Avg. path length     2       0.331

Source: Authors' own calculations
COPYRIGHT 2018 Romanian Academic Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2018 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Fierascu, Silvia I.; Parvu, Mihai; Topirceanu, Alexandru; Udrescu, Mihai
Publication:Romanian Journal of Political Science
Article Type:Report
Geographic Code:4EXRO
Date:Jun 22, 2018

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |