Between Accommodating and Activating: Framing Policy Reforms in Response to Workforce Aging across Europe.

In the past decade, European governments have implemented activating policy reforms to maximize older workers' employment and employability, representing a paradigmatic change in approaches to work and retirement. This study isolates the factors that explain the relative success and failure of competitive frames that are either in favor of or against activating policies in European news coverage, by applying time-series analysis (ordinary least squares with panel-corrected standard errors) to monthly aggregated news coverage in the Netherlands, the United Kingdom, Denmark, and Spain over the timespan 2006-2013. The results show that pro-activating and counteractivating frames generally coincide in competitive framing environments. The pro-activating frame proliferated in times of high employment protection, whereas the counteractivating frame prevailed stronger in conservative compared with progressive newspapers, and gained momentum during the aftermath of the financial crisis and in times governments on the economic left were in power. The study advances knowledge of competitive issue framing by demonstrating how the economic, policy, and political context matters for the emergence and evolvement of competing frames. In addition, the findings contribute to the understanding of the factors that contribute to news representations that promote active aging in European news, which may foster support for policy reforms that sustain older workers' employability.

Foreseeing population aging, European governments placed the issue of older workers’ sustainable employability in the mid-1990s on the policy agenda. Since then, policy reforms have focused on activating older workers by replacing accommodating measures (e.g., decreasing workload and offering possibilities for early retirement) with activating policies (e.g., prolonging working lives and stimulating job mobility) (Organisation for Economic Co-operation and Development [OECD] 2006; van Selm and Van der Heijden 2013). Despite such substantial policy changes, today, European countries continue to face serious challenges in retaining and sustaining older workers’ employability. Recent studies suggest that older workers still intent to retire before the formal retirement age (Hofäcker 2015), whereas European employers generally do not promote active aging (see Van Dalen et al. 2015).

Support for activating policy reforms is increasingly important to maximize and sustain the employability of aging workers (Hofäcker 2015). Compulsory measures imposed by governments, such as raising the retirement age, yield little effect if employers remain unwilling to hire or retain older people and if individual workers do not want to work longer (Van Echtelt et al. 2014). The successful implementation of policy reforms thus largely hinges upon public support from (older) workers as well as employers.

The news media serve as an important arena where opponents and proponents compete for perspectives in the battle for public opinion (Fowler et al. 2012). As the news media are key in informing the general public about current affairs, they can have a strong impact on public knowledge of, and support for, specific policy proposals (Chong and Druckman 2007; Vliegenthart 2012). The news environment, then, fulfills a crucial role in reinforcing or challenging public beliefs about the necessity of pro-activating policy reforms in the light of sustaining older workers’ employability, with consequences for the successfulness of such reforms.

Although all European countries have to deal with workforce aging, large between-country differences exist in response toward the European Union’s “active aging paradigm” (Conen 2013). This makes it a particularly interesting case to investigate how news coverage about the issue is shaped by outlet characteristics (i.e., newspapers’ political leaning) and factors that vary across time and countries (i.e., temporal factors and the political and policy climate). The current study traces the influence of these sources of variation in attention for the issue of workforce aging and variation in competitive frames that are either in favor of or against activating policies.

The resulting analysis makes important contributions to the understanding of why different European news media frame activating policy reforms in the way they do. Situated against the backdrop of a fiercely debated and often-disputed issue, our case makes a unique contribution to the literature on competitive framing, which has predominantly focused on framing effects at the expense of examining the sources that explain competitive frame variation (but see Fowler et al. 2012). This study advances knowledge in this area by conceptualizing competing frames as dynamic and by tracing the factors that contribute to the failure or success of these frames over time. The data come from a content analysis of conservative and progressive newspapers in the Netherlands, the United Kingdom, Denmark, and Spain (2006–2013).

Policy Responses toward Workforce Aging across Europe

“Active aging” stresses the rights of older people to maintain their participating role in the community and political life, by remaining healthy and employed (Foster and Walker 2015). With respect to workforce aging, active aging policies are aiming at extending working lives and labor market participation of older workers (OECD 2006), and vary from increasing the formal retirement age, facilitating work continuation, reducing early exit routes from the labor market, and lowering benefits schemes for older workers to enhance their job mobility (Hofäcker 2015; OECD 2006). In this study, we refer to this set of policy types as pro-activating policies.

