From popularity to vulnerability: An application to dynamic representation in coalition governments

Abstract

Electoral vulnerability matters for policymakers’ responsiveness to the public. While coalition governments are the norm in Europe, research on government responsiveness to public opinion has studied the effects of electoral pressures mostly for single-party governments and employed measures of government popularity. This article draws on and extends this research by developing two alternative measures of electoral vulnerability—government potential vulnerability and formateur potential vulnerability—that account for popularity limitations. An illustration of the measures is given by Germany (1987–2005) as a case of agenda responsiveness in coalition governments. Data from the Comparative Agendas Project on executive speeches in Germany are combined with data on vote intentions and the most important problem. Empirical analyses find support for the expectation that electoral vulnerability conditions agenda responsiveness to public issue priorities and that the proposed measures reflect more accurately the sources of vulnerability than government popularity.

Keywords

agenda responsiveness coalition governments dynamic representation electoral vulnerability government popularity

Although research recognized the role of electoral pressures for government responsiveness, studies have mostly focused on cases of single-party governments (e.g. Canes-Wrone and Shotts, 2004; Hakhverdian, 2010; but see Hobolt and Klemmensen, 2008). To capture electoral pressures, previous work has relied on measures of government popularity based on opinion polls data and found that electoral incentives matter for governments to respond. In turn, the wide research on party responsiveness to voters (for a review, see Adams, 2012) recognizes the importance of the electoral context (e.g. Spoon and Klüver, 2014) but has only recently shifted attention on how participation in coalition governments affects party responsiveness (Klüver and Spoon, 2016).¹

Our article contributes to these research lines in two ways. On the one hand, we explore the conditional effect of electoral pressures on agenda responsiveness by developing two alternative measures. Giving an illustration of these measures is the primary goal of this article. On the other hand, we extend the limited research on dynamic representation in coalition governments by analyzing agenda responsiveness in Germany, a case so far understudied. We believe this is important for at least three reasons.

First, incumbent vulnerability is at the core of the connection between dynamic representation and political competition. Second, coalition governments are the norm in Europe. Third, measures of government popularity permit studying electoral pressures dynamically but leave aside important issues. Focusing on how well or badly the government is doing at the polls, government popularity does not explicitly account for the fact that governing parties can be electorally vulnerable to opposition parties. Further, government popularity considers governments as monolithic entities and fails to account for the possibility that, in coalition governments, vulnerability can come from the inside—junior coalition partners—and not only from the outside—main opposition parties.

Although government popularity has important limitations, we think that using vote intentions makes sense not only because polls are a reliable predictor of the electoral outcome, especially when elections are close (Jennings and Wlezien, 2016), but also because responsiveness occurs between elections (Manin et al., 1999; Narud and Esaiasson, 2013; Soroka and Wlezien, 2010). And so, we build on the government popularity approach and propose two alternative measures.

One measure is based on the assumption that governments do not only care about their own popularity, but also about how well the opposition is doing at the polls. The other measure builds on the literature on formateur selection and is based on the idea of the PM party as the government formateur and focuses on the concept of pivotality of the PM party, which becomes vulnerable when it loses this pivotal advantage.

These measures are validated against a traditional measure of popularity and applied to dynamic representation in Germany (1987–2005), using data on executive speeches collected by the Comparative Agendas Project (CAP). We find evidence of short-term effects of electoral vulnerability on agenda responsiveness, in that changes in government agendas respond to changes in public issue priorities conditional on changes in government’s and PM party’s vulnerability. Our results support previous studies on government responsiveness which find that electoral pressure affects responsiveness. However, our measures reflect more accurately the country’s specific situation than measures of government popularity. We discuss the implications of our measures and results in the conclusions.

Electoral vulnerability and responsiveness: Going beyond popularity

Policymakers’ responsiveness to public opinion is a key feature of representative democracy (Dahl, 1971). The opinion-policy link has been widely explored through dyadic representation to collective representation and dynamic representation. Extensive work on dynamic representation has recognized the importance of both public preferences and public priorities. This duplicity has generated two parallel lines of research. The first perspective looks at responsiveness in terms of position (e.g. Lax and Phillips, 2012; Soroka and Wlezien, 2010), while the second perspective, the one adopted in this article, looks at responsiveness in terms of attention (e.g. Bevan and Jennings, 2014; Mortensen et al., 2011). From previous research, we know that different policy issues promote different levels of responsiveness and that the latter also depends on issue salience. Policymakers are in fact more likely to respond on issues that are salient and important to the public, whereas, given the complexity and the amount of public demands, policymakers’ attention is scarce and varies by agenda (e.g. Jones and Baumgartner, 2005).

Competitive democratic theory posits that the key mechanism that brings about responsiveness, that is, “the need to respond,” is based on elections and its key assumption is that representatives aim to be re-elected. Politicians are obliged to take account of voters’ preferences and priorities in order to pursue their goal of vote maximization (Barry, 1970; Downs, 1957; but see Strøm 1990). Only if politicians are worried about the reactions of voters, they will be “constantly piloted by the anticipation of those reactions” (Sartori, 1977: 350). So, responsiveness is achieved by introducing Friedrich’s (1963) “mechanism of anticipated reactions.” If this mechanism relies on the desire of being re-elected, the incumbent will need to anticipate sympathetically voters’ preferences and demands. This mechanism will perform better if incumbents perceive themselves to be vulnerable (Bartolini, 1999; Fenno, 1977; Mayhew, 1974; Strøm, 1989).

