Individual amending activity in the European Parliament committee system

ABSTRACT Individual amendments at the committee stage offer Members of the European Parliament (MEPs) a valuable tool to change legislation, prior to its passage at the plenary stage. And until a recent rule change, they were also one of the last meaningful ways for MEPs to self-select into the EP legislative process. Applying theories of legislative organization, internal positions, and differences in national institutional background, we use novel data to examine why MEPs proposed amendments to select EP committees during a one-year period between October 2015 and 2016. Our findings address a remaining gap on the individual determinants for legislative behavior in the EP, while also highlighting the effects of uneven prestige and contestation found in the EP committee system.

Whereas rapporteurs shepherd the committee's opinion on proposed legislation through to a plenary vote and must account for the diverse and sometimes rival preferences of the shadow rapporteurs and party group coordinators, individual MEPs are also provided with an opportunity to weigh in at the committee stage via legislative amendments. Importantly, this process has traditionally been removed from the direct control of the EPGs and the national parties. Remarkably, little is known about what drives individual MEPs to sponsor amendments at the committee stage, although these amendments may lead to substantial changes in the course of EP legislation.
Accordingly, we evaluate a variety of likely drivers for individual amendment activity. Drawing upon a combination of theoretically relevant influences from the individual, party, national, and committee levels, we collect data on all amendments tabled to the committees on Agriculture and Rural Development (AGRI), Economic and Monetary Affairs (ECON), Foreign Affairs (AFET), and Transport and Tourism (TRAN) during the second year of the 8 th EP, October 2015-2016. Importantly during this period, and prior to a change of rules in 2017 that limited some access to this procedure, any MEP could table an amendment to any committee, irrespective of their own membership on the committee in question.
First taking stock of research on committee work in the EP, we posit and test three sets of potential explanatory factors for amendment sponsorship -each derived from existing theories of legislative orientation, internal positions, and committee standing. Findings from our novel dataset provide empirical heft to an important individual power for MEPs to single-handedly seek influence in the course of the EP legislative process. Shepsle and Weingast (1987) encapsulate the importance of committees to the legislative process, viewing them as both creators and guardians of legislation from inception to final plenary vote. Strøm (1998) further develops this point within the context of European democracies, referring to parliamentary committees as 'privileged members' of a legislature that are able to draw upon multiple bodies of policy specialists and that are able to work in parallel, in order to coordinate the trading of both policy information and benefits, while also efficiently passing legislation.

EP committees as drivers of the EU legislative process
Within the EP, Ringe (2010) argues that the supranational body's 20 standing committees and two subcommittees are central to its ability to pass a large volume of highlyspecialized policies. What the general body of the EP lacks in terms of policy knowledge can be made up for at the committee level, whose reports make recommendations that are typically respected by the EP's final plenary votes. In addition to being 'repositories of policy expertise' (Ringe 2010, 20), the standing committees are also vehicles for the formation of a common line within the EP party groups (EPGs) and national delegations. In this regard, the EP resembles working parliaments like the German Bundestag that combine strong, specialized committees with a preponderant role for party groups. At the same time, there is some evidence that EP committees are also venues for agenda-setting and policy-shaping, through which external actors -such as national governments or interest groups -seek to advance their preferences (see, for example, Bouwen 2004;Marshall 2010Marshall , 2015McElroy 2008;Settembri and Neuhold 2009).
The impression of EP committees from the literature is therefore one of important legislative gatekeepers, but also of specialized policy workhorses and the occasional agent of partisan, national, or even external interest group principals. Within this chaotic and critical milieu, it is no wonder that Nikoleta Yordanova comes to the conclusion that EP committees are 'in need of a theory' (2011b). However, inasmuch as committees are the collective foot soldiers for EP policymaking, they are also one of the final remaining fora for MEPs to directly express their individual legislative preferences. This expression can take place via the tabling of legislative amendments.

Legislative amendments in the EP
Up until a recent rules change in January 2017, 1 any individual MEP was allowed to table an amendment to a Commission proposal during the EP's first reading, while that proposal was at the committee (as opposed to the full plenary) stage. 2 In the event of a second reading, amendments at the committee stage would be restricted to full members or substitutes of the committee responsible. 3 And during the plenary stage, amendments to EP proposals may typically only come from the committees that are responsible for the report (and would therefore require the acquiescence of the relevant rapporteur), from the EPGs themselves, or from groups of MEPs meeting a certain threshold (currently set at a threshold of 5% of all MEPs). Amendments to reports at the committee stage were thus a key moment for individual MEPs to directly weigh in on legislative proposals.
As Marshall (2010, 566) has observed, successful amendments at the committee stage by far outnumber successful amendments in later stages, which suggest that amendments can have a substantial impact on the committee report that is sent to plenary. If, as Coen, Lehmann, and Katsaitis (2020) put it, 'a legislative proposal passes [the committee] stage it is very likely to pass the plenary, making the committee's proposed draft resolution also the EP's final position' (p. 3). It is therefore somewhat surprising that the scholarly literature has given relatively scant attention to the dynamics of such amendments.
Most work on EP amendments examines either their successful passage or the fact that lobbyists influence MEPs' tabling of them. For example, Kreppel's (1999) work examines the success of EP amendments at an inter-institutional level and Whitaker (2011) notes the power of amendments as an important individual tool. Marshall (2010) examines how external actors choose who to target for amendment sponsors, while both Hurka (2013) and Baller (2017) trace the process of committee amendments that have been tabled. What is not taken up by the literature, however, are the micro-foundations of what drives an individual MEP to offer an amendment at the committee stage in the first place.

Theorizing the determinants of committee-stage amendments
We can distinguish at least two incentives to propose an amendment at the committee stage. Naturally, the most obvious is to shape the content of legislation. However, amendments -especially those that do not require much coordination or prior authorization -can also be used as a device to signal to principals (i.e. the party group leadership, national party, citizens) or interest groups a commitment to a particular policy position, dissent with the draft report and/or committee majority, or simply document legislative activity (Thierse 2015). A record of activism in tabling (substantively perhaps even unimportant) amendments may be enough to secure committee membership in the next term or renomination by the national party. For the purposes of our discussion, we examine possible determinants for committee-stage amendment sponsorship along three different strands of the comparative legislatures literature that we believe should have the most bearing on this impulse: theories of legislative organization and legislator experience, internal legislative positions, and committee differences within the EP itself. Yordanova's (2013) work on the EP committee system suggests that legislators may take on a 'distributive', 'informational', or 'partisan' characteristic. Building upon literature from the US Congress, she refers to those MEPs who have 'outlying preferences' far away from the median legislator as being 'high demanders' and most likely to carry out committee work in a way that is self-serving or based upon the unique preferences of their home districts. These outlying preferences may privilege special regional or sectoral interests, pork barreling, or coordinated legislative logrolling.

Legislative organization and legislator experience
Conversely, the 'informational' perspective views committee members as 'experts' in the policy substance of their committees who can specialize at low cost and translate general policy preferences of the parent chamber into substantive legislative proposals (Ringe 2010). Others also identify the interplay between the informational and distributive logics of legislative work and the ability of organized interests to sway MEPs at the committee stage (e.g. Bouwen 2004;Marshall 2010;Ringe 2010). If MEPs are influenced by either of these logics in their committee behavior, then we should see evidence of this in their tabling of legislative amendments.
Taking first the informational logic into consideration, we assume that EP committees are most likely to be driven by policy experts who are MEPs with either an educational or a professional background in the subject matter of their policies. Here, the logic is similar to Daniel and Thierse's (2018) finding that MEPs with relevant 'pure expertise' in the substance of their committees will be more likely to become party group coordinators on these committees. From this, we derive the expectation that MEPs with expertise in the policy subject matter addressed by a committee will also be more active in tabling amendments to it: H1. MEPs who have pure expertise in the subject matter of a given committee will be more active in tabling amendments to reports being taken up by that committee.
Returning to Yordanova's distributive logic, we might also assume that 'high demanding' MEPs -those whose previous ties to organized special interests related to the committee subject matter lead them to having 'outlying preferences' -will be more active in tabling amendments to that committee. This assumption is supported by Hurka's (2013) discussion of MEPs connected to green groups as 'demanding' of outlying policy options on the ENVI committee. It also has broader standing in the comparative legislatures literature, such as in Cross et al.'s (2019) finding of a 'strong parliamentary bias in favour of business interests' in the tabling of committee amendments to the German Bundestag (p. 19). Therefore, if MEPs with background connections to organized special interests are more likely to 'demand' specific policies, then we might expect that they will be more active in tabling amendments at the committee stage: H2. MEPs who have special interest ties in the subject matter of a given committee will be more active in tabling amendments to reports being taken up by that committee.
Legislative organization and experience may also come from previously held positions in national political life, as MEPs themselves come from a broad array of differing national political institutional traditions (Arter 2006). Given that MEPs come from 28 different national backgrounds and may have varying degrees of previous career experience, it may be reasonable to expect that MEPs who have had both previous legislative experience in national parliamentary systems -and who have had that experience in national parliaments with stronger committee organizations -will be more likely to project that previous experience onto their work in the EP and table written legislative amendments at the committee stage.
On the other hand, Pemstein, Meserve, and Bernhard (2015) suggest that the EP's unique legislative organization means that 'legislative experience is not necessarily transferable, and a background in the national legislature may not prepare a politician to be effective in the EP' (p. 1425). This suggests that MEPs with previous national experience in strong committee parliaments may not be more active in tabling amendments than their colleagues from other systems, as the professional requirements of the two jobs are not like for like. Treib and Schlipphak (2019) relatedly find that committee leadership positions are not more likely to go to MEPs with previous experience in national legislatures, just as van Geffen (2016) finds that former national MPs are less likely to be active in amending committee reports (because they are more likely to become rapporteurs, themselves). We therefore expect that previous experience in national parliaments is meaningful, but could affect amendments in one of two ways: H3a. Previous national legislative experience in parliaments with strong committee systems will positively influence an MEP's likelihood of tabling amendments at the committee stage.
H3b. Previous national legislative experience in parliaments with strong committee systems will negatively influence an MEP's likelihood of tabling amendments at the committee stage.

Internal EP positions
A third, 'partisan' characteristic of EP legislative organization set forth by Yordanova (2013) views MEPs as agents of their political parties (or in this case, EPGs). While this logic is supported by the classic literature on party groups' bidding for rapporteurships (see, for example, Kaeding 2004), it is also reasonable to expect that MEPs might rationally use the committee amendment procedure to highlight an individual position, when their EPG has not successfully 'won' the rapporteurship. In this case, we would expect to see MEPs whose EPG is not the rapporteur for a given piece of legislation to be more active in the levying of committee amendments: H4. MEPs whose party group affiliation differs from the rapporteur's for a given committee report will be more active in tabling amendments to that report.
A related point is that rapporteurs may themselves table an outsized number of amendments on legislative dossiers that they currently shepherd. Although these are likely to be 'friendly' corrections that take place during the committee process, they are nonetheless important to consider empirically alongside the two previous roles: H5. MEPs who serve as rapporteur on the committee report in question will be more likely to table amendments to that report.
Furthermore, MEPs whose group differs from that of the committee rapporteur may become the 'shadow' rapporteur and work to find consensus with other shadows and the main rapporteur, all whilst advocating for their party group interests. Häge and Ringe's (2020) recent study of shadow rapporteur selection indicates that MEPs are most likely to self-select into the role, based primarily upon a longstanding and demonstrated interest in the policymaking area of the dossier. Whereas shadows are often selected in consultation with group coordinators and other legislative positions, Häge and Ringe find that this process is much more 'bottom-up', which we view as compatible with the individually driven dynamics of tabling single-author committee amendments. Accordingly, we expect that: H6. MEPs who serve as a shadow rapporteur on the committee report in question will be more likely to table amendments to that report.

Committee-level considerations
Taken in concert, we might expect each of the above determinants to behave similarly across any and all EP committees. However, we also know that not all EP committees are created equally, particularly with respect to their degree of legislative consensus in passing proposed reports onto the full plenary (Settembri and Neuhold 2009;Winzen 2011), but also in terms of their perceived prestige by MEPs during the committee assignment process (Whitaker 2019).
While the EP is typically viewed as a consensus-driven body, on the whole, the increasingly politicized nature of the legislature does allow for more overt displays of partisan contestation. Accordingly, we might generally expect MEPs to use the amending process as a means of mitigating or indeed magnifying the partisan nature of the committee's policy debates. Combined with the first set of hypotheses on legislative organization in the EP, we also view a potential scenario for those committees in which the degree of internal contestation is particularly high (viz., where consensus is low).
For example, expert MEPs may have a heightened desire to table amendments to more contested committees if they feel a particular need to either 'course correct' for errant substantive positions of the committee rapporteur or to lend informational efficiency that is informed by their expertise. Similarly, MEPs who are preference outliers, MEPs from different party groups than the rapporteur, and shadow rapporteurs may each have additional incentives to table (public) amendments on legislation, registering dissent when legislation passes through divisive (and thus more politicized) committees.
H7. Individual-level effects of amendment sponsorship will be more pronounced on more contested committees.