Accommodative and protective measures may stand in the way of achieving the goal of active aging (Remery et al. 2003; Van Dalen et al. 2015). First, accommodative measures “compensate for the possible fall in physical and cognitive capacities that accompany the process of aging” (Van Dalen et al. 2015: 816). This compensation includes reductions of working hours, decreasing workload, offering possibilities for additional leave or early retirement, and exemptions from working irregular shifts and overtime for older workers. Although such measures are often framed as “age conscious” policies, they paradoxically limit older workers’ opportunities (Conen 2013; Euwals et al. 2009), as they tend to incur additional costs without generating returns for employers (Remery et al. 2003) and depreciate the human capital value of older workers (Kooij and Van de Voorde 2015). Second, generous protective policies are argued to contribute to the cost–productivity gap that employers associate with the aging of the workforce (Conen et al. 2012). This, in turn, negatively harms both the recruitment and retention of older workers (Conen et al. 2012; Saint-Paul 2009). The aforementioned accommodative and protective measures are in this study referred to as counteractivating measures.

Attention, Pro-Activating, and Counteractivating Frames: The Dependent Variables

To investigate news representations of activating policy reforms, we rely on framing theory. Framing theory has been frequently used to study how news media represent issues. The theory’s major premise is that by presenting an event or issue in a particular way, media may shape public opinion (De Vreese 2005). Although there is wide variation in the conceptualization of the framing concept, most studies focus on elements that are part of standard definitions of framing, such as a diagnosis (what is the problem?) and prognosis (what should the solution look like?) (Entman 1993; Snow et al. 2007). In this study, we follow this approach and focus on two competing issue frames: pro- and counteractivating policy frames. Both frames are presented as the appropriate manner to deal with the issue of workforce aging and defined in terms of the advantages and disadvantages that policy reforms are argued to have.

A key goal of the study is to identify the causal mechanisms that drive variation in these two issue frames, as well as attention to the issue of workforce aging more generally. The selection of our independent variables is ingrained in theoretical insights on the construction of news messages (Reese 2001; Scheufele 1999; Scheufele and Tewksbury 2007; Shoemaker and Reese 1996). According to this literature, news content is the outcome of both inter- and extra-media factors (Shoemaker and Reese 1996). Most empirical studies, however, do not probe the wider political, economic, and policy context in which frame contests emerge (Carragee and Roefs 2004).

The current study theoretically acknowledges and empirically models the larger societal structure in which journalists operate. We identify four determinants of the attention devoted to the issue of workforce aging, and its framing in pro- and counteractivating terms: (1) the ideological orientation of newspapers, (2) temporal aspects, (3) governments’ stance on economic issues, and (4) the policy arena.

Sources of Variation in Attention and Competitive Frames: The Antecedents

Newspapers’ ideology

First, we consider the influence of the newspapers’ ideology. Specifically, we look at the difference between progressive (characterized by advancement and progression) and conservative (seeking to preserve traditional social institutions) newspapers (Berkowitz 1997).

It is argued that conservative forces more strongly reject pro-activating reforms for the following reasons. For generations, it has been common practice in several European countries to retire before the formal retirement age, and to apply accommodation strategies for aging workers. Pro-activating policy reforms put an end to this. Ultimately, this breaks the promise of retirement and accommodation at early age, causing older workers to feel that they do not receive what they deserve (Henkens et al. 2016). This makes pro-activating policy reforms generally unpopular (Hofäcker 2015)—and arguably especially among conservatives as pro-activating reforms pose a threat to the status quo and traditional labor relations. Conservatives are generally more likely to advocate the retrenchment of new policies (such as pro-activating reforms) and oppose “new” social policy reforms (Häusermann 2006, 2012).

Following this reasoning, it is expected that counteractivating arguments resonate with conservative news media ideologies. Journalists’ ideological stance affects how issues are framed in the media (Shoemaker and Reese 1996). Journalists with a conservative ideology are argued to have “respect for tradition” and said to more fiercely “defend the social order” (Gans 1979: 179). These value systems, in turn, “come through clearly in the news” (Shoemaker and Reese 1996: 79)—causing journalists, generally unconsciously, to put forward a perspective that resonates with their personal and organizational values. It is expected that compared with progressive-leaning newspapers, conservative-leaning newspapers address workforce aging (a) less often in pro-activating terms and (b) more often in counteractivating terms (Hypothesis 1 [H1]).

