Multilevel Venue Shopping Amid Democratic Backsliding in New European Union Member States

Recently, various Central and Eastern European countries have experienced a regression of democratic quality, often result‐ ing in the emergence of competitive (semi‐)authoritarian regimes with an illiberal governing ideology. This has often been accompanied by a closing political space for civil society groups. Based on a survey of more than 400 Polish, Hungarian, Czech, and Slovenian interest organizations, we explore, in the context of backsliding, the conditions under which orga‐ nized interests shift their lobbying activities to alternative, i.e., EU or regional levels. Our statistical analyses indicate that it is rather exclusive policy‐making in general than a lack of individual group access to domestic policy networks that motivate organizations to engage in multilevel lobbying. However, it appears that organizational self‐empowerment and inter‐group cooperation are the “name of the game.” Even under the adverse conditions of democratic backsliding, organizations that are accumulating expertise, professionalizing their operations, and cooperating with other organizations not only can sus‐ tain access to (illiberal) national governments but also branch out their operations to the European and regional levels.


Introduction
Recently, democratic quality has declined in parts of Central and Eastern Europe (CEE).Semi-authoritarian regimes with an illiberal governing ideology have emerged.This has often been accompanied by what Sata and Karolewski (2020) call "caesarean politics": Instead of promoting pluralist democracy, such governments may engage in patronage by deliberately allocating rewards to political allies and punishing and/or demonizing political adversaries.This is accompanied by state monopolization of public institutions and media, but often also by privileging informal contacts over formal interest intermediation forums (Labanino, 2020).The "battle against the establishment" proclaimed by various national-conservative populist gov-ernments has often resulted in further political centralization.Such developments are particularly pronounced in Hungary and Poland (Bozóki & Hegedűs, 2018;Tworzecki, 2019).
So far, though, the ramifications of democratic backsliding for civil society, in general, and specific civic organizations remain unclear.Olejnik (2020) argues that various CEE governments are characterized by elaborate systems of rewards and punishments, the predominance of informal networks, and state capture.Moreover, illiberal governments are engendering their own new illiberal civil societies (Ekiert, 2019;Greskovits, 2020).Instead of "hollowing out" civic activism through overt oppression, governments are re-engineering civic organizations aligned with authoritarian-nationalist objectives to generate grassroots support.Indeed, the literature on civil society and organized interests in the region has boomed recently (Dobbins & Riedel, 2021;Novak & Fink-Hafner, 2019;Rozbicka et al., 2020).Scholars have covered, among other things, population ecologies (Rozbicka & Kamiński, 2021) and advocacy strategies (Czarnecki, 2021) while also exploring state-interest group interactions (Olejnik, 2020;Ost, 2011) and lobbying regimes (Vargovčíková, 2017).Yet we still know little about how individual interest organizations are navigating the new environment.A recent contribution by Pospieszna and Vetulani-Cęgiel (2021) indeed demonstrated that disadvantaged Polish interest groups are coalescing and doubling down on their networking strategies to navigate the increasingly authoritarian context.Yet the closure of the political opportunity structure (POS) might prompt groups to withdraw from policy-makers altogether (Gerő et al., 2020).
Against this background, we explore how the closing political space affects the relative importance of the levels organizations lobby on.Is there a shift from the national to the regional and/or EU level?Multilevel venue shopping across Western EU member states has previously been addressed by Beyers and Kerremans (2012), while Poloni-Staudinger (2008) explored how the relative openness of the domestic POS affects EU-level lobbying.We move beyond these accounts, however, by factoring in the reality of democratic backsliding both at the contextual as well as organizational and interorganizational levels.
Our study sheds light on how the changing political environment has pushed organizations into a mode of "defiant responsiveness" through strategic choices on where to lobby (national vs. EU or regional levels), as well as by enhancing their internal and external capacities.Thus, we provide new insights into the responsiveness of organizations in a region historically characterized by weak civil society and distorted patterns of political participation.The centerpiece of the analysis is a survey of more than 400 interest groups operating in the healthcare, energy, and higher education sectors in Poland, Hungary, the Czech Republic, and Slovenia.