The dynamics of contestation within the EP committee system are not only limited to partisan divides, but also to the perceived prestige that an assignment on each committee holds. As shown by Whitaker (2019), the process of committee assignment is unevenly contested, with committees like Foreign Affairs (AFET) particularly in demand, whereas others such as Development (DEVE) or Petitions (PETI) are rarely requested. Although Whitaker's broad survey of MEPs speaks to the mix of distributive, informational, and partisan goals sought after by MEPs in requesting a committee assignment, it also implies a perceived general hierarchy of prestige across each of the standing committees (see, for instance, the listing from Table 1, p. 173).
Broadly speaking, we observe the committees that are the most prestigious as those that can communicate clear legislative priorities or whose nature would be easily understood by constituents. These include both larger committees with less legislative power (such as AFET), alongside active committees that convey direct benefits to one's constituents (such as ECON). Less in demand, by contrast, are the narrowly focused committees (e.g. PECH or DEVE) and those whose purpose may not be obvious to external actors (such as PETI).
To develop upon Whitaker's (2019) own hypotheses, we anticipate that more broadly prestigious committees will attract amplified levels of MEP amendments. To summarize our expectations, therefore: H8. Individual effects of amendment sponsorship will be more pronounced on more prestigious committees.
Alternative explanations Naturally, a number of other factors may lead MEPs to be more active in the tabling of amendments. For instance, we might expect individual-level demographic differences to make MEPs more predisposed to tabling amendments. This might include, for instance,  (0) 7 (3) 28 (7) 11 (6) Source: Own calculations based on data from the European Parliament (2017) and Whitaker (2019: 173). Agreement index taken from Hix (2007).
relevant backgrounds in political work (such as office-holding at the national or subnational level). Similarly, Whitaker's (2001) and Finke's (2012) discussions of the EP committee players may make group coordinators more likely to offer further amendments. This logic may also apply to committee chairs, vice chairs, or even certain EPGs. Longer reports might also have the mechanical effect of demanding additional amendments, as a product of their complexity -just as reports taken up as a part of Ordinary Legislative Procedure (OLP, formerly referred to as 'co-decision') proceedings may systematically attract more amendments. We control for each of these eventualities, as well as other MEP demographic information, in our empirical analysis.

Case selection of EP Committees
Because of the huge volume of legislative dossiers and thus amendments in the EP, we deliberately choose to look at the tabling of amendments on four committees -AFET, AGRI, ECON, and TRAN -during a year-long legislative cycle that is removed from both internal leadership turnover (which would come at half-term) and the stress of an upcoming EP election (at end-of-term). Our committee selection also demonstrates variation along the two dimensions of committee consensus and prestige.
Looking at the average rates of committee contestation, we see that TRAN is a highly consensus-driven body, whereas ECON is quite contested. On the other hand, Whitaker's (2019) classification of committee prestige finds that ECON is quite prestigious, whereas TRAN is less so. AFET is somewhat contentious, while being the most prestigious, and AGRI is relatively moderate on both dimensions. We are therefore able to exploit variation in the combinations of prestige and consensus and address the underlying hypotheses discussed above.
A third factor in selecting these committees is in their usage of the OLP. Whitaker (2011) documents how three of these committees have been quite active in taking up OLP reports (ECON and TRAN in the top third and AGRI in the top half), whereas AFET is among the least likely to work on co-decision reports -which mirrors the EU's reduced competency in foreign affairs. In other words, we are able to consider variation in how seriously these four committees are taken at a policy level. The four committees also differ with respect to their legislative workloads. Whereas ECON and TRAN accounted for 13% and 11% of all co-decision files adopted in the first half of the 8 th EP, respectively, the share of AGRI was only 5%, while AFET handled no co-decision file as lead committee (see Table  1). 4 Finally, the committees differ with respect to the total number of dossiers which they have handled as lead committee: ECON leads both in terms of full workload and in terms of files subject to the OLP. While AFET is responsible for a relatively large number of files, none of them are among the most legislatively relevant.
In sum, we view the selection of the four committees as containing a mix of strengths, legislative consensus, and political prestige that should help us to gauge comparative committee-level dynamics in legislative decision-making. Table 1 summarizes the committee-level differences.