Temporal influence

Second, we take temporal aspects into account. Several scholars have demonstrated that the occurrence of significant, real-world events makes the content of the news fairly predictable (Brosius and Eps 1995; De Vreese 2001; Vliegenthart and Boomgaarden 2007). Such key events generally represent the unusual and can often be characterized by significant societal importance. The combination of these characteristics makes these events highly newsworthy and attractive to journalists (Eilders 2006; Harcup and O’Neill 2001).

Here, the focus is on the influence of the financial crisis. This event is clearly one of the most significant events that occurred during the research period in the countries under study. Unemployment rates accelerated rapidly in the second phase of the financial crisis (2012/2013)—the so-called “double dip”—especially in the Netherlands and Spain (Van den Berge et al. 2014). In particular, long-term unemployed workers, among which older workers are overrepresented, experienced difficulties in finding new employment (De Graaf-Zijl et al. 2015). It can be anticipated that the excessive unemployment rates among older workers triggered the newsworthiness of the issue of workforce aging, resulting in higher attention to the issue of workforce aging during the second phase (vs. the pre- and initial phase) of the financial crisis (Hypothesis 2 [H2]).

In addition, the second phase of the crisis may have affected the framing of the issue, as the focus shifted from long-term problems related to workforce aging to present-day problems with older workers’ employment. During the second phase of the financial crisis, future labor shortages were no longer a primary concern, as labor pools enlarged due to rapidly deteriorating labor market conditions and increasing unemployment rates (Van den Berge et al. 2014). As a consequence, public pressure to protect employment and welfare statement arrangement may have increased, as the perceived urgency of activating older workers’ labor participation decreased (see Conen 2013). We hypothesize that in comparison with the pre- and initial crisis period, the presence of the counteractivating frame increases during the second phase of the financial crisis (Hypothesis 3 [H3]).

Political influence

Previous research has stressed the importance of the political environment for the emergence and evolvement of competitive frames (Bell and Entman 2011; Chong and Druckman 2007). Governments are—to some extent—able to influence the framing of issues by providing accessible and reliable information to journalists. In the tradition of public accountability and openness, governmental agencies are easily accessible sources for journalists (Shoemaker and Reese 1996). Previous research shows that journalists prefer official governmental sources because they believe their information is factual and important (Livingston and Bennett 2003; Paletz and Entman 1981).

In the current study, we focus in particular on governments’ economic ideology. Governments with a left-wing ideological stance on economic issues, characterized by their active role in the economy and willingness to spend taxes on welfare arrangements, are likely to experience greater public pressure to uphold traditional welfare arrangements and prevent activating policy reforms compared with governments with a right-wing leaning. Most germane to the counteractivating argument is that the baby-boom generation will not receive the same benefits as their parents did, as they have to work longer while contributing more to comparable or even smaller pensions (Auer and Fortuny 2000). This argumentation resonates well with the solidarity and equality values that are supported by partisans on the economic left.

Based on the notions that (1) official governmental sources have a significant influence on news content and (2) governments on the economic left are likely to oppose activating reforms; we anticipate that when governments with economically a left signature are in power, the availability of official governmental sources propagating counteractivating frames is higher, which will resonate in the news content. As journalists try to bring balance to the news—and present both perspectives—we do not expect a decrease in pro-activating frames. We hypothesize that the incidence of the counteractivating frame will increase under the rule of governments on the economic left, aiming for active intervention in the economy, compared with governments on the economic right, aiming for a leaner welfare state (Hypothesis 4 [H4]).

Policy influence

The current study focuses on two policy variables that are key to understanding the position of older workers: the level of social insurance and employment protection legislation (EPL). The here-studied countries have exemplary positions on these variables (see Table 1).

Table 1.

Selected Countries’ Position on the Level of Employment Protection and Social Insurance.

		Employment Protection
		High	Low
Social insurance	High	The Netherlands	Denmark
	Low	Spain	The United Kingdom

Source. Derived from T. M. Andersen and Svarer (2007).