The conditional effect of government electoral vulnerability on government responsiveness finds recognition in both theoretical (e.g. Bartolini, 1999, 2000; Sartori, 1987; Strøm, 1992) and empirical studies (e.g. Hakhverdian, 2010; Hobolt and Klemmensen, 2008; Manza and Cook, 2002; Pickup and Hobolt, 2015). Whereas the former argue that electoral vulnerability has a beneficial effect on responsiveness, the latter suggest that the vulnerability hypothesis is reflected in different institutional arrangements and argue that electoral pressure or uncertainty is a powerful incentive, increasing government responsiveness to citizens’ preferences and priorities. Note, however, that empirical evidence is far from being unanimous. For instance, other studies from the United States report no particular impact of presidential popularity on responsiveness to public concern (Cohen, 1995) and that “unpopular presidents are not more likely than popular ones to support positions endorsed by majority opinion” (Canes-Wrone, 2004: 487).

Scholars of responsiveness and dynamic representation assign a great deal of attention to the electoral pressures that governments face between elections. Studies analyzed pressures by selecting two main electoral incentives: how governments are doing in the polls (e.g. Canes-Wrone and Shotts, 2004; Hakhverdian, 2010; Hobolt and Klemmensen, 2008; Pickup and Hobolt, 2015) and how close or proximate elections are (e.g. Canes-Wrone and Shotts, 2004; Stimson et al., 1995). Government popularity is thus used as a proxy for capturing how potentially vulnerable governments are during the electoral cycle.²

While mostly focusing on cases of single-party governments, previous work does not effectively address the issue of vulnerability in coalition governments. Studies measuring government popularity use vote intentions and presidential approval as such. This approach is a good starting point but we think it is problematic, for it leaves some relevant issues unsolved. Under the popularity perspective, what counts is the government’s own popularity, no matter how the other parties are performing in the polls. We suspect that the argument in its support relies on the fact that governing parties do not care how their competitors are placed in the opinion polls; rather they only care about themselves and whether their own popularity goes up or down, and react accordingly. Yet, firstly, if government popularity declines, it does not necessarily mean that opposition popularity is going up, since voters might also prefer to abstain rather than reward opposition parties. Second, even if vote intentions for the government decline, the government might still be safe: vulnerability occurs when the potential success of main competitors is also included in the picture. In addition and more importantly, government popularity fails to consider different variations of vulnerability that come from governing status. That is, in coalition governments, the source of vulnerability can be internal rather than external.

We propose two measures, which both depend on the concept of uncertainty about future election outcomes (Elkins, 1974) and both use vote intentions. We call the first measure, government potential vulnerability (GPV), for it is based on the assumption that governments can feel uncertainty with respect to the possibility of losing the government at the next election. The second measure, formateur potential vulnerability (FPV), focuses on the idea that vulnerability ensues when the PM party stops holding the competitive advantage of being the government formateur. If the PM party is performing badly in the polls and its vote intentions decrease, it might nevertheless not be feeling vulnerable until it loses the potential of forming the government in favor of another party. These measures are described in the subsequent sections and further tested as a conditioning factor in a model of dynamic representation in Germany.

Government potential vulnerability

The measure of GPV emphasizes voters’ willingness to shift their vote. There is then no need to weight the measure by the strength or size of government and opposition, since it is already embedded in the vote intention. GPV is computed by subtracting the vote intentions for the relevant opposition parties from the vote intentions for the governing parties:

GPV (i t) = \sum Vote Intentions GOVT (i t) - \sum Vote Intentions OPP (i t)

where i is the party and t is time. Note that values below zero in the measure denote that the government is vulnerable and the more negative the values, the higher the vulnerability. The main question becomes which parties to include. For the government, the job is easy, as all parties in government should be considered. What is harder is defining what the relevant opposition is. By relevant opposition it means those parties receiving vote intentions, the government might be vulnerable from, including those who are not direct rivals in the competition for government but also those ones that can steal votes from governing parties. Since our second measure of FPV is quite restrictive, we aim to be as inclusive as possible with our measure of GPV as a kind of baseline measure. Hence, we consider relevant parties that are represented in parliament throughout the period of reference. This is not a perfect criterion but a simple decision to avoid ad hoc criteria of party selection.³ After all, governing parties compete for all votes, even if marginally. In Germany, this leaves us with CDU/CSU, SPD, FDP, Greens, and PDS/Die Linke (see Table 1). However, as some readers might consider this criterion too unrestrictive for some political systems where very small parties gain parliamentary representation, we replicated our models using two alternative selection criteria. One criterion is a quantitative adaptation of Sartori’s (1976) notable criteria of coalition potential and blackmail potential. According to this criterion, the party must either have been in government throughout the period of reference or won at least 5% of the votes and five seats in at least two elections (see Lühiste et al., 2017). Taking parties’ ability to gain media attention into account, such a decision has been already applied in other studies of representation (Bischof, 2018) and gives us exactly the same parties selected based on the parliamentary representation criterion. The other alternative criterion builds on the coalition formation research, especially with the ideas of size and incumbency status (e.g. Martin and Stevenson, 2001, 2010; Warwick, 1996). Based on this criterion, we include the largest opposition party in terms of vote shares and those parties that during the period of reference were at some point in office. Applied to Germany, this more restrictive criterion leaves us with the same parties with the exception of PDS/Die Linke. We report the analyses on agenda responsiveness based on these two alternative criteria in Table 1G, and our substantive results remain the same.

Table 1.

Measures of electoral vulnerability in Germany, 1987–2005.

	GPV		FPV (a)		FPV (b)		FPV (c)
	GOVT	OPP	PMPARTY	LARGESTJUNIOR	PMPARTY	LARGESTOPP	PMPARTY	LARGEST COMP
1987	CDU/CSU + FDP	SPD + Greens	CDU/CSU	FDP	CDU/CSU	SPD	CDU/CSU	SPD
1991	CDU/CSU + FDP	SPD + Greens + PDS/Linke	CDU/CSU	FDP	CDU/CSU	SPD	CDU/CSU	SPD
1994	CDU/CSU + FDP	SPD + Greens + PDS/Linke	CDU/CSU	FDP	CDU/CSU	SPD	CDU/CSU	SPD
1998	SPD + Greens	CDU/CSU + FDP + PDS/Linke	SPD	Greens	SPD	CDU/CSU	SPD	CDU/CSU
2002	SPD + Greens	CDU/CSU + FDP + PDS/Linke	SPD	Greens	SPD	CDU/CSU	SPD	CDU/CSU
2005	CDU/CSU + SPD	FDP + Greens + PDS/Linke	CDU/CSU	SPD	CDU/CSU	Greens	CDU/CSU	SPD

Note: The year refers to the time when a government took office. FPV: formateur potential vulnerability; GPV: government potential vulnerability.

Formateur potential vulnerability

Unlike GPV, the measure of FPV is based on the different assumptions that votes are not the only goal governing parties aim for. This is relevant given that most of European democracies have coalition governments. Hence, this measure accounts for the fact that government vulnerability can also depend on motivations other than re-election. Indeed, although incumbent governments have, for various reasons, a high chance of returning immediately to office, not all incumbents ‘should desire to re-form’ (Martin and Stevenson, 2010: 503).

Borrowing from the coalition formation research, the proposed argument is that a government would be considered safe to the extent that the PM party is still the formateur party.⁴ At the core of FPV, there is the idea of “pivotality.” We focus on the PM party because, unlike junior coalition partners, it is the most visible party in the coalition, the most likely to set the agenda and influence government policy, and the party that typically has the opportunity to call early elections. Being the most visible party in the coalition, voters are more likely to make judgments regarding the PM party than its junior coalition partners, as research on coalition heuristics shows (e.g. Fortunato and Stevenson, 2013), and so the safety of the PM party has important implications for the stability of the whole government, even though the PM party can decide to form a coalition with another party if the polls look favorably (see, e.g. Lupia and Strøm, 1995).

Although research on formateur selection proposes alternative criteria, we prioritize the “largest party status” in our measure. That is, the government becomes vulnerable at the time that the PM party is no longer the largest party in the opinion polls (equation 4).⁵ Together, we propose two variations of FPV, useful to detect the source of vulnerability: (a) the largest junior coalition partner or (b) the largest opposition party. Scenario (b) is crucial to apply the measure to single-party (minority) governments, given that in such cases scenario (a) would not be empirically feasible. The measure of FPV will then be operationalized in the following ways:

{FPV}_{i t} (a) = Vote Intentions PM (i t) - Vote Intentions LJUNIOR (i t)

{FPV}_{i t} (b) = Vote Intentions PM (i t) - Vote Intentions LOPP (i t)

{FPV}_{i t} (c) = Vote Intentions PM (i t) - Vote Intentions LCOMP (i t)

where PM is the PM party, LJUNIOR is the largest coalition partner (defined as the coalition partner with the highest vote intentions), LOPP is the largest opposition party (defined as the opposition party with the highest vote intentions), and LCOMP is the largest competitor in absolute terms (defined as the party with the highest vote intentions besides the PM party). Note, again, that values below zero in the measure denote that the PM party is vulnerable and the more negative the values, the higher the vulnerability.

An illustration: Agenda responsiveness in Germany (1987–2005)

We apply these vulnerability measures to Germany, which is a very versatile example for empirically testing the impact of government vulnerability on dynamic representation. Since one of our measures takes into consideration vulnerability of the PM party from its largest coalition partner, a case of coalition government illustrates the variants of vulnerability better than a case of single-party government such as the United States.

Allowing the formation of pre-electoral coalitions which produce a bipolar pattern conducive to high levels of cabinet durability (Saalfeld, 2005), the German case offers a high variability in terms of types of government, experiencing both multi-party coalition governments and grand coalitions.⁶ Table 1 and Figure 1 report how the measures of GPV and FPV are constructed and their level for the period of reference.

Figure 1.

Measures of government and formateur potential vulnerability in Germany, 1987–2005.

The left-upper panel shows the level of government vulnerability according to the GPV measure. Most of governments are below the zero line (which means maximal uncertainty) but not the SPD-Green coalition government between 1998 and 2002. The 2005 grand coalition makes the CDU-SPD government safe again, given the high vote intentions assigned in the polls. The right-upper side shows, instead, the level of vulnerability of the PM party against its junior coalition partner. The formateur is never really vulnerable, meaning that the PM party in Germany is never afraid of losing its pivotal advantage in favor of another coalition member. Differently, the lower panels show the vulnerability of the PM party against the largest opposition party (left-hand side) and against the largest absolute competitor (right-hand side). In Germany, the pattern of these two variants of FPV is similar to the pattern of GPV, except that in 2005 the largest competitor of the CDU is its junior coalition partner SPD.

Germany is also an ideal case for dynamic representation because of data availability issues. To measure dynamic representation data on public issue priorities and government agendas are used. To measure government agendas, data from the CAP on executive speeches (Regierungserklärungen) are available from 1987 to 2005 (Breunig and Schnatterer, N.d.). To measure public priorities, the most important problem (MIP) question is used. Time-series data on vote intentions and for the MIP question are available from the Politbarometer. The simultaneous availability of data on voting intentions, the MIP, and government agendas makes Germany an invaluable case to test the effects of government vulnerability on dynamic representation.

Executive speeches are delivered annually by the head of state or the head of government and are formal statements that set out the government’s agenda for the year ahead (Jennings et al., 2011). These speeches are available yearly and communicate the government’s general priorities as well as more specific measures that it plans to address. Hence, they are costly signals that ‘create future potential costs for the government, if the priorities in the speech are not followed by policy outputs’ (Bevan et al., 2011). Previous studies, in fact, document the translation of governments’ policy agendas into legislative outputs in the United States (Edwards and Wood, 1999) and the United Kingdom (Bara, 2005; Bevan et al., 2011) and, moreover, comparative research documents that executive speeches reflect the issues governing parties emphasize in other venues, suggesting that these speeches are reasonable proxies for the government’s more general rhetorical emphases (Green-Pedersen et al., 2018). Similar to the Comparative Manifestos Project’s codings of party manifestos (Budge et al., 2001), the CAP coding scheme takes the quasi-sentences in executive speeches as the unit of analysis, with each quasi-sentence assigned a single topic code.

For our analysis, we can rely on the following CAP major topics: macroeconomics (1), health (3), education (6), environment (7), law and crime (12), social welfare (13), housing (14), defense (16), and international affairs and foreign trade (19) (see Table 1H for summary statistics).⁷

The model

A regression model is specified to evaluate the effect of vulnerability on responsiveness. Scholars’ attention has recently moved to applications of error correction models (ECMs) to time-series data. Using ECMs became common practice particularly in the studies of dynamic representation (e.g. Bevan and Jennings, 2014; Jennings and John, 2009), for they allow one to estimate both short-term and long-term effects (De Boef and Keele, 2008) of changes in public opinion on government activity. The choice in favor of an ECM is also methodologically appropriate, given that unit root tests reveal that not all executive speech series are stationary and the null hypothesis that all the panels contain a unit root cannot be rejected.⁸ Since the first-difference models often perform poorly and throw out long-run effects, the adoption of an ECM is a valuable solution (e.g. Beck and Katz, 2011).

Another reason why ECMs are appropriate for these data is the presence of time dependencies in the data. In fact, visual inspection of autocorrelation and partial autocorrelation functions suggests the first-order autocorrelation in the dependent variable. The pooled models are estimated with panel-corrected standard errors (PCSEs) (Beck and Katz, 1995), which control for panel heteroscedasticity and contemporaneous correlations of the errors.

The model specifies overtime changes with emphasis on executive speeches as a function of the levels of (and changes in) public issue priorities. The dependent variable, [Δ Govt Speech (t)], is the difference between issue emphasis in the executive speech in the current year and issue emphasis in the executive speech in the previous year, that is, positive values on the dependent variable denote that the government’s issue emphasis on the issue has increased over the past year. The independent variables are [Govt Speech (t − 1)], government’s issue emphasis in the executive speech in the preceding year; [Public Priorities (t − 1)], the proportion of the previous year’s public issue emphasis; [Δ Public Priorities (t)], the change in public issue emphasis in the current year compared to public issue emphasis in the previous year; [Vulnerability (t − 1)], the level of vulnerability (measured as GPV or FPV) in the previous year; and [Δ Vulnerability (t)], the change in current vulnerability (measured as GPV or FPV) compared to its vulnerability in the previous year. The following pooled model is estimated yearly over all the governments in the study:

\begin{array}{l} Δ Govt Speech (t) = α 0 + α 1 [Govt Speech (t - 1)] \\ + β 1 [Δ Public Priorities (t)] + β 2 [Public Priorities (t - 1)] \\ + β 3 [Δ Vulnerability (t)] + β 4 [Vulnerability (t - 1)] \\ + β 5 [Δ Public Priorities (t) \times Δ Vulnerability (t)] \\ + β 6 [Public Priorities (t - 1) \times Vulnerability (t - 1)] \\ + β 7 [Govt Ideology (t)] \end{array}

To evaluate the vulnerability hypothesis, the key coefficients are those on the interaction between [Δ Public Priorities (t)] and [Δ Vulnerability (t)] variables, for short-term effects, and the interaction between the [Public Priorities (t − 1)] and [Vulnerability (t − 1)] variables, for long-term effects. A negative coefficient β5 on the short-term interaction between public priorities and government vulnerability would denote that an increase in public priorities on a given issue in the current year—compared to the previous year—is associated with an increase in issue emphasis in government speeches conditional on government vulnerability, that is, that the government responds to short-term changes in public priorities when vulnerable. Similarly, a negative coefficient β6 on the long-term interaction between public priorities and government vulnerability would denote that an increase in public priorities on a given issue in the previous year is associated with an increase in issue emphasis in government speeches conditional on government vulnerability, that is, that the government responds to public priorities in the previous year when vulnerable.

To assess dynamic representation independently of the conditional effect of vulnerability, the coefficients of interest are β1 and β2. A positive and significant coefficient β1 on the [Δ Public Priorities (t)] variable would denote that increases in public priorities in the current year (compared to the previous year) are associated with increased government emphasis in its speech, a short-term effect, while a positive and significant coefficient β2 on the [Public Priorities (t−1)] variable would denote that the more an issue becomes important to the public at the previous time period, the more the government emphasizes it in its speech at the current period.

A negative and significant coefficient β3 [Δ Vulnerability (t)] variable would denote that an increase in government vulnerability in the current year (compared to the previous year) is associated with increased government emphasis in its speech, a short-term effect, while a negative and significant coefficient β4 on the [Vulnerability (t−1)] variable would denote that the more the government is vulnerable at the previous time period, the more the government emphasizes a given issue in its speech at the current period.

The model specification also includes the government’s lagged issue emphasis in its speech, [Govt Speech (t−1)], to control for the government’s long-term level of issue emphasis in its speech and to evaluate whether governments that were emphasizing a given issue at the previous time period tend to emphasize it less at the current time period. The model also controls for government ideology. Including the variable [Govt Ideology (t)] in the equation allows testing the mechanism through which public opinion influences policy (see Hakhverdian, 2010: 849–850).⁹

Finally, in the empirical analyses, years when a new government emerged whose ideology differed from the previous government, such as Schroeder I in 1998 and Merkel I in 2005, are omitted. This is because in these years the lagged and current levels of government rhetoric pertain to different governments, so that the public plausibly does not hold the current government responsible for the previous government’s lagged behavior. However, successive governments with the same Chancellor are considered as the same.

Results

Table 2 reports the parameter estimates (with PCSEs in parentheses) for the pooled model given by equation (5) above, estimated over all issues. Column 2 reports the parameter estimates computed using a common measure of government popularity, column 3 reports the parameter estimates computed using the GPV measure, and columns 4–6 report the parameter estimates computed using the FPV measures.¹⁰ Before turning to effects pertaining to public priorities and government vulnerability, note that the coefficient on the variable [Govt Speech (t − 1)] is negative and significant in all four sets of analyses, while the coefficient on the intercept is positive, which implies a “regression to the mean” in government issue emphasis, that is, when government issue emphasis was unusually high (low) at the previous time period, then emphasis tended to subsequently decline (increase) at the current period.

Table 2.

The vulnerability hypothesis in Germany (1987–2005).

	POP	GPV	FPV (a)	FPV (b)	FPV (c)
Govt Speech (t − 1)	−1.015*** (0.161)	−1.018*** (0.161)	−1.016*** (0.159)	−1.006*** (0.158)	−1.006*** (0.158)
Δ Public Priorities (t)	0.184* (0.104)	0.184* (0.103)	0.175* (0.104)	0.192* (0.103)	0.192* (0.103)
Public Priorities (t − 1)	0.386 (0.306)	0.198*** (0.053)	0.185 (0.166)	0.194*** (0.051)	0.194*** (0.051)
Δ Vulnerability (t)	0.013 (0.048)	0.010 (0.025)	0.031 (0.040)	0.006 (0.023)	0.006 (0.023)
Vulnerability (t − 1)	−0.119 (0.077)	−0.067 (0.042)	−0.109** (0.055)	−0.062 (0.039)	−0.062 (0.039)
Δ Public Priorities (t) × Δ Vulnerability (t)	−0.029* (0.015)	−0.018** (0.009)	−0.032* (0.017)	−0.021** (0.010)	−0.021** (0.010)
Public Priorities (t − 1) × Vulnerability (t − 1)	−0.004 (0.007)	−0.001 (0.003)	0.000 (0.005)	−0.001 (0.003)	−0.001 (0.003)
Govt Ideology (t)	0.915* (0.507)	0.848* (0.481)	0.147 (0.491)	0.331 (0.437)	0.331 (0.437)
Constant	7.215** (3.452)	1.499*** (0.572)	5.627*** (1.788)	1.826*** (0.498)	1.826*** (0.498)
N	128	128	128	128	128
R ²	0.58	0.58	0.56	0.58	0.58

Note: The coefficients are reported with PCSEs in parentheses. Column 2 compares the government popularity (POP) model with the government vulnerability models (GPV and FPV (a), (b), (c)). The variables are defined in the text. FPV: formateur potential vulnerability; GPV: government potential vulnerability; PCSE: panel-corrected standard errors.

*p < 0.10; **p < 0.05; ***p < 0.01.

We next consider the effects of public priorities and government vulnerability. If dynamic representation in Germany occurs, we would expect significant coefficient estimates on either (or both) of the variables [Δ Public Priorities (t)] and [Public Priorities (t − 1)]. This is indeed the case.¹¹ If government vulnerability was equal to zero, the coefficient estimates on these variables, in combination with their standard errors, would imply that public issue salience has both short-term and long-term positive and significant effects on government emphasis in its executive speeches, that is, that the government responds in its policy agendas to both changes in and past levels of public priorities. The analysis also shows that there is no direct effect of electoral vulnerability—either in the short term or in the long term—on changes in government emphasis in executive speeches, in that that coefficients of the variables [Δ Vulnerability (t)] and [Vulnerability (t−1)] are insignificant at conventional levels, except in one case.

The results on the vulnerability hypothesis are now considered. If electoral vulnerability had a beneficial effect on government responsiveness, a negative and significant coefficient on either (or both) of the interacted variables [Δ Public Priorities (t) × Δ Vulnerability (t)] and [Public Priorities (t − 1) × Vulnerability (t − 1)] would be expected. According to the level of vulnerability based on the different measures proposed, it has been argued above that a stronger effect of both government vulnerability and the PM party vulnerability on government responsiveness would be expected in all but one scenario, namely in the case of vulnerability of the PM party against its junior coalition partner, which in Germany seems to be none. While the empirical analysis finds no long-term effects of potential vulnerability on government responsiveness in any of the vulnerability measures, the vulnerability hypothesis is, instead, supported in the short run. There is, indeed, evidence of short-term effects of electoral vulnerability on government responsiveness in all models, but differences apply.

The conditional effect of the interacted variable [Δ Public Priorities (t) × Δ Vulnerability (t)] is negative and weakly significant (p < 0.10) for the vulnerability of the PM party against its largest junior coalition partner (model FPV (a)), which is in line with our expectations. The effect is instead stronger (p < 0.05) for the vulnerability of the government as a whole against opposition (model GPV), the vulnerability of the PM party against its largest opposition party (model FPV (b)), as well as against its strongest competitor (model FPV (c)).