Data and method
We investigate the amending activity of individual MEPs by considering the separate number of single-authored amendments that a MEP has tabled for a particular draft report or opinion on each of the four standing committees included in our analysis (AFET, AGRI, ECON and TRAN) during a one-year period from October 2015 until October 2016. Whereas we also view co-sponsored amendments as critical to the legislative process, we expect the determinants of their sponsorship to differ considerably from the individual initiatives discussed above. Moreover, we expect the above-mentioned rules change to most crucially restrict the prerogatives of individual MEPs to table amendments at the committee-stage, reflecting a broader trend of curtailing individual legislators' rights in the course of the ongoing professionalization of the EP (McElroy 2007). The unit of analysis that we model is MEP-committee-report, as each of the committee considered multiple reports during the period of analysis and any MEP can theoretically table amendments to multiple reports on each of the four committees, throughout the year considered. 5 In order to create our dependent variable, we consulted the eMeeting website of the EP, 6 which documents the proceedings in the standing committees and contains records for all tabled amendments. For reasons of data curating and availability on the website, October 2015 was the earliest date for which information on the committee meetings was available. This is helpful, as our period of investigation thus precedes the latest change to the rules that restrict individual legislators' rights to table amendments. 7 Our approach in collecting data on committee proceedings means that we consider the respective committees both in their leadership and in their opiniongiving roles.
We operationalize the dependent variable as a count variable, which is heavily skewed towards 0, since many MEPs never tabled any amendments to committee draft reports or opinions. One obvious reason is that MEPs, although still formally allowed to do so during the period we considered, are less likely to table amendments to committees where they are not a member or a substitute. Accordingly, we employ a zero-inflated negative binomial (ZINB) regression modeling strategy, which allows us to distinguish between the processes that drive the decision to author any amendment at all and the determinants of authoring many amendments.
Put plainly, the model first considers full-and substitute-committee members of the relevant committee, which we consider to be the two groups that are least likely to table no amendments. The second and main stage of the model then considers the full complement of determinants for tabling amendments, in light of the previous process that expected some MEPs would be more likely to table none at all. The analysis therefore assesses the number of individual amendments tabled by an individual MEP to a particular report on a given committee, while assuming that non-members of committees will be most likely to table no amendments at all. 8 We restrict our analysis to those MEPs that were continuously seated throughout our period of investigation and for which sufficient biographical and background information was available. In total, our baseline sample includes 718 MEPs. 9 We retrieved most information from the personal webpages on the EP homepage, which also featured a CV template at the time of data collection. Since more personal details such as educational background or previous work experience are optional and are often not disclosed on the EP homepage, we also turned to other publicly searchable sources such as personal websites, websites of national parties, EPGs or national parliaments (for previous MPs), and Wikipedia.
We generate the variable Pure Expertise to reflect whether an MEP has acquired policy-relevant expertise in the jurisdiction of a committee as a result of vocational training, higher education or work experience. For the AGRI committee, we code MEPs in the positive direction if they hold a degree in agricultural science or were themselves farmers. For ECON, MEPs with a degree in economics or business administration or with work experience in the financial sector are coded as experts. For TRAN, previous employment in the transportation and telecommunications sector is considered as policy-relevant expertise. For AFET, we consider MEPs with a degree in international relations, with previous work experience in the diplomatic service of their home countries, professional experience in EU institutions or with international organizations as holding policy-relevant expertise. In line with H1, we expect expert MEPs to table more amendments.
Special interest tie is a dichotomous variable that captures whether an MEP has revealed affiliations with interest groups or has a special interest because of previous employment in the management of a company or the running of a business. Thus, MEPs becoming involved in the affairs of the AGRI committee are coded as having a special interest if they have previously owned a farm or if they have ties to farming associations that promote the interests of (conventional) agriculture or, conversely, if they entertain ties to environmental groups that may advocate a reform of the agricultural sector. For the ECON committee, MEPs who have run a business or have served in the management of (large) firms are coded as having special interest ties. In line, with H2, we expect MEPs with special interest ties to be preference outliers and to table more amendments. In line with Yordanova's (2009: 256) distinction between information-driven and interest-driven committees, we did not code the special interest tie variable for AFET and TRAN. 10 To consider the mitigating effect of national parliamentary culture, we interact two variables. First, committee strength is measured by an index proposed by André, Depauw, and Martin (2016) that captures the degree of specialization in the committee structure, the capacity of parliamentary committees to acquire information, and their prerogatives to set and shape the legislative agenda. Higher values correspond with more powerful legislative committees in national parliaments. Second, we use a dummy variable that measures whether or not an MEP held a mandate in a national parliament before serving in the EP. MEPs coded as zero do not have the requisite national experience to have been influenced by a strong committee structure at the national level and MEPs coded as greater than zero reflect those politicians with a combination of both their national experience and the strength of parliamentary committees in their national legislatures. The interacted term forms the basis for our test of H3 and the constituent terms are left in the model as additional controls.
In terms internal positions held at the committee level, we use a dummy that takes the value 1 if both the particular MEP in question and the rapporteur on the given committee report came from the same EPG and 0 if their EPG affiliation differs to test H4. We then collect information on the (shadow) rapporteurs for each legislative report considered and code MEPs with 1 that served as the (shadow) rapporteur on that particular report to test H5 and H6.