The aging of the workforce places increasing strains on public finances. This is especially the case for countries with generous social insurance systems, such as the Netherlands and Denmark. These countries have strong incentives for the activation and maximization of labor market potential, as public finances can only be maintained by keeping the number of people receiving benefits as low as possible. However, and although a sufficient level of social security is crucial to the provision of social welfare, generous social insurance may hurt labor participation incentives (Euwals et al. 2009). Generous insurance benefits discourage older workers to search for new jobs after dismissal (Bovenberg and Wilthagen 2008; Ljungqvist and Sargent 1998). Once unemployed older workers become eligible for social security benefits, they are more likely to permanently leave the labor force (Marmora and Ritter 2014).

Likewise, strict EPL has been identified as a potential barrier for activating older workers (OECD 2006; Schömann 2014). Due to stringent EPL rules, it may be more expensive to hire and retain older workers. In the face of generous employment benefits, “insider” older workers may avoid changing employers, with the consequence that they become “locked” in their jobs. At the same time, high levels of EPL have a depressing effect on hiring rates of workers of all ages (Daniel and Siebert 2004), which is particularly problematic for older workers who already face difficulties finding employment (Dorn and Sousa-Poza 2007). Consequently, labor mobility rates tend to be lower in countries with high EPL (OECD 2006).

Based on this literature, one may expect that in countries with generous social insurance systems and strict EPL, the urgency to reform social policies and activate older workers will be more pronounced. This, in turn, is likely to trigger several actors to articulate the need to activate the aging workforce. For example, politicians, interest groups, and experts may decide to speak up, and communicate their stance on the issue. In addition, journalists themselves may identify the need to reform labor market policies, and choose to select sources that promote the need to reform the labor market. In sum, in the context of high social insurance and EPL, we expect increased willingness of journalists to select from the pool of sources adopting a pro-activating perspective. We, therefore, expect that the incidence of the pro-activating frame will be positively affected by the level of social insurance (Hypothesis 5 [H5]) and the incidence of the pro-activating frame will be positively affected by the level of EPL (Hypothesis 6 [H6]).

Method

A systematic content analysis was carried out of newspaper articles dating from January 2006 to December 2013. The selected countries (the Netherlands, the United Kingdom, Denmark, and Spain) are geographically dispersed and include all types of Europe’s welfare state regimes. For each of the four countries, we selected two national newspapers based on the following criteria. First, we included newspapers that reach a large proportion of countries’ inhabitants and are diverse in terms of format and political leaning. Per country we selected one newspaper with a conservative leaning, and one newspaper with a liberal or progressive leaning (see Table 2). Last, our selection was guided by the availability of newspapers in Lexis-Nexis and Infomedia database. The selected newspapers were analyzed with the same search string, encompassing references to “older (worker* OR employee*)” AND/OR “workforce aging” AND/OR “(sustainable OR lifelong) employability.” This resulted in N = 1,612 newspaper articles. From the Danish sample, every second article was coded. In the remaining countries, the census of selected articles was coded, making the final sample used for frame analysis N = 1,362.

Table 2.

Sample Characteristics.

Newspaper	Country	Conservative Leaning	No. of Articles Selected	No. of Articles Analyzed
El Mundo	Spain	Yes	142	142
El Pais	Spain	No	186	186
Total (Spain)			328	328
Jyllands-Posten	Denmark	Yes	303	150
Politiken	Denmark	No	201	104
Total (Denmark)			504	254
Daily Mail	The United Kingdom	Yes	173	173
The Guardian	The United Kingdom	No	269	269
Total (the United Kingdom)			442	442
Telegraaf	The Netherlands	Yes	172	172
NRC Handelsblad	The Netherlands	No	166	166
Total (the Netherlands)			338	338
Total			1,612	1,362

Development Codebook and Coding

The codebook was developed during a qualitative and inductive prestudy. A random sample of articles from all included countries was studied in great detail, herewith focusing on frames of older workers’ employability, in terms of the capacities older workers are believed to have, the potential problems that exist around their employment and employability, and the solutions. Each newspaper article was investigated with the following questions: (1) “What is the problem?” and (2) “What should the solution look like?” The answers to these questions were used to construct frame categories pertinent to the issue of workforce aging. The prestudy sample of newspaper articles was rigorously and repeatedly analyzed until frame-category descriptions were as inclusive as possible and new frame elements were no longer encountered. The frames were selected based on explanatory power and commonality. This resulted in a diverse set of issue-specific diagnostic and prognostic frames, which were included in the codebook.