In the next section, we set up the theoretical framework and the hypotheses.In Section 3, we describe the sample and survey underpinning the analysis and explain the research design in detail.Section 4 contains the descriptive and statistical analyses.Section 5 discusses the results and future research avenues.

Theoretical Framework and Hypotheses
One of the most basic strategic choices for an organization is in which venues and at what level (e.g., regional, national, international) it lobbies.Particularly in multilevel polities such as a federal state or the EU, this choice is affected by contextual factors such as the system of national (that is, member state) interest representation or the general openness of the political system towards societal interests (Beyers & Kerremans, 2012;Poloni-Staudinger, 2008).European integration has opened a multitude of transnational lobbying opportunities (Mahoney & Baumgartner, 2008).However, Europeanization also strengthened the regional level, even in unitary states.The economic, administrative, and entrepreneurial capacities of subnational territorial entities have gained significance, as most CEE countries have undergone substantial territorial reforms (Pitschel & Bauer, 2009).Europeanization and regionalization processes thus offer interest associations the opportunity to assert their demands in alternative venues.
Indeed, several authors argue that a compensation logic is at work: If groups are constrained in articulating their preferences in an institutional context, they will seek to do it elsewhere (Baumgartner & Jones, 2009;Guiraudon, 2000).Another perspective, however, emphasizes that European integration reinforces the national constellations and rewards strong and influential groups (Eising, 2007).That is, those groups with strong access to national policy-making venues and abundant organizational and financial resources will also be the ones engaging in multilevel lobbying in EU institutions.
Testing different multilevel governance accounts, Beyers and Kerremans (2012) show that the two perspectives are not necessarily mutually exclusive.Whereas organizational resources and strong access to national policy-making venues are indeed among the best predictors for multilevel venue shopping, the structure of the national polity and system of interest representation also matter.Interest groups from federal and/or neocorporatist states are more likely to engage in EU-level lobbying.Beyers and Kerremans (2012) show that national-level exclusion does not necessarily induce a flight to the EU level.However, the compensation hypothesis cannot be entirely rejected.Interestingly, higher proximity to peripheral (as opposed to relevant) opposition parties-as a proxy for a group's distance from the political mainstream-leads to more extensive multilevel venue shopping.Indeed, in her study of French, German, and UK environmental NGOs, Poloni-Staudinger (2008) found that groups use EU action to bypass undesirable national conditions.As national POS closed, groups were more likely to target their activities to the EU.
Democratic backsliding can also be conceptualized as a closure of the POS as the political space for independent organizations may shrink.Illiberal incumbents may restrict funding and harass interest groups opposing their policy agenda (Bromley et al., 2020;Buyse, 2018;Carothers, 2016) and weaken or abolish formal interest intermediation mechanisms (Olejnik, 2020).Hence, the closure of the POS associated with backsliding forces organized interests to recalibrate their lobbying strategies.
Following these theoretical considerations, our first hypothesis looks at a set of organizational factors explaining venue shopping.Financial and organizational resources were found to be among the most important predictors of accessing the EU level (Bernhagen & Mitchell, 2009;Eising, 2007).Having professional staff, trained lobbyists, and fundraising capacities, that is, being a professionalized organization is key to access to policy-makers (Albareda, 2020;Beyers, 2002).H1: The more organizational resources a group possesses, the more likely it engages in lobbying at every level (regional, national, EU).
However, Beyers and Kerremans (2012) found that relational aspects such as gaining access to domestic policymakers are a strong predictor of interest groups' political strategies).According to this "persistence hypothesis,'' the skills that organizations invest in and accumulate by lobbying at one level may boost their ability to expand to other areas (Beyers, 2002).Moreover, personal ties cultivated at one level may better enable organizations to branch out to other levels.At the same time, the representativeness and legitimacy which organizations gain by accessing the national level may make policy-makers at alternative levels more receptive to interactions with them (Dür & Mateo, 2014).H2: The greater the access interest groups have to domestic policy-makers, the greater are their regional and EU-level activities, too.
It is also conceivable, however, that organizations unable to exert influence in a closing political space may compensate by turning to the regional or EU levels (Beyers & Kerremans, 2012;Poloni-Staudinger, 2008).
H3: Interest groups excluded from national-level policy networks turn to the regional and EU levels.
Conceptualizing backsliding at the organizational level is everything but straightforward (see Section 3.2).Lacking access to national policy networks is not necessarily a sign, let alone a measurement, of backsliding.The closure of the POS and diminishing access are, of course, related phenomena, but they cannot be naively treated as the two sides of the same coin.Therefore, we will employ several variables to test the effect of backsliding differently than access.H4: Interest groups under pressure from illiberal incumbents will intensify their regional and EU-level lobbying activities as opposed to the national level.
We test several control variables, which should alleviate the effects of resources, access, and backsliding at the organizational level, but simultaneously also affect multilevel lobbying strategies.In line with the findings of Beyers and Kerremans (2012) for the supranational level, we expect neo-corporatism to have a positive effect on both EU and regional level lobbying, Moreover, intergroup cooperation is a key strategy organizations can employ to reduce "environmental uncertainty" (Pfeffer & Salancik, 2003).Hanegraaff and Pritoni (2019) find that groups with decreasing influence and at risk of organizational failure are more likely to cooperate with other groups.We also test domestic networking and national and EU umbrella organization memberships and expect to find a positive relationship between intergroup cooperation and national and EU-level lobbying.Expertiselegal, economic, technical/scientific-is also a significant exchange good for interest groups for access and influence both at the national and EU levels (Bernhagen, 2013;Chalmers, 2011;De Bruycker, 2016).Interest groups not only may disburden policy-makers from the complexities of all the pieces of legislation they simultaneously juggle but, as voices of civil society, may also provide information on how their constituents are affected by policies (Bouwen, 2002).However, expertise provision might be an additional resource for already privileged organizations, particularly in the context of backsliding.Indeed, Horváthová and Dobbins (2019) found that, in Hungarian energy policy, the government relies heavily on the expertise of numerous preferred business groups with frequent access, while opponents of governmental policy are largely excluded.We also include organizational longevity as a control, as it is treated in lobby research as a decisive factor explaining access and influence (Kohler-Koch et al., 2017).

Sample and Data
First, we applied Eising's (2008) definition of interest groups while identifying relevant organizations.He defines three attributes of interest groups: organization, political interest, and informality.They strive to "influence policy outcomes….Political interest refers to attempts…to push public policy in one direction or another on the behalf of constituencies or a general political idea" while "informality relates to the fact that interest groups do not normally seek public office but pursue their goals through informal interactions with politicians and bureaucrats" (Eising, 2008, p. 5).Based on these criteria, we compiled population ecologies of all Czech, Hungarian, Polish, and Slovenian healthcare, higher education, and energy policy organizations currently active on the national level.The three selected policy areashealthcare, higher education, and energy-are diverse and not interrelated, thus increasing the generalizability of the findings.All represent a large portion of public budgets and include both public, non-state, and business interest groups.
As a rule, we collected data from public registries of civil society organizations.We cross-checked the data with internet searches and lists from parliaments and different ministries that invited organizations to various committees and meetings.We identified a total of 1,345 interest organizations on the national-level active as late as 2019 and conducted an online survey targeting the active organizations between March 2019 and May 2020 in the four national languages.The multiple-choice survey questions addressed their interactions with parties, parliaments, regulatory authorities, and policy coordination with the state.We sent 1,264 invitations and received 427 responses, which corresponds to a 33.7% response rate (with strong country variations: Slovenia 51.8%, Hungary 35.3%, Czech Republic 33.6%, Poland 24.6%).Organizations perceived as critically important (e.g., large students' organizations, labor unions, medical chambers, and energy business organizations) were contacted more intensely and nearly all responded.For a detailed description and the two datasets see Dobbins, Labanino, et al. (2022).