To test the variation in committee contestation assumed by H7, we use the agreement index proposed by Hix et al. (2007, 91) 11 to calculate the general level of divisiveness on each committee, which is displayed in Table 1. For H8, we rely upon Whitaker's (2019) ranking of committee assignment preferences, where AFET is viewed the most prestigious of 18 committees, ECON is third most prestigious, and TRAN is seventh and AGRI is eighth.
We control for additional individual factors that may impact on the extent of amendment sponsorship, including: the MEP holding a formal position in the committee hierarchy (i.e. status as committee chair, vice-chair, or group coordinator, respectively), having drafted an opinion by a committee involved in the respective dossier, membership as a full or substitute member on the committees examined, and for the length (measured in years) that a MEP has already served in the EP. We account for the potential trade-off between tabling single-and co-authored amendments by controlling for the number of amendments that an MEP has sponsored jointly with his or her colleagues -which we view to be driven by a more complex set of dyadic factors that differ from the individual interests of amendment sponsorship. We also control for MEP age and gender. Finally, we control for the size of the national delegation of an MEP at the beginning of the 8 th term to proxy for the majority or minority status of an MEP within his or her EPG. At the dossier and committee level, we control for whether or not the dossier in question was subject to the OLP -which may raise the stakes of the legislation and thereby merit additional amendments, whether the respective committee handled a dossier as lead committee or simply delivered an opinion, and the number of opiniongiving committees. 12 We retrieve this information, along with details on rapporteurships, from the EP Legislative Observatory (OEIL). 13 Since longer pieces of legislation are likely to be subject to more amendments, we control for the length of the proposal as a logged word count. 14 Analysis Our analysis features separate models for each committee. The interpretation of the main results is based upon a model specification that includes fixed effects for EPGs and assumes that MEPs who are full or substitute members of the committee in question will be far less likely to propose zero amendments. Table 2 reports the main results from the ZINB regression models. To preview our findings, we find conditional support for many of our hypotheses; however, this support does vary by committee.
We begin by assessing the binary logistic regression that feeds into the main model. Here, we see very strong and consistent evidence that MEPs who are on a given committee (either as a full or substitute member) are significantly less likely to propose no amendments to that committee legislation. In other words, even though the EP changed its rules in 2017 to formalize this restriction between membership and the ability to propose amendments at the committee stage, it is apparent that this was already consistent practice. MEPs tend only to table amendments to committees on which they serve.
Moving to the core results from the count portion of the model, H1 assumed that MEPs with pure expertise relevant to the committee where a dossier was being discussed would be more likely to propose amendments. Our hypothesis finds support, but only within the ECON committee (model 1), where the effect of the pure expertise variable reaches conventional standards of statistical significance (p < 0.05). This may be related to more MEPs having expertise in economics than in the other fields considered. 15 Similar, support is found for MEPs with special interest ties (H2), who are more likely to propose amendments to ECON. Figure 1 plots predicted average marginal effects for key dichotomous variables in Table 2, using the package developed by Jann (2014). With respect to H1 and H2, we can more clearly assess the substantive effect of pure expertise and special interest ties, which initially appears quite minimal. However, it is worth noting that the mean predicted number of single-authored amendments tabled on the ECON committee is 0.177, so this indicates that MEPs with pure expertise and/ or special interest ties relevant to ECON dossiers are tabling about 65% more single-authored amendments than MEPs who do not. This effect is similarly amplified for MEPs with special interest ties related to ECON, albeit with a wider confidence interval. In tandem, this suggests that MEPs with both expertise and special interest ties in the economic sector are significantly more likely to participate actively in the ECON committee amendment process. However, it is important to note that the effect of these variables also varies between committee members and non-members. Consider a hypothetical MEP with special interest ties and no formal position as (shadow) rapporteur, tabling an amendment on a file subject to the OLP. Having policy-relevant expertise increases the predicted number of amendments by roughly 0.192 for full members, but only by 0.002 for non-members, a difference that is statistically significant at p < 0.01. This indicates a differential effect for expertise on committee members.