In the second phase, an international and well-trained team of five native speaking coders coded the identified frames. After repeated pretesting of the closed-items codebook, the final sample was coded. The coding unit was the newspaper article. During regular meetings and researchers’ supervision of the coders, questions were resolved. Multiple frame categories could be coded for per news article. The intercoder reliability of all participating coders was established on a random sample of 30 English-language news articles. For all frames, reliability was above Krippendorff’s alpha .60, which is comparable with previous content analytical investigations relying on rather complex categories and multiple coders (Mcmenamin et al. 2012; Van der Pas 2013). Results are reported below in Table 3.

Table 3.

Frames across Countries.

	Spain		Denmark		The United Kingdom		The Netherlands		Total
	n	%	n	%	n	%	n	%	n	%	α	Percent Agreement
Pro-activating frame	165	40.64	125	28.34	106	21.29	209	30.33	605	29.74	0.60	94.67
Contra-activating frame	141	34.73	82	18.59	69	13.86	87	12.63	379	18.63	0.60	90.22
Ageism frame	16	3.94	48	10.88	114	22.89	71	10.3	249	12.24	0.70	90.46
Labor market opportunities frame	62	15.27	146	33.11	157	31.53	234	33.96	599	29.45	0.75	92.25
Individual responsibility frame	22	5.42	40	9.07	52	10.44	88	12.77	202	9.93	0.92	92.25
Total	406	19.96	441	21.68	498	24.48	689	33.87	2,034	100	0.70	91.97

Note. α = Krippendorff’s alpha.

Issue-Specific Frames of Workforce Aging

For the presentation of the results, issue frames about workforce aging were clustered in the following five categories. The presented frames differ systematically from each other in terms of diagnosis and prognosis.[1] The first two issue frames capture a substantial amount of the total framing and represent opposing perspectives on policy reforms in response to workforce aging. (1) Pro-activating frame: Policies that accommodate older workers in their workload are argued to have detrimental side effects on their position on the labor market. Advanced systems of social security and EPL that stand in the way of a more dynamic labor market should be changed. Lowering the level of EPL and dismantling preferential treatments of older employees is necessary to improve their position in the labor market. Active working lives are encouraged, and increases in the formal retirement age promoted. (2) Counteractivating frame: The abolishment of established protective rules and rights, such as EPL, further complicates the position of older workers. The negative consequences of changing seniority-based wages and breaking down the protected position of older employees are highlighted. The outward shift of the retirement age is perceived to be problematic, because it implies that the baby-boom generation will not receive the same benefits as their parents did, but have to work longer while contributing more to comparable or even smaller pensions. Privileges and protective policies aimed at older workers must be preserved or extended. (3) Labor market opportunities: This frame stresses the troublesome position of older employees on the labor market. Participation of older workers within organizations should be facilitated and encouraged. Older workers should have equal and fair opportunities to gain employment. (4) Individual responsibility frame: This frame stresses the importance of professional development of older workers. If older employees refrain from actively investing in their own professional development, they will not be able to meet the demands of modern-day, fast-paced work environments. (5) Ageism frame: Older workers are hampered in their chances of employment by age stereotyping and/or age discrimination. Negative images and stereotypes about older workers should be combated and countered, for example, through the use of information campaigns.[2]

Operationalization

Dependent variables

This study considers three key dependent variables. First, we look at the attention of workforce aging in news media across Europe. An attention score was calculated for all articles, largely following the operationalization of previous research (Boomgaarden and Vliegenthart 2007; Vliegenthart and Boomgaarden 2007). We consider attention as a proxy for the relative editorial importance that journalists attach to the issue. The score is weighted upon the positioning of the news article in the newspaper (front page is weighted as 2; all other pages are weighted as 1), and the position and frequency of search terms within the article. In addition, attention is assumed to be sublinear dependent upon the frequency with which the search terms used to select the articles are present in the article. Last, attention is weighted by the mean amount of news articles published on a daily basis by the selected newspapers, as newspapers that publish a relatively limited amount of articles per day have to be more selective in terms of the issues that they cover. The following formula was used:

Here, is attention for the issue, dependent upon the number of referrals in the headline or lead and the number of referrals in the body of the text When terms are used more frequently, one additional single term adds less. Finally, refers to the mean amount of articles that newspapers publish per day. When newspapers publish more articles on a daily basis, the relative editorial attention for the issