Issue substitution or volume expansion? How parties accommodate agenda

How do political actors respond when an issue suddenly jumps to the top of the public agenda? While conventional theories of party behaviour predict that parties increase their attention to that issue, they tell us little about how they will do so. One approach is to increase attention to the focal issue while maintaining the messaging level on other issues ( volume expansion ). Alternatively, political actors can increase their attention to the focal issue while decreasing their emphasis on other issues ( issue substitution ). We theorize that the overall volume of communication determines which approach dominates: Parties with high communication volumes will tend towards issue substitution, whereas those with lower communication volumes will prefer volume expansion. We confirm this hypothesis using a data set covering all press releases issued by members of the Austrian parliament between 2013 and 2017—a period that includes the 2015 ‘refugee crisis’ as an agenda shock.


Introduction
How do parties adapt to changes in the public issue agenda? While theories of issue competition typically assume that parties are responsive to external events and public opinion (Ansolabehere and Iyengar, 1994;Sides, 2006;Sigelmann and Buell, 2004;Spoon and Klüver, 2014), they tell us next to nothing about how parties enact this response, and what explains cross-party variation in responsiveness strategies.
This paper makes a contribution by presenting two alternative responsiveness strategies of parties: one reaction to a changing issue agenda is to increase attention to the focal issue while maintaining the level of attention to other issues (volume expansion). Alternatively, parties can raise attention to the focal issue while de-emphasizing other topics (issue substitution). We posit that a party's volume of communication determines how it chooses from these two options. Parties that communicate a lot should pursue issue substitution as they are likely to run into diminishing returns when producing more messages overall. Parties with lower volumes of communication do not face this constraint and can therefore opt for volume expansion to show responsiveness whilst maintaining attention to issues that are more strategically favourable to them (for example, based on issue ownership considerations).
We test these expectations using data on all press releases issued by members of the Austrian Nationalrat between 2013 and 2017. This A widespread view in the literature assumes that all parties should have strong incentives to respond to the issue priorities of the public. Theories of democratic representation and responsiveness would suggest that (changes in) voter concerns should feed into party agendas (Dalton et al., 2011). Thus, parties are expected to adapt their issue emphasis to changes in issue salience among the electorate. By 'riding the wave' on issues that are publicly salient (Ansolabehere and Iyengar, 1994), parties may demonstrate that they take the concerns of voters seriously and thus enhance their popularity. This strategy may also help parties generate media attention for their messages (Hopmann et al., 2012;Meyer and Wagner, 2016).
Other theories of party competition consider parties as having their distinct issue profile which is relatively stable over time. For example, theories based on issue ownership and issue saliency suggest that parties should try to draw attention to issues on which they have a competence advantage over their political opponents. Particular parties are perceived as better at handling certain policy issues, or solving certain problems (Petrocik, 1996;Petrocik et al., 2003), such that specific issues are reliably linked with certain parties in the minds of citizens (Walgrave et al., 2012;Tresch et al., 2015). Hence, parties should follow a strategy of selectively emphasizing their 'best' issues to shape the public issue agenda (Budge and Farlie, 1983;Robertson, 1976). If the political debate focuses on those favourable issues, then the issue owner should be able to gain electoral support, as citizens will view the party more positively (Bélanger and Meguid, 2008). At the same time, parties should seek to downplay other issues on which their opponents enjoy an advantage (Petrocik, 1996). This also implies that parties should not compete with each other by directly fighting over the same issues, but rather focus on their own areas of concern (Budge and Farlie, 1983).
A third approach is the issue yield theory developed by De Sio and Weber (2014) which identifies the electoral risks and opportunities for different parties associated with various issues stances (see also De Sio et al., 2018). It provides a dynamic and party-specific view on issue competition. Specifically, it is argued that parties should primarily focus on issue positions that are popular among both, the general electorate and their own voters. As these 'bridge policies' provide an opportunity to expand the electoral support without losing existing supporters, parties are expected to strategically emphasize those issues with a higher yield. A more recent extension of the issue-yield framework also takes valence issues into account by focusing on party credibility on an issue (De Sio and Weber, 2020;D'Alimonte et al., 2020).
While parties may not find a newly arising issue especially appealing and thus try to avoid it (Budge et al., 2001), this is only possible to a limited extent. If the importance of an issue is high enough, there is no way political actors can ignore the subject, as failure to engage on this topic can threaten their credibility in the eyes of the public. Ultimately this might result in a loss of electoral support if voters perceive political parties to be 'out of touch' with their concerns (Sides, 2007;Spoon and Klüver, 2014). Even if parties agreed in a cartel-like fashion to keep an issue off the agenda (Blyth and Katz, 2005), they would risk opening up space for new actors -issue entrepreneurs -to gain support by being more responsive to voters' concerns than established parties (De Vries and Hobolt, 2020). This electoral threat should be a sufficiently strong incentive for parties to respond to highly salient topics and increase their attention to the issue at hand. Hence, parties should even address disadvantageous issues that are owned by their rivals if these issues are important to the electorate (Damore, 2004).
The literature on issue responsiveness confirms that parties typically respond to voters' issue priorities (Klüver and Sagarzazu, 2016;Spoon and Klüver, 2014), even though responsiveness varies with the political context. It increases with electoral competition (Abou-Chadi, 2018) and voter polarization (Spoon and Klüver, 2015). It is also more pronounced for opposition parties and larger parties with a broader target electorate (Klüver and Spoon, 2016;Wagner and Meyer, 2014).
Even if parties may not want to respond to the public's issue priorities, responsiveness may be forced upon them by the pressures of party competition. Once other parties in a polity have taken up an issue (and thus increased its 'systemic salience' Steenbergen and Scott, 2004), the issue becomes more difficult to avoid Mortensen, 2010, 2015). As no party has 'monopolistic agenda control' (Steenbergen and Scott, 2004, 169), the party-system agenda both influences and constrains the issue emphasis of individual parties. Ignoring an issue on the party-system agenda may be costly, as parties might lose their ability to frame an issue in their favour (Jerit, 2008;Nadeau et al., 2010). Additionally, if a party chooses to completely disregard an issue that features high on the public agenda, it will not get a chance to inform the electorate about its specific position on the topic (Nadeau et al., 2008).

Responsiveness strategies of parties
The scholarly work discussed above clearly stresses parties' incentives to respond to voters' issue priorities, while highlighting the constraints generated by the competitive environment and the electorate's perceptions of party issue competence. Whereas responsiveness to the public's priorities seems like an obvious strategic choice, there is always a trade-off in diverting attention from one issue to another. To see how parties navigate this trade-off, we have to re-consider how issue emphasis is typically understood in the literature.
Studies of party competition mostly conceptualize (and operationalize) issue emphasis as the relative frequency of an issue in a political actor's agenda: the percentage -rather than the absolute number -of observations dedicated to an issue, be they manifesto quasi-sentences, TV ads, Facebook posts, Tweets, press releases, or paragraphs in a speech. To be sure, there is often a good reason to follow this approach: Relative instead of absolute frequencies allow for large-scale comparisons across actors, periods, and communication channels. After all, parties and politicians differ systematically in the resources and incentives they have to produce amounts of political text, and communication channels differ in the volume of messages they can process in a given period. Many research purposes require eliminating these differences.
Yet, these practical necessities should not blind us to the fact that, conceptually, absolute frequencies are a crucial dimension of a party's issue communication, with considerable variation across parties and over time (Däubler and Benoit, 2013;Dolezal et al., 2012).
Thinking in terms of absolute instead of relative frequencies also points us towards the importance of considering marginal costs in political communication. Generally, we can assume that the marginal cost of message production -the costs of producing one additional message -in political campaigning is very low (though not zero). Once a party has established a communication infrastructure that produces a certain volume of messages, it is very cheap to add another paragraph to a manifesto or speech, issue one more press release, or put out another social media posting (a somewhat different calculus may apply to paid advertisements, even though economies of scale operate in this area, too). Therefore, political actors have some freedom to vary not only the relative emphasis they put on issues but also the overall volume of messages they produce. To be sure, no party's resources are infinite, and there is certainly the possibility of running into diminishing returns. This should keep parties from producing an unending stream of messages -a point to which we return below.
Once we accept the premise that absolute frequencies are a relevant dimension of issue communication, it becomes clear that parties have several ways of adopting their message production to changes in the issue agenda. 1 First, they may not respond at all -an option that becomes dramatically more unlikely as the size of the agenda change increases. Second, they may increase their attention to the focal issue in absolute terms, while maintaining their communication effort on other issues. As this strategy produces a higher overall number of messages, we call it volume expansion. Third, parties may increase their messaging on the focal issue, while decreasing their emphasis on other issues. This strategy of issue substitution produces the largest relative change in issue attention. 2 What determines which of these two responsiveness strategies a party will follow? We theorize that one important factor is a party's overall communication effort. Parties vary in how much personnel and money they can mobilize to produce messages (Webb and Keith, 2017) and resources can play a significant role in issue engagement (Meyer and Wagner, 2016). On balance, resource-rich parties will be able to maintain a higher overall volume of message production (Gibson and Rommele, 2001). Parties that usually produce larger message quantities should react differently than those with a low communication activity, because the marginal utility of a message depends on the overall level of message production. The origin of this argument goes back to our earlier point about the importance of theorizing responsiveness and issue competition not just in terms of relative, but also absolute frequencies.
In theory, a party could issue an almost infinite number of statements about different issues on any given day to be responsive to a broad range of voter concerns while also maintaining attention to its own priorities. Yet, given the limited capacity (and willingness) of the political and media system to absorb messages from one single actor, such a strategy would run into diminishing returns quite quickly -a problem that has long been an issue in marketing research (Johansson, 1979;Ha and Litman, 1997). After all, media attention is scarce and parties are competing with each other and other actors and events to push their messages into the news . This logic extends to unmediated communication channels, such as social media, where user engagement declines in the number of daily postings (Xenos et al., 2017). Moreover, parties may risk watering down their message by issuing huge quantities of posts .
Given that the marginal cost of producing a message is low but not zero, the law of diminishing returns dictates that the net utility from issuing an extra message reaches negative territory at some point. Communicating more would then be worse than ineffective -it would be outright damaging. In addition, given that political and media attention is a limited resource, an overly high quantity of messages would dilute the effectiveness of each single message, thus turning the calculus further negative.
The logic of diminishing returns thus imposes a ceiling effect on the volume of party communication. Under such conditions, more input (issue demand by the public) no longer leads to more output (message production on the issue), unless some other issue is de-emphasized to make space for the focal issue.
Parties with lower communication volumes are much less likely to run into this problem and can therefore simply be responsive by (temporarily) increasing message production on the newly important topic without having to curtail their communication efforts on other issues. We should therefore expect volume expansion as a strategy from parties whose prior communication levels are low.
By contrast, parties communicating at higher volumes are much more likely to be plagued by the problem of diminishing returns. As simply adding more messages is not advisable, these parties have to curtail their communication efforts on some issues in order to make space for a topic that becomes newly important. Hence, for parties that already produce a lot of messages, the most plausible response to agenda changes should be to redistribute attention from non-focal issues to the focal issue (issue substitution).
Parties' overall communication volume is likely to be a function of their resources, as higher levels of message production require more communication infrastructure and staff. While adding more resources to a party's media operation is always possible (within limits), it makes sense only as long as the marginal utility of a message remains positive. This is much more likely for parties with fewer resources and thus lower communication volumes.
Our central hypothesis is therefore that volume expansion is more likely to occur when message production is low, whereas issue substitution is more likely to prevail at high levels of communication.

The Austrian case
For our empirical analysis, we rely on data collected from almost 20,000 press releases issued during the 25th legislative term of the Austrian Nationalrat, spanning the period between the general elections in 2013 and 2017.
Our period of observation covers the so-called European 'refugee crisis' during the summer of 2015, which upset the political agenda of many democracies in Europe. Even though the issues of immigration and immigrant integration had been central to political conflict and party competition before, the influx of large numbers of asylum seekers during 2015 and 2016 produced an 'agenda shock' -a sudden and dramatic shift in the issue priorities of the public, the media, and political actors. While annual asylum applications in the EU-28 had averaged just under 400,000 between 2010 and 2014, the numbers increased to 1.3 m in 2015 and 1.2 m in 2016, before returning to around 700,000 for the years 2017-19. In per-capita terms, Austria was one of the most affected countries in Europe, receiving 10 asylum applications per 1,000 inhabitants in 2015 (only topped by Sweden (17) and Hungary (18)) (see the migr_asyappctza indicator in the Eurostat Database).
As Fig. 1 shows, the proportion of Austrian respondents who named immigration as one of the two most important issues facing the country shot up from around 20% to 56% during the year 2015, clearly overtaking unemployment (30% in late 2015) as the top concern. Similar peaks in issue importance were registered throughout Europe, especially in strongly affected countries such as Germany (76%) and Sweden (53%). In parallel to voter perceptions, media attention to asylum and refugeerelated issues increased dramatically (Greussing and Boomgaarden, 2017). Although the empirical analysis of this paper focuses on Austria during the 'refugee crisis', the theoretical framework reaches beyond this specific case: We argue that it may be equally applicable for other instances when one issue very rapidly rises in importance, strongly influences not only public opinion and media attention, but also the party-system agenda, and has severe consequences with regard to policy measures. Therefore, the theoretical framework can help us gain a better understanding of parties' responses to other examples of 'agenda shocks', such as the COVID-19 pandemic, the Fridays-for-Future protests, the economic and financial crisis, the 9/11 terrorist attacks, or the nuclear disasters in Chernobyl and Fukushima.
Austria is a suitable case because it features several characteristics typical of a European parliamentary democracy, such as a proportional electoral system, multiparty competition, and coalition government. Furthermore, as much of the literature focuses on political communication and issue competition during election campaigns, our study also provides valuable insights into party communication during non-election periods.  We include all six parties with parliamentary representation for the period under study: The Social Democratic Party of Austria (SPÖ), the Christian democratic Austrian People's Party (ÖVP), the populist radical right Freedom Party of Austria (FPÖ), the left-libertarian and environmentalist Greens, the liberal NEOS (The New Austria and Liberal Forum), and the short-lived populist Team Stronach (TS). The two latter parties were founded in 2012, only one year before the 2013 election. In terms of government formation, the most frequent coalition type in post-war Austria has been a coalition of SPÖ and ÖVP (Müller, 2006), which was also in office during this period.

Data
Dependent variables. For the investigation of how Austrian parties responded to the 'refugee crisis', we measure their issue emphasis using press releases 3 issued by members of the Austrian Nationalrat (MPs) 4 during the legislative period from 2013 to 2017 (Müller et al., 2022). Press releases are a suitable source to examine political communication strategies over the legislative cycle: During routine political periods, when public attention to everyday politics is rather low, press releases are a useful tool for parties and politicians to keep the public informed about their political agenda. If picked up by newspapers, their authors can reach out to a large audience . Furthermore, press releases allow politicians to respond quickly to the dynamics of the political agenda during the electoral cycle (e.g. Klüver and Sagarzazu, 2016;Meyer et al., 2020). Analytically, they provide for high temporal granularity because they are usually issued on a daily basis.
Trained coders manually recorded the issue content of the title and subtitle of each press release. We expect title and subtitle to cover the most important messages of the press release, especially since press releases target journalists who work under considerable time pressure (Dolezal et al., 2017, 671-2). After initial training, the coding was executed independently by ten student coders. The inter-coder reliability tests for a random sample ( = 250) of press releases yielded satisfactory results (Krippendorff's = 0.78).
3 Press releases issued by political parties and their representatives are freely available from www.ots.at, a website run by the Austrian Press Agency (APA). 4 Press releases are issued by individual politicians and access is not restricted to party leaders. The publication is organized in different 'channels' (e.g. for the party's central office, the parliamentary party group, or its regional branches).
For our analysis, we construct two different dependent variables: The first one counts the number of press releases per month that address the topic of immigration 5 either in the title or subtitle of the press release (2,335). Complementary to that, our second dependent variable counts all press releases where a non-immigration issue is addressed in the title or subtitle (16,910). Press releases where no policy issue was coded either for title or subtitle, were excluded from our analysis (e.g. press releases announcing personnel decisions or upcoming press conferences) (770).
Independent variables. To test how parties react when faced with an agenda shock, we use three independent variables: the monthly number of asylum applications (from the Federal Ministry of the Interior), the salience of immigration among voters, and the number of media reports on immigration per month. 6 While much of the literature on party responsiveness examines reactions to public opinion changes, we believe that public opinion may be a lagging indicator of agenda change. At times, events, media reporting or even political elites themselves may lead public opinion. In the case at hand, however, all three variables are highly correlated and yield very similar results (see below). Still, among these independent variables, the monthly number of asylum applications should be most appropriate to capture the theorized mechanisms, because we view it as causally antecedent to the other two: asylum applications produce media and voter attention, not the other way round. What is more, the voter data are available at six-month intervals only (see Fig. 1), thus requiring most data points to be imputed (see Online Appendix for details on the imputation procedure).
As we have theorized above, the strategy a party chooses to respond to an agenda shock should depend on the size of its overall communication effort. To test this hypothesis, we use the average number of press releases issued by a party in the previous three months (including those press releases where no policy issue was coded). 5 Immigration is defined by aggregating 56 categories (see Table A.1 in the Online Appendix) from a coding scheme with more than 700 categories (see Dolezal et al., 2016 for details on the coding process), including subtopics like asylum and migration policies, social, economic and cultural integration of immigrants, as well as migration management on the European level. 6 We use a short list of keywords (refugee, migration, migrant, asylum and Islam) to obtain the number of articles on the topic from eight national daily newspapers. Any article with at least one matching keyword was counted as a hit. Results were scraped from the Austrian Press Agency's (APA) online media archive.
Control variables. Because we use press releases issued by MPs, we control for the monthly number of plenary speeches held on immigration and non-immigration issues in the Nationalrat (per party). MPs use press releases to draw attention to their parliamentary work (Huber et al., 2020), so that the parliamentary agenda will likely affect our dependent variables. We also include a predictor for the 'systemic salience' of immigration and non-immigration issues. This is simply a logged count of the number of other parties' press releases for both categories.

Model specification
We set up our data in a month (49) × party (6) format. By counting party press releases per month, we obtain a sufficiently high number of observations (N=264) 7 , while we reduce the risk that short-term fluctuations of the issue agenda largely affect our measure, as could be the case if counting party press releases on a weekly or daily basis. Moreover, this data structure corresponds to our main independent variable (number of asylum applications), which is reported by the Ministry for months only. We take the natural log of our dependent variables to remove skew, as this allows us to use linear models with lagged dependent variables and fixed effects at the party level. We also include dummies for months (January through December) to control for seasonal variation in message volume (e.g. the summer recess) and add a time trend variable (a simple count of months running from 1 to 45) to account for the fact that, with much party communication moving to social media channels, the overall number of press releases is slowly declining over the course of our observation period. Standard errors are clustered on parties.

Analysis
Our two dependent variables are logged counts of monthly press releases (PRs) issued by a party's MPs: one counting immigration-related press releases, the other counting all PRs focusing on other policy issues. How do both numbers change when the issue of immigration rises in importance? We first look at the effect of asylum applications in more detail, but we also provide models that use voter salience of the immigration issue and media reports on immigration as independent variables (these regressions are reported in the Online Appendix).
We display our dependent variables in Fig. 2. The dashed blue lines show smoothed trends for the immigration-related press releases. For all parties, there is an increase as the number of asylum applications rises (indicated by the histograms in the figure), and a decrease as applications ebb back in 2016 and 2017. The number of immigrationrelated messages thus correlates strongly with the number of asylum applications.
However, some parties are clearly more responsive than others. In 2015, Team Stronach issued on average 12 immigration-related press releases more per month than in 2014. The FPÖ is a close second with 11 press releases, followed by the Greens (8), SPÖ (5), ÖVP and NEOS (both 4). Thus, in line with issue yield theory (De Sio and Weber, 2014;De Sio et al., 2018), there were different incentives for Austrian parties to emphasize the issue of immigration when it was rising in importance. While Team Stronach and FPÖ might have seen this as an opportunity to increase their electoral appeal (due to their position or reputation on immigration), responsiveness might be rather forced upon the other parties (especially SPÖ, ÖVP and NEOS) by the tremendously raising salience of the issue.
Compared to the immigration-related press releases, the party trends for non-immigration press releases (solid red lines in Fig. 2) do not only vary in their extent, but also regarding their direction. Some parties show ups and downs or flat lines (SPÖ, ÖVP), whereas others display almost steady downward slopes (FPÖ, Greens, Team Stronach). The non-immigration trend for NEOS is even increasing over time.
Turning to the multivariate analysis, Model 1 in Table 1 shows the effects of asylum applications on press releases about immigration, Model 2 displays the effects on non-immigration press releases. Parties are clearly responsive to the changing issue agenda. The number of asylum applications has a significant effect on the number of press releases addressing the topic of immigration. Given that both, the independent variable (asylum applications) and the dependent variable are log-transformed, we can express the coefficient (0.498) as follows: A ten-percent increase in asylum applications is associated with an increase in immigration-related press releases by 1 − (1.10 0.498 ) = 0.049, that is, 4.9 percent. This positive effect is in line with previous findings on issue responsiveness (Spoon and Klüver, 2014;Wagner and Meyer, 2014;Klüver and Sagarzazu, 2016;Abou-Chadi, 2018). It is also matched by our alternative independent variables. A onepoint increase in voter salience of immigration is associated with a 2.1-percent increase in immigration-related press releases ( Table A.2). Similarly, 100 additional media reports on immigration are associated with a 2.5-percent increase in press releases on the issue (Table A.4).
However, in this paper we want to go a step further and investigate how parties enact responsiveness when an issue raises in importance. To do so, we test whether the number of non-immigration press releases is influenced by the rising importance of the immigration issue. As can be seen in Model 2, we find no effect for our main independent variable. While the sign is negative, pointing to a decline of non-immigration press releases when asylum applications increase, the coefficient is very close to zero and does not reach statistical significance. 8 Looking at the overall pattern, volume expansion thus appears to be the main way for parties to react to a changing issue agenda. Parties increase the number of press releases on the focal issue while keeping attention on other policy issues stable. Our alternative independent variables (voter issue salience, media reports) yield very similar results: positive and significant effects on immigration-related PRs, non-significant effects for non-immigration PRs (see Tables A.2 and A.4).
To test our expectation that parties' reactions to shifts in the issue agenda should vary according to their overall communication effort, we run additional regression analyses in which we interact our key independent variables (asylum applications, voter salience, media reports) with a party's overall communication effort, measured as the average number of press releases in the past three months (see Model 2 in Table 2).
Our main expectation is that parties with high communication volumes will engage in issue substitution, whereas those with lower numbers of press releases can afford to expand their volume of communication, resulting in a stable number of press releases on nonimmigration issues and an increasing number of press releases on immigration. Empirically, this implies a negative interaction effect in the models with non-immigration PRs as the dependent variable. The larger a party's prior communication effort, the more negative the effect of our three independent variables on non-immigration press releases should be. Fig. 3 displays the average marginal effects of our three independent variables (asylum applications, voter salience, media reports) on immigration and non-immigration press releases by a party's prior communication effort -the average number of press releases in the previous three months. The effect of the three variables on immigrationrelated press releases are positive and significant at low communication L. Ennser-Jedenastik et al.

Table 1
The effects of asylum applications on press releases (PRs).
(1) volumes, yet they all decline substantially as a party's communication volume increases. For example, a ten-percent increase in asylum applications is associated with a seven-percent rise in immigrationrelated press releases for a party issuing 30 such statements in the previous month. For a party producing 130 press releases, the same scenario yields an increase of only three percent in immigration-related statements.
Even so, the effect of asylum applications on immigration PRs remains positive and statistically significant across the empirical range of the communication volume variable. Yet, this is not the case for the two alternative measures: both become statistically insignificant at higher communication volumes. The effects of asylum applications, voter salience, and media reports on non-immigration press releases also vary with overall communication activity, as indicated by the L. Ennser-Jedenastik et al.

Table 2
The effects of asylum applications on press releases (PRs) conditional on prior communication levels. (1) (2) Immigration PRs Non-immigration PRs Standard errors in parentheses * < 0.05, ** < 0.01, *** < 0.001.  Tables 2, A.3, and A.5. While results for all three specifications are aligned with our theoretical argument, the effects might also capture different trends for various issues, which calls for a more cautious interpretation. 9

Discussion and conclusion
Parties may not only vary in the extent to which they are responsive to agenda change but also in the way they enact responsiveness. We introduce two strategies for parties to respond to changes in the public issue agenda: volume expansion and issue substitution. We also theorize that the choice of strategy will depend on a party's overall communication effort. Our analyses show that parties that communicate a lot are more likely to substitute their existing agenda to increase the importance of the focal issue in both, absolute and relative terms. By contrast, parties with lower volumes of communication -unaffected by the problem of diminishing returns to additional messages -opt for volume expansion, which allows them to signal responsiveness to pressing public concerns whilst maintaining attention to other issues. 10 Our findings have several implications for democratic representation and party competition. Democratic representation is based on party responsiveness to core voter concerns. From this perspective, the results from our study are reassuring as all parties were sensitive to massive changes in the public issue agenda. Thus, notwithstanding their policy profiles or whether the immigration issue was especially favourable to them, all Austrian parties devoted a substantive amount of their communication efforts to the issue when it suddenly jumped to the top of the public agenda during the 2015 'refugee crisis'.