With respect to the other dichotomous predictors of amendment sponsorship, we find a highly significant effect (p < 0.01) for MEPs from the same EPG as the committee rapporteur to table more amendments to the ECON committee -although this runs counter to our expected direction in H4. One possible explanation is that party group colleagues may anticipate a higher chance of their amendments to be taken on board by 'their' rapporteur. Alternatively, it may be that individuals from the same EPG as the rapporteur have lost the bid for the position and therefore have an increased interest in tabling amendments to that particular dossier. This assumption is also in line with H6, which indicates that shadow rapporteurs are clearly more likely to table amendments within the ECON committee -perhaps because, in this case, their EPG failed to secure the rapporteurship. This could also be viewed as in function of ECON's heightened level of contestation (as in H7).
Turning to continuous predictors of amendment sponsorship, we use Royston's (2013) package to plot the interacted effect for national parliamentary strength and previous national MP experience, as discussed in H3. Here, the aggregate coefficient value for the interaction in Table 2 suggests that there may be a significant difference in trend between those MEPs with and without previous national MP experience. Plotted marginal effects in Figure 2 suggest that national parliamentary strength may affect legislators in different ways, based upon their national political experience. While the wide margins and unclear trends in the right panel suggests very little in the way of conclusive evidence for H3, the left panel does seem to indicate that MEPs from countries with a culture of strong parliamentary committees do trend towards being more active on EP committeeseven when they have no history in national parliaments themselves. 16 This muddled result suggests the need for additional, qualitative consideration.
Moving to H5 and H6, we find consistent and highly significant results for shadow rapporteurs to table more amendments within the ECON, AFET, and TRAN committees. Rapporteurs are also significantly more likely to table amendments to AFET and TRAN. This could indicate the presence of more formalized hierarchies of leadership within these committees -unlike in AGRI where none of the variables relating to rapporteurships (H4-H6) were shown to be significant.
Looking comparatively at the four committees, we notice from Table 1 that ECON is among our more contentious (H7) and prestigious (H8) EP committees; it is also among the most consistent in its results. This may indicate that individual MEPs have an increased incentive to propose amendments to this committee in the expectation of influencing the content of policy proposals and showcasing this activity to voters. The same trend does not hold true for AFET, where very few of our hypothesis are supported by the data. This may also have something to do with service on AFET itself being more of an empty signal to voters, given the relative lack of OLP reports that pass through it.
Looking at less contested and less prestigious committees, we note that our key predictors for the effect of individual backgrounds are mostly insignificant. This suggests indirect support for the amplifying effect of H7 and H8, given the positive and significant result on the more contested and prestigious ECON committee. We also note an interesting result on AGRI that relates to co-sponsored amendments. MEPs who were the most active in co-sponsoring amendments to AGRI reports are routinely the least likely to table single-authored amendments. This relationship is plotted in Figure 3 and suggests further nuance to explore between those MEPs who are 'team workers' and those who 'go it alone.' In light of how the four committees fall in somewhat neat pairs of high prestigelow consensus/low prestige-high consensus, it is also important to note that observational equivalence makes it difficult for us to infer exactly which dimension of the committee quality is driving this behavior.

Conclusion
Our analysis confirms that MEP backgrounds -both along partisan and individual lines, as well as legislators' internal positions -do have bearing on their propensity to table single amendments to legislation at the committee stage and especially for those committees on which they serve. This underlines the importance of committee amendments, as suggested by Coen, Lehmann, and Katsaitis (2021). However, our findings also vary along committee lines.
First, we find tentative support for the hypothesis that expert MEPs typically offer up more single-authored amendments at the committee stage. However, this finding is only related to ECON dossiers and may be due in part to the large number of MEPs with formal expertise in economics. Preference outliers who have demonstrated special interest ties to the substance of the committee at hand are similarly more likely to offer amendments to ECON, which does suggest that particular committees may make better targets for outside interests to lobby MEPs. In terms of the partisan theory for legislative behavior and the meaningfulness of previous national experience, we also find the most support for these theories on ECON dossiers, although not always in the expected direction.
More broadly speaking, we note that our findings are oftentimes highly conditional on which committee we are investigating. Although the systemic factors of (shadow) rapporteurs consistently drive amendment sponsorship, co-sponsorship only has a negative (albeit very strong) effect on amendments tabled to the more technical and niche AGRI committee. In line with the spirit of H7 and H8, the prestigious and contentious ECON committee contains more discernable drivers to sponsorship than any of the others. We hope that our meso-level approach strikes the balance between Hurka's (2013) case study and Baller's (2017) more global picture.
Our results also indicate that both full and substitute committee membership significantly discriminates between those MEPs who never table amendments at the committee stage and those who sometimes do. This finding corresponds to extant research on amending activity in the EP committee system offered by Marshall (2010): within committee hierarchy, it is usually a fairly small circle of MEPs -not necessarily made up of legislators with a formal position of authority -who account for the bulk of the work. Our findings empirically substantiate the considerable difference in impact between these 'active' and 'inactive' legislators.
Finally, our findings also speak to the shifting nature of inter-institutional collaborations in the EU policymaking process. While some have stressed the importance of the trilogue system between the Commission, Council and EP for reaching an informal consensus on legislative proposals (see, for example, Delreux and Laloux 2018; Ruiter 2020), others view the rise of trilogues as problematic for the already-weak lines of democratic representation in the EP (Rosén and Stie 2020). Our findings somewhat assuage these fears. Even prior to a rules change that limited the ability of MEPs to weigh in on committee proposals, we show that most MEPs were already very unlikely to table amendments on committees where they were not a member. And while expertise and external connections increased the amending activity of members, the bulk of amending takes place around those already involved (directly or indirectly) in the trilogue process as (shadow) rapporteurs. Individual amendments retain an important link, both to the representation of MEP individual interests, as well as to the influential trilogue process.
In sum, MEPs have differing incentives to table amendments, both within and across committees. Future research can address this calculation with the additional consideration of co-authored amendments, which we find appears to have a set of different drivers. In addition, our analysis has only been able to consider the number of amendments, without a deeper examination of their content. Future research could draw on quantitative text analysis to make more informed inferences on the content and quality of amendments that are tabled at the committee stage.
While it is clear from our analysis that amendments offer MEPs an important venue for individual input into the legislative process, it also appears that this may be yet another example of the EP as being 'in need of a theory' (to borrow from Yordanova 2011b). That said, future work on the EP committees should continue to look for theories that work conditionally on certain committees, rather than broadly generalizable claims. Notes 1. Rule 208 (1), EP Rules of Procedure in the version of July 2014, changed such that committee full members or substitute members must sign -and thus officially sponsor -all amendments for consideration in a given committee (see Rule 218 (1) (3). 4. On a side note, our committees also vary by their share of first-reading agreements, which reflects both differing degrees of inter-institutional conflict about policy and distinct cultures to 'fast-track' legislation. ECON has concluded a stunning 95% of its legislation during first reading, while this figure drops to 71% for AGRI and 55% for TRAN (European Parliament 2017: 11). 5. Appendix A in the supplementary materials displays the distribution of amendments across each committee examined. 6. http://www.emeeting.europarl.europa.eu/ 7. What is more, since our dataset covers a period in the middle of a legislative term, we are confident that our figures on amending activity are not biased upwards, which might be the case towards the end of a legislative term -when the EP may be eager to close a greater number of (legislative) dossiers and MEPs table amendments in preparations for an electoral campaign. 8. As the amendments offered by MEPs are nested in committee reports, we encounter repeated measures on our units of observation. This implies that we can neither assume the residuals to be uncorrelated nor to follow a normal distribution. In addition, unobserved between-group heterogeneity may bias the regression point estimates and standard errors. Hence, we use robust standard errors, clustered by individual MEP. 9. André et al.'s (2016) dataset on committee strength excludes Bulgaria, Croatia, Cyprus, and Romania from their analysis and further restricts our MEPs, as reported in Table 2. 10. While it is possible that these committees generate some targeted externalities, special interest ties in foreign affairs and/or transportation and telecommunications will likely be heavily correlated with the biographical determinants of MEP expertise on those committees. In that sense, these background traits are already subsumed into our 'pure expertise' indicator. 11. The agreement index (AI) is calculated as follows: , where Y i is the number of yeas cost in vote i, N i is the number of nays, and A is the number of abstentions. The index ranges from 1 (perfect unity) to 0 (tie between the three voting options). Our indicator is computed as the mean of agreement index scores per final committee vote for each committee in our period of investigation. See also Table 1. 12. Appendix B in the supplementary materials displays the distribution of files according to procedure across the four committees analyzed. 13. Committee reports that are debated and voted in plenary contain information can be accessed via the EP homepage and are conveniently linked to the OEIL documentation gateway. The reports also include vote results in the responsible and opinion-giving committees, which were used to calculate the agreement index. 14. For files subject to the budgetary (BUD), ordinary legislative (COD) and discharge (DEC) procedures, we used the respective Commission initiative as the basis for our calculations. For a handful of resolutions on topical subjects (RSP) that were handled by the AFET committee, we treated Council conclusions as the relevant proposal. 15. Refer to Appendices D through G in the supplementary materials for summary statistics from each committee. 16. It is worth noting here, however, that one of the strongest national parliaments in the dataset is Germany -whose MEPs do have an outsized impact in EP legislative behavior.