The country selection also controls for the system of interest representation and lobbying regulation.Slovenia is a model neo-corporatist state (Jahn, 2016) and arguably CEE's most coordinated market economy (Bohle & Greskovits, 2012), but lobbying regulation is weak.The Czech Republic exhibits a very weakly regulated market economy and lax lobbying regulations (Šimral, 2015).The Polish economy is also relatively weakly coordinated.However, extensive lobbying regulations exist, which may stymie the influence of interest groups (McGrath, 2008).Hungary exhibits stronger market coordination (Duman & Kureková, 2012), and elections are publicly funded, lobbying activities, however, have only been loosely regulated since 2010 (European Commission, 2020;Laboutková et al., 2020).
The four countries also represent different levels of democratic backsliding.Hungary and Poland, but arguably also the Czech Republic and Slovenia, have embraced authoritarian governance styles (Bozóki & Hegedűs, 2018;Hanley & Vachudova, 2018;Przybylski, 2018).The 2021 Nations in Transit report by Freedom House reveals that Hungary and Poland underwent the steepest decline in democratic quality ever recorded (Csaky, 2021).Hungary was downgraded first from a stable to a semi-consolidated democracy and in 2020 to a "transitional hybrid regime."Freedom House still rates Poland as a semi-consolidated democracy.Although still considered a consolidated democracy, the quality of Czech democracy declined somewhat under the Babiš government (Buštíková, 2021).Democratic quality also deteriorated in Slovenia under Janez Janša's premiership since 2020 (Lovec, 2021).Nevertheless, our survey period mostly preceded his government forming (and only in a few cases coincided with the first weeks of his being in power).However, during 2021 and 2022, voters resoundingly ousted both the Babiš and Janša governments, whereas Viktor Orbán retained his constitutional majority for the fourth time since 2010.
While we cannot thoroughly engage here with the country-specific complexities of the regionalization of fiscal, administrative, and policy-related competencies, it is safe to say that all four countries emerged from the socialist era as unitary states but then experienced a profound strengthening of subnational governance, not least due to Europeanization processes.
While Poland and the Czech Republic led the way (Yoder, 2003), Hungary (Temesi, 2017) and Slovenia (Setnikar-Cankar, 2010) also experienced a shift towards more regional autonomy.Regionalization was generally driven by regionalist political movements, the devolution of social services by the central state, and often pressures from wealthier regions for more economic control (Yoder, 2003).The European Commission also strongly pushed for a governance model based on autonomy for subnational entities in order to receive EU structural funds (Pitschel & Bauer, 2009).However, backsliding meant a significant re-centralization of governance structures and policy competencies (Antal, 2019; for healthcare, see Mikuła & Kaczmarek, 2019;Szigetvári, 2020; for energy, see Szulecki, 2020; Temesi, 2017; for higher education, see Vlk et al., 2021).Despite this, there is still a strong argument for subnational activities of organized interests.First, healthcare and higher education services are, by nature, provided at the regional and local levels, while major sources of energy are often derived from peripheral regions (e.g., the Polish coalmine basin in Silesia, the Temelín nuclear reactor in southern Bohemia).Yet, more importantly, EU cohesion funds are generally distributed by subnational authorities, meaning that-despite different countryspecific polity-related dynamics-regional authorities remain potent actors and potential points of access for organized interests despite centralization trends.

Research Design
Our dependent variables are based on a survey item asking respondents to evaluate how the importance of the following levels of representation has changed compared to 10-15 years ago: the regional, national, and EU-levels.The respondents gave their answers on a five-point scale for each level ranging from 1 (much less), 2 (less), 3 (the same), 4 (more), and 5 (much more).We estimated models for all three separately.However, we also created two new variables expressing the relative importance of the national and the EU and the national and regional levels for representation, respectively.We recoded the EU and the regional level variables: much less importance takes the value −2, less −1, the same 0, more +1, and much more +2.The variable on the importance of the national venue is coded as a mirror image of the EU and the regional variable.That is, if the importance of the national level for an organization has increased much more, it takes the value of −2; if it increased more, it takes the value −1; if it became less important, it takes the value +1; if it became much less important, it takes the value +2; finally, if it is the same, it takes the value 0. We summed up the recoded national and EU and the national and regional variables and created two new variables: National vs.European levels and National vs. regional levels.These two new variables range from −4 to +4, negative values indicating that the national level has gained in importance relatively to the EU or regional levels, whose importance did not change or decreased.Positive values indicate the opposite: The EU or regional levels gained importance relative to the national level.The value 0 indicates that the national level gained in importance exactly as much as the EU or the regional levels lost their importance, respectively, or that there was no change (Table 1).This operationalization highlights the relational importance of the three representational levels and grasps the move in relative terms towards or away from the national level to either the regional or EU levels.
Access to policy-makers is a composite variable of five survey questions measuring interest group access to different venues.Our respondents rated the difficulty of accessing regulatory authorities, parliamentary committees, governing parties, opposition parties, and a general assessment of policy coordination between the state.The questions are measured on a 1 to 5 scale, 1 being no access or extreme difficulties in access, and 5 being full access or very easy access.We summed the scores, which resulted in a continuous scale from 1 to 25.Nevertheless, as a robustness test, following Beyers and Kerremans (2012), we also estimate our models with access to parties in general and subsequently to governing and opposition parties, respectively.In POS models, the legislature is treated as an especially important venue explaining the openness of a political system to societal interests because legislators have a direct democratic mandate giving them incentives to seek contact with civil society actors (Kitschelt, 1986).Thus, we also test our models with only access to parliament.
Our most important variable measuring organizational resources is a composite variable, professionalization.The variables are based on a survey item asking respondents to evaluate on a five-point scale (from much less to much more) to what extent their organization focuses on the following activities as opposed to 10-15 years ago: (a) organizational development, (b) human resources development, (c) fundraising, (d) strategic planning, and (e) evaluation of efficiency and effectiveness.Following Klüver and Saurugger (2013), we included staff in our composite variable of professionalization, as we believe having paid employees reflects organizational development.We operationalize financial resources using a variable measuring the financial planning horizon of an organization on a 5-point scale ranging from 1 (less than one year) to 5 (more than five years).
Backsliding is first measured with country dummies, with Slovenia being the baseline.We capture these country-level processes with country dummies and simultaneously control for omitted variable bias at the country level.The country-and policy-field dummies also enable us to account for particularities in interest group activity driven by varying levels of regionalization and subnational policy competencies.
Additionally, we introduce an organizational-level proxy for backsliding based on a survey item measuring the perceived frequency of governmental meetings in an organization's respective area of activity (not between the organization and the government).That is, the answers give a general assessment of the frequency of government consultations measured on a five-point scale: never, once a year, twice a year, monthly, and weekly.There are pronounced differences across countries in this measurement (Figure 3).For the analysis, we created a new dichotomous variable, which takes the value 1 for organizations reporting no or only yearly consultations in their field of activity, and 0 for all other organizations in the dataset.
We measure intergroup cooperation with three variables: two on the national and one on the EU levels, respectively.For national and EU umbrella memberships, we have two dichotomous (dummy) variables taking the value 1 for membership and 0 for non-membership.For domestic cooperation, we created a composite variable from four survey items measuring different forms of cooperation between interest groups in fundraising, representation on advisory boards, issuing joint statements, and formulating joint political strategies.For each answer never we assigned the value 0; for occasionally we assigned the value 1; for frequently we assigned the value 2. That is, we received a continuous scale ranging from 0 to 8. We also asked organizations to evaluate the importance of four types of expertise for policy influence: technical/scientific, legal, economic, and impact assessment (De Bruycker, 2016).We created a composite expertise index, which takes the value 0 for any type of expertise evaluated as unimportant by the respondents in their interactions with policy-makers, 1 evaluated as somewhat important, and 2 marked as very important.Thus, the composite expertise index is a continuous scale between 0 (if all four types of expertise were marked as unimportant) and 8 (if all were marked as very important).Finally, longevity is operationalized as organizational age (logged).See Table 1 in the Supplementary File for a summary of variables.

Descriptive Analysis
To explore our dependent variables, we first depicted the means of the variables measuring the change in the importance of regional-, national-, and EU-level representation per policy field and country (Figure 1).The bar charts reveal that, on average, all representational levels became somewhat more important compared to a decade ago (3 means no change, 4 means more importance, 5 means much more importance on the scale).However, there are clear differences between countries and policy fields.
Hungarian respondents, on average, reported the lowest increase in importance of any representation levels, and Slovenian and Czech respondents reported the highest (Slovenians slightly more).Interestingly, Hungarian energy policy organizations report the least positive change in the importance across all levels of representation.This lends support to the findings of Horváthová and Dobbins (2019) that the government provides access to a few influential business groups while excluding others.The graphs do not indicate a flight from the national level of excluded groups.
In the other three countries, the EU level has become particularly important for energy groups.In all four countries, the regional level gained the least in importance.However, despite recentralization processes, the balance is positive even for the regional level except for Hungarian higher education groups, for whom it became less important.This is likely a sign of the increasing government control and weakening academic freedom in Hungarian higher education (Kováts, 2018).
We also plotted our relational dependent variables (Figure 2).Here the differences are more pronounced between the four countries.As expected, it is neocorporatist Slovenia where the national level gained the least importance relative to the European and regional levels (with a relatively large increase in importance at all three levels; see Figure 1).However, the numbers are more difficult to interpret for the other three countries.It is Hungary where the national level gained the least importance relatively after Slovenia compared to both other levels of representation, much less than either Poland or the Czech Republic.Now, looking at the previous graphs in Figure 1, this result is somewhat less surprising.After all, Hungarian interest groups reported the least increase in importance across all three levels for representation.However, we need to estimate multivariate models to better understand these patterns.
It is important to also look at the dispersion of our individual level proxy for backsliding, a variable indicat-ing the respondents' assessment of the frequency of government consultations in their field of activity on a five-point scale where 1 is no consultations, 2 is once a year, 3 is twice a year, 4 is monthly, and 5 is weekly (Figure 3).The boxplots show that Slovenian groups report the least frequent governmental consultations in their field of activity.This again might be an effect of neocorporatism, whereby there are formal interest representation forums and bigger, encompassing organizations enjoy an advantage over smaller ones.Hungarian and Slovenian respondents have the same median; however, the interquartile range is narrower for Hungarian ones.Czech and Polish organizations report more frequent consultations than Hungarians or Slovenians.Again, we do not know whether more frequent consultations translate to easier access and more frequent participation, let alone higher policy influence.As already elaborated above, we use a dichotomous variable based on this measure in our multivariate models to better highlight the effect of exclusive governmental policy-making.

Statistical Analysis
We estimated seven multivariate OLS models to explore the determinants of the changes in the importance of the levels of representation.In every model, we control for country (baseline Slovenia) and policy field (baseline higher education).For each model, we plotted the kernel density over the normal distribution and estimated the variance inflation factor (for each coefficient and the mean for the model as a whole) to control for multicollinearity.
In Models 1, 2, and 3, we estimated the three levels separately.Access to domestic policy-makers is in a positive relationship with an increased importance for the national level lobby activity, whereas in a negative one with both the EU and regional levels.That is, closer proximity to national level policymakers does not seem to induce higher EU or regional level lobby activity (H2).On the contrary, groups with lower domestic access seem to value the supra-and subnational levels somewhat more (H3).However, we do not find support for either hypotheses, as the coefficent for the composite access index remains insignificant in all three models.Estimating the models with party access in general, or access to governing and opposition parties, respectively, or access to parliament does not change the results (see Supplementary File, Tables 2-5).
At the national level, the coefficient of the dummy for having no or only yearly governmental consultations in the field of activity-our proxy for backsliding at the organizational level-is negative and significant (H4).However, at the regional and EU levels, it loses statistical significance.In Model 1, the coefficient of the composite expertise index is significant and indicates a positive relationship.That is, expertise provision has a positive effect of placing greater importance on national-level representation compared to 10-15 years ago.However, expertise provision loses significance at the EU and regional levels.
Turning to the effect of organizational resources (H1), professionalization has a positive and significant effect at all three levels, suggesting that professionalization is a key organizational resource.More professionalized organizations engage in more active lobbying not only at the national but also at the EU and regional levels.We plotted the effect of professionalization on the change in importance for EU-level representation (Model 2).The marginal effect plot clearly reveals a sub-stantial and robust effect (Figure 4).Organizations that did not invest in enhancing their organizational capacities in the past 10-15 years place less importance on EU-level representation than those that did.
While the policy field itself turned out to be insignificant in all models, national differences indeed are pronounced at the EU and regional levels: Hungarian organizations place less importance on both levels, whereas Polish groups are less active only at the regional level.Hence, Polish organizations seem to be adapting to the general trend towards centralization (Rozbicka et al., 2020) while-unlike their Hungarian counterparts-also still branching out to the EU level.Organizational age is a positive predictor for more EU-level lobbying and a negative for regional lobbying (the coefficient is negative but not significant at the national level).This might indicate that older, more influential organizations have turned to the EU level in the past 10-15 years.From our interorganizational variables, only EU umbrella membership has a significant effect, a positive one at the EU level (unsurprisingly).
These models, however, do not tell us how the different representation levels relate to each other.To answer this question, first, we regressed the national level on the EU and the regional level, excluding access from the models but leaving all other variables in.Models 4 and 5 lend support to H2.The main driver for placing more importance on both EU and regional level representation is indeed increased importance for the national level.We learned from Model 1 that these are privileged, highly professionalized organizations providing expertise to national policy-makers active in policy fields with relatively frequent governmental consultations.
To model what determines organizations to move towards or away from the national level, either to the EU or regional levels, we need a different, relational operationalization of the dependent variable.For a detailed description of the two relational variables, please refer to Section 3 and Table 1.As a reminder: The two variables range (potentially) from −4 to +4.Negative numbers indicate higher importance for the national level and, at the same time, no change or reduced importance of the EU or the regional level, while positive numbers indicate just the opposite.Models 6 and 7 clearly show that our proxy variable for backsliding-the perceived frequency of governmental consultations in the area of activity-affects a move toward the EU and regional levels away from national representation positively.That is, organizations reporting no or only yearly government consultations place more importance on both EU and regional level representation compared to 10-15 years ago.This lends support to H4.The coefficient for access is, however, still insignificant in both models suggesting that a general closure of the POS-a structural condition, that isis more important in explaining the strategic choice of moving toward the supra-or subnational levels than individual group inclusion/exclusion.It is also interesting that domestic inter-group cooperation becomes significant  in these relational models and positive both for the EU and regional levels.This finding tentatively suggests that inter-group cooperation is indeed a "weapon of the weak," an important asset against environmental uncertainties (Hanegraaff & Pritoni, 2019;Pfeffer & Salancik, 2003).The significant and negative effect of the composite expertise index strengthens our findings in Model 1.Those organizations engaging in expertise-provision to domestic policy-makers evaluate national-level representation more importantly than 10-15 years ago.It also lends tentative support to H3: As access to policy networks is a necessary condition for informationoriented lobbying, exclusion means a move away from the national towards the EU and regional levels.

Conclusions
This article assessed whether organized interests operating amid (varying degrees of) democratic backsliding are "defiantly responding" to the new playing field.Specifically, we explored whether increasing engagement in alternative political venues-i.e., the EU and/or regional level-is driven by organizations' (lacking) individual access to policy networks or rather by a general closure of the POS.We also assessed how multilevel lobbying is influenced by organization-specific variables such as financial resources, professionalization, and expertise, as well as the meso-level of inter-group cooperation.Are organizations under pressure from illiberal governments "taking their business elsewhere"?First, it depends somewhat on how the question was posed-and the country's context.Our variables linking (lacking) access to national policy-makers with increasing regional and international activity revealed neither a multiplication of lobbying capacity (i.e., better national access = more multilevel lobbying) nor evidence that excluded groups seek alternative lobbying forums.Hence, excluded organizations seem to be coping with or adapting to the situation.Hungary, the country most significantly affected by backsliding, stood out as an extreme case, as Hungarian organizations are even less likely to "go abroad" or regionalize despite the adverse climate.This also holds, to a somewhat lesser extent, for Polish organizations.
However, when we applied our organization-specific proxy variable for backsliding, namely the perceived (lacking) frequency of governmental consultations in the specific policy area and our relational variable for venue shopping (expressing the relative importance of the national vs. the EU or the regional levels, respectively), our regressions indeed reflected a move towards the EU and the regional levels.Once again, though, Hungarian and Polish organizations appear less willing to engage in venue shopping.This ultimately lends evidence to a depressing effect of democratic backsliding on organizational responsiveness (and potentially a positive effect of Slovenian corporatism on EU-level lobbying).
Yet there is hope for organizations operating in backsliding contexts.Nearly all models show that organizational self-empowerment is crucial.Organizations focusing on accumulating expertise, and professionalization, i.e., through training lobbyists, monitoring effectiveness, and cooperating with other organizations, not only can sustain access to governments but also branch out to other (regional and European) lobbying forums.Hence, the self-induced micro-level development of organizations enables them to react, regardless of their proximity to policy-makers.In other words, "better businesses," i.e., more professionalized, expertise-oriented, and cooperative organizations, appear capable of operating in multiple venues, regardless of whether included or excluded at the national level.On another positive note, our results show a strong correlation between expertise provision and increasing national-level activity, thus indicating that illiberal, populist governments are still interested in gathering expertise from civil society organizations.However, it may also be the case that illiberal governments are specifically propping up such well-endowed, expertise-intensive organizations and potentially even supporting their EU-level activities.In other words, organizational resources and lobbying capacity may be driven more by symbiotic relationships with governments than own personal initiative (for rather weak evidence for this phenomenon in CEE see Dobbins, Horváth, et al., 2022).
Aside from these complex causal processes requiring further exploration, our perspective opens numerous avenues for future research.Clearly, our bird's-eye quantitative approach somewhat overlooks the dynamics of individual organizations.Thus, case studies might provide more lucid insights into specific organizational decision-making processes, specifically with regard to governments' leverage over organizational development.Importantly, multilevel lobbying might be driven by factors other than democratic backsliding.For example, an increasing density of the organizational population might crowd out the playing field and compel organizations to shift operations elsewhere.Moreover, future scholarship might explore whether targeted funding from the EU or other foreign donors enhances multilevel lobbying capacity among CEE organizations.Finally, authors should also assess whether venue shopping at the European level ultimately strengthens (i.e., due to learning effects) or weakens (i.e., due to administrative overburdening) organizations' influence on national policy-making.

Figure 1 .Figure 2 .
Figure1.Mean change in the importance of regional-, national-, and EU-level representation levels compared to 10-15 years ago per country and policy field.

Figure 3 .
Figure 3.The frequency of governmental consultations in the field of activity.

Figure 4 .
Figure 4. Predictive margins for the change in the importance of EU-level representation as compared to 10-15 years ago for the levels of professionalization, 95% confidence interval.

Table 1 .
Descriptive statistics of the two relational dependent variables (national vs. EU levels/national vs. regional levels).