Moreover, we show that parties' responsiveness strategies are constrained by diminishing returns. Scholars sometimes implicitly assume that, given enough resources, parties are unconstrained in their ability to produce political messages. Yet, our findings show that even resource-rich parties are limited by the absorptive capacity of the political and media system. As a party's communication volume increases, each additional message carries less benefit for the sender. Assuming that the marginal cost of producing a message is low but not zero, the net utility from issuing an extra message thus reaches zero at some point. Therefore, parties with high communication volumes face different incentives in response to an agenda shock than parties with low communication volumes.
In addition, our findings have implications for issue competition more broadly. Extant research sometimes treats 'riding the wave' and issue ownership not only as distinct but as mutually exclusive strategies. In terms of the underlying theoretical rationale of systemic arguments this applies to all parties for the former but functions as a party-specific heuristic for the latter. However, despite their mutually exclusive rationales, both strategies can produce the same empirical results: i.e. if a party's best issue is salient, issue ownership and riding the wave will both dictate more attention to the issue. Our analysis implies that the extent to which these differences materialize in empirical terms will depend on parties' ability and willingness to expand their overall communication effort. Producing more messages overall allows parties to respond to the public while maintaining a higher level of attention to their core issues. Volume expansion can thus soften the trade-off between riding-the-wave and issue-ownership strategies. By contrast, parties pursuing issue substitution face a much sharper trade-off between these two approaches.
Finally, our analysis points to a blind spot in the literature on responsiveness and issue competition: the 'agenda costs' of responsiveness. If parties increase their attention to one issue, what happens to the rest of the issue agenda? Which concerns are more or less likely to be displaced by a newly dominant topic?
Of course, our study also comes with limitations. First, it examines a small set of parties in a single country over one legislative 10 To be sure, stretching resources in such a manner may not be sustainable over the long term. Eventually, volume expansion may have to give way to issue substitution, if no new resources can be mobilized. In our case, however, the public salience of the immigration issue declined markedly after its 2015 peak, as did parties' attention to the issue.
term. Cross-national research will therefore be needed to examine how party characteristics (e.g. activist-vs. leader-dominated) or countrylevel factors (e.g. electoral systems) affect variation in responsiveness strategies.
Second, whereas we have focused on a massive short-time shock to the issue agenda, future research could explore whether and how responsiveness strategies vary with less dramatic instances of agenda change. Similarly, research with a longer time-frame could investigate under which circumstances temporary fluctuations in party issue emphasis may induce more enduring changes in political competition and issue salience.
Third, we have focused our study on issue emphasis. Recent research suggests that (large) parties may adjust their policy positions towards the median voter if the salience of an issue increases (Abou-Chadi et al., 2020). Future studies could therefore investigate whether and how parties adapt their positions in response to changes in the issue agenda.
Fourth, the study's focus on party press releases could bias our results as MPs (in particular those from parties with previously higher communication volumes) could distribute their messages over other platforms, such as social media. Of course, limited public outreach of individual MPs makes it hard to compensate for setbacks encountered on party channels. Moreover, quantitative constraints such as diminishing returns extend to social media platforms , which limits the room for manoeuvre. Yet, future research should investigate whether and how parties distribute their messages across various communication channels to cope with space constraints in the face of agenda shocks. While much research remains to be done, the differentiation between volume expansion and issue substitution presented in this paper certainly provides a useful conceptual starting point to examine variation in issue responsiveness and its determinants.

Funding details
The authors gratefully acknowledge funding from the Austrian Cooperative Infrastructure for Electoral Research (ACIER), financed by the Austrian Federal Ministry of Education, Science, and Research (BMBWF).

Data availability
Data will be made available on request.