Column 2 reports the results for the government popularity (POP) measure, where neither junior coalition partners nor opposition parties are explicitly taken into account. In comparison with our vulnerability measures, short-term effects of government popularity on government responsiveness to public priorities are also present. The coefficient in the interactive variable is in fact in the right direction, namely, lower government popularity is associated with higher responsiveness. However, it is worth noting that the coefficient of the interaction is only statistically significant at p < 0.10.

Figure 2 presents the marginal effects of the [Δ Public Priorities (t) × Δ Vulnerability (t)] variable on the [Δ Govt Speech (t)] variable, computed for the coefficient estimates reported in Table 1. According to the plot, an increase in public priorities on a given issue in the current year—compared to the previous year—is associated with an increase in issue emphasis in government speeches conditional on government vulnerability. The effect of vulnerability on responsiveness stops being statistically significant at around zero, suggesting that only negative and not positive changes in vulnerability are associated with an increase in government responsiveness to public priorities.

Figure 2.

Effects of electoral vulnerability on agenda responsiveness in Germany (1987–2005), by measure. Note: The figure displays the marginal effects of the [Δ Public Priorities (t) × Δ Vulnerability (t)] variable on the [Δ Govt Speech (t)] variable, computed for the coefficient estimates reported in Table 2. The dashed lines denote 95% confidence intervals. The variables are defined in the text.

To sum up, our findings show that there is a relationship between electoral pressures and government responsiveness to public priorities in the German Chancellor’s policy agenda. The way these electoral pressures are measured leads to somewhat different results. In particular, short-term effects of electoral vulnerability on responsiveness are stronger when the whole government faces the major opposition and when the PM party faces the largest opposition party or the largest competitor, which in the period covered by our analysis always coincide. Short-term effects are, instead, weaker with a classic measure of government popularity and when the PM party faces its junior coalition partner, which in Germany never really constitutes a big threat to the Chancellor party.

We conducted additional analyses to assess the robustness of our findings. First, the models have been reestimated by including a trend variable that controls for time effects (Table 1C) and by including policy dummies to control for policy effects (Table 1D). Second, since the state of the economy might have an impact on government policy, we reestimated the models while controlling for national levels of unemployment and inflation, along with changes in these levels (Table 1E). These analyses, which are reported in the SM memo, continue to support our substantive conclusions. Third, since the causal relationship between opinion and policy can also be reciprocal (e.g. Soroka and Wlezien, 2010), in that popular governments can lead public opinion (Hakhverdian, 2012), the counter movement hypothesis is also tested (Table 1F). No evidence, however, is found in Germany that governments manipulate the public by influencing issue priorities through their policy agendas when they are popular.

Conclusion and discussion

Governing parties’ behavior in opinion polls is important not only for election outcomes but also for policy and representation. Our article makes some contributions for the latter. First, work on responsiveness tended to capture dynamics in electoral pressures using a measure of government popularity. We discussed the limitations of this approach and went beyond by proposing two more fine-grained measures of vulnerability using opinion polls that can be easily replicated by political analysts for undertaking comparative research. This was the main goal of the article. Second, our contribution is also empirical. Both measures are tested on the framework of dynamic representation with data from the CAP on executive speeches in Germany (1987–2005). Our findings are the following.

First of all, dynamic representation in Germany works, whereby past public issue priorities influence changes in issue emphasis in German government agendas (Table 1B), and this is in line with previous findings on agenda responsiveness (e.g. see Bevan and Jennings, 2014). More importantly for our purposes, we find short-term effects of electoral vulnerability on agenda responsiveness to public priorities. In particular, we find that German coalition governments are more under siege when pressure comes from outside rather than from inside the government. That is, the PM party in Germany is never really vulnerable against its junior coalition partner. Interestingly, government attention to public priorities is not much conditional on the level of government vulnerability, when the latter is measured in relation to the performance of the junior coalition partner at the polls or when traditional measures of popularity are used. The electoral incentive, instead, occurs when vulnerability of the PM party is measured as a function of the strength of the major opposition party. In this case, in fact, vulnerability does have a short-term effect on government responsiveness and governing parties adjust their policy agendas to public priorities. Hence, we think our measures depict the effect of vulnerability on responsiveness more reliably than the effect of government popularity on responsiveness.

We believe that our measures can find application in other research areas such as the one of coalition heuristics (e.g. Fortunato and Stevenson, 2013) to answer some of the questions arising from coalition politics and its consequences on voters. For instance, recent research by Sagarzazu and Klüver (2017) finds that coalition parties need to compromise in order to maintain the coalition and, at the same time, need to differentiate from their partners to strengthen their own policy profile. On the same line, Klüver and Spoon (2016) find that, as conflict over an issue among coalition partners increases, parties will pay less attention to voters’ issue priorities. Would, then, the loss of the formateur advantage have consequences for the longevity and unity of the coalition? Would this have consequences for voters’ perceptions of coalition partners’ positions?

Our article comes with some limitations. Although agenda responsiveness in Germany represents an illustrative case for our measures, our findings are not easily generalizable and are limited to the data at our disposal. Consistent with previous studies, our findings are based on issues that are considered salient in the public sphere but cannot say much for low or not salient issues. However, previous research by Hobolt and Klemmensen (2008) and Hakhverdian (2010) using executive speeches found evidence that popularity has an impact on agenda responsiveness in different contexts and in the expected direction. We want to avoid speculations but, taken together, our findings and their findings seem suggesting that there is some scope for electoral incentives influencing dynamic representation in symbolic/rhetorical policy venues.

Future research should study whether electoral vulnerability of different coalition partners influences their responsiveness differently and whether the vulnerability effect varies across the election cycle, such as in the campaign (Bevan and Krewel, 2015). The German case showed that government attention responds when vulnerability comes from outside the government. Future research should also explain the source of vulnerability in countries with more fragmented party systems where more than one coalition partner is in government, and whether the potential of losing the pivotality advantage would be sufficient for governments to respond.

Supplemental Material

Supplemental Material, sj-pdf-1-ppq-10.1177_1354068818790117 - From popularity to vulnerability: An application to dynamic representation in coalition governments

Supplemental Material, sj-pdf-1-ppq-10.1177_1354068818790117 for From popularity to vulnerability: An application to dynamic representation in coalition governments by Luca Bernardi in Party Politics

Footnotes

Acknowledgements

The author is grateful for European Research Council. The author is especially grateful to Christian Breunig for his generosity in sharing data on executive speeches, without which this article would not have been possible. This research benefitted from excellent feedback at various levels from a number of colleagues. In particular, the author is thankful to Jim Adams, Daniel Bischof, Mark Franklin, Sara Hobolt, Heike Klüver, Laura Morales, Oriol Sabaté, and Francesco Visconti. The author thanks the three anonymous reviewers for their feedback that substantively improved the article and the editors of the journal for supporting this research.

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research leading to these results has received funding from the European Research Council under the European Commission’s Seventh Framework Programme through a Starting Grant (FP7/2007-2013 grant agreement 284277) to the project “Democratic Responsiveness in Comparative Perspective: How Do Democratic Governments Respond to Different Expressions of Public Opinion? (ResponsiveGov)” () led by Prof Laura Morales.

Supplemental material

Supplemental material for this article is available online.

Notes

References

Abou-Chadi

Orlowski

(2016) Moderate as necessary: the role of electoral competitiveness and party size in explaining parties’ policy shifts. The Journal of Politics 78: 868–881.

Adams

(2012) Causes and electoral consequences of party policy shifts in multiparty elections: theoretical results and empirical evidence. Annual Review of Political Science 15(1): 401–419.

Anderson

(1995) Blaming the Government: Citizens and the Economy in Five European Democracies. Armonk: M. E. Sharpe.

André

Depauw

Martin

(2014) Electoral systems and legislators’ constituency effort: the mediating effect of electoral vulnerability. Comparative Political Studies 48: 1–33.

Bäck

Dumont

(2007) Combining large-n and small-n strategies: the way forward in coalition research. West European Politics 30(3): 467–501.

Bäck

Dumont

(2008) Making the first move: a two-stage analysis of the role of formateurs in parliamentary government formation. Public Choice 135(3–4): 353–373.

Bara

(2005) A question of trust: implementing party manifestos. Parliamentary Affairs 58(3): 585–599.

Barry

(1970) Sociologists, Economists and Democracy. Chicago: University of Chicago Press.

Bartolini

(1999) Collusion, competition and democracy: part I. Journal of Theoretical Politics 11(4): 435–470.

10.

Bartolini

(2000) Collusion, competition and democracy: part II. Journal of Theoretical Politics 12(1): 33–65.

11.

Beck

Katz

(1995) What to do (and not to do) with time-series cross-section data. The American Political Science Review 89(3): 634–647.

12.

Beck

Katz

(2011) Modeling dynamics in time-series-cross-section political economy data. Annual Review of Political Science 14(1): 331–352.

13.

Bevan

John

Jennings

(2011) Keeping party programmes on track: the transmission of the policy agendas of executive speeches to legislative outputs in the United Kingdom. European Political Science Review 3(3): 395–417.

14.

Bevan

Jennings

(2014) Representation, agendas and institutions. European Journal of Political Research 53(1): 37–56.

15.

Bevan

Krewel

(2015) Responsive elections: the effect of public opinion on political campaigns. Electoral Studies 40: 548–555.

16.

Bischof

(2018) Ideological congruence between party rhetoric and policy-making. West European Politics 41(2): 310–328.

17.

Breunig

Schnatterer

(N.d) German Policy Agendas - Data Set and Descriptive Insights. Working Paper Uni Konstanz.

18.

Budge

Klingemann

H-D

Volkens

, et al. (2001) Mapping Policy Preferences. Estimates for Parties, Electors, and Governments 1945-1998. Oxford: Oxford University Press.

19.

Canes-Wrone

(2004) The public presidency, personal approval ratings, and policy making. Presidential Studies Quarterly 34(3): 477–492.

20.

Canes-Wrone

Shotts

(2004) The conditional nature of presidential responsiveness to public opinion. American Journal of Political Science 48(4): 690–706.

21.

Cohen

(1995) Presidential rhetoric and the public agenda. American Journal of Political Science 39(1): 87–107.

22.

Dahl

(1971) Polyarchy: Participation and Opposition. New Haven: Yale University Press.

23.

De Boef

Keele

(2008) Taking time seriously. American Journal of Political Science 52(1): 184–200.

24.

Downs

(1957) An Economic Theory of Democracy. New York: Harper and Row.

25.

Edwards

GC III

Wood

(1999) Who influences whom? The president, congress, and the media. The American Political Science Review 93(2): 327–344.

26.

Elkins

(1974) The measurement of party competition. The American Political Science Review 68(2): 682–700.

27.

Fenno

Jr (1977) U.S. house members in their constituencies: an exploration. The American Political Science Review 71(3): 883–917.

28.

Fortunato

Stevenson

(2013) Perceptions of partisan ideologies: the effect of coalition participation. American Journal of Political Science 57(2): 459–477.

29.

Friedrich

(1963) Man and his Government. New York: McGraw Hill.

30.

Green-Pedersen

Mortensen

(2018) The agenda-setting power of the prime minister party in coalition governments. Political Research Quarterly OnlineFirst.

31.

Hakhverdian

(2010) Political representation and its mechanisms: a dynamic left-right approach for the United Kingdom, 1976-2006. British Journal of Political Science 40: 835–856.

32.

Hakhverdian

(2012) The causal flow between public opinion and policy: government responsiveness, leadership, or counter movement? West European Politics 35(6): 1386–1406.

33.

Hobolt

Klemmensen

(2008) Government responsiveness and political competition in comparative perspective. Comparative Political Studies 41(3): 309–337.

34.

Immergut

Abou-Chadi

(2014) How electoral vulnerability affects pension politics: introducing a concept, measure and empirical application. European Journal of Political Research 53(2): 269–287.

35.

Jennings

John

(2009) The dynamics of political attention: public opinion and the Queen’s speech in the United Kingdom. American Journal of Political Science 53(4): 838–854.

36.

Jennings

Wlezien

(2016) The timeline of elections: a comparative perspective. American Journal of Political Science 60(1): 219–233.

37.

Jennings

Bevan

John

(2011) The agenda of British government: the Speech from the Throne, 1911-2008. Political Studies 59(1): 74–98.

38.

Jones

Baumgartner

(2005) The Politics of Attention: How Government Prioritizes Problems. Chicago: University of Chicago Press.

39.

Kayser

Lindstädt

(2015) A cross-national measure of electoral competitiveness. Political Analysis 23: 242–253.

40.

Klüver

Spoon

J-J

(2016) Challenges to multiparty governments how governing in coalitions affects coalition parties’ responsiveness to voters. Party Politics 23: 793–803.

41.

Lax

Phillips

(2012) The democratic deficit in the states. American Journal of Political Science 56: 148–166.

42.

Lupia

Strøm

(1995) Coalition termination and the strategic timing of parliamentary elections. The American Political Science Review 89(3): 648–665.

43.

Lühiste

Morales

Bernardi

, et al. (2017) ResponsiveGov’s codebook and appendices. DOI: 10.7910/DVN/PINCO4, Harvard Dataverse, V1, UNF:6: wlfV9HEyxkYtEGXMcITAdQ==

44.

Manin

Przeworski

Stokes

(1999) Elections and representation. In: Adam Przeworski

Stokes Manin

(eds) Democracy, Accountability, and Representation. Cambridge: Cambridge University Press, pp. 29–54.

45.

Manza

Cook

(2002) The impact of public opinion on public policy: the state of the debate. In: Jeff

Lomax

Page

(eds) Navigating Public Opinion: Polls, Policy and the Future of American Democracy. Oxford: Oxford University Press, pp. 17–32.

46.

Martin

Stevenson

(2001) Government formation in parliamentary democracies. American Journal of Political Science 45(1): 33–50.

47.

Martin

Stevenson

(2010) The conditional impact of incumbency on government formation. American Political Science Review 104(3): 503–518.

48.

Mayhew

(1974) Congress: The Electoral Connection. New Haven: Yale University Press.

49.

Mortensen

Green-Pedersen

Breeman

, et al. (2011) Comparing government agendas: executive speeches in the Netherlands, United Kingdom, and Denmark. Comparative Political Studies 44: 973–1000.

50.

Narud

Esaiasson

(eds) (2013) Between-Election Democracy. The Representative Relationship After Election Day. Colchester: ECPR Press.

51.

Pickup

Hobolt

(2015) The conditionality of the trade-off between government responsiveness and effectiveness: the impact of minority status and polls in the Canadian house of commons. Electoral Studies 40: 517–530.

52.

Saalfeld

(2005) Germany: stability and strategy in a mixed-member proportional system. In: Michael

Mitchell

(eds) The Politics of Electoral Systems. Oxford, New York: Oxford University Press, pp. 209–229.

53.

Sagarzazu

Klüver

(2017) Coalition governments and party competition: political communication strategies of coalition parties. Political Science Research and Methods 5(2): 333–349.

54.

Sartori

(1976) Parties and Party Systems: A Framework for Analysis. Cambridge: Cambridge University Press.

55.

Sartori

(1977) Democrazia competitiva ed elites politiche. Rivista Italiana di Scienza Politica 7: 327–355.

56.

Sartori

(1987) The Theory of Democracy Revisited. Chatham: Chatham House Publishers, Inc.

57.

Soroka

Wlezien

(2010) Degrees of Democracy: Politics, Public Opinion, and Policy. New York: Cambridge University Press.

58.

Spoon

J-J

Klüver

(2014) Do parties respond? How electoral context influences party responsiveness. Electoral Studies 35: 48–60.

59.

Stimson

Mackuen

Erikson

(1995) Dynamic representation. The American Political Science Review 89(3): 543–565.

60.

Strøm

(1989) Inter-party competition in advanced democracies. Journal of Theoretical Politics 1(3): 277–300.

61.

Strøm

(1990) A behavioral theory of competitive political parties. American Journal of Political Science 34(2): 565–598.

62.

Strøm

(1992) Democracy as political competition. American Behavioral Scientist 35 (4/5): 375–396.

63.

van der Eijk

Oppenhuis

(1991) European parties’ performance in electoral competition. European Journal of Political Research 19(1): 55–80.

64.

Warwick

(1996) Coalition government membership in west European parliamentary democracies. British Journal of Political Science 26(04): 471.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.20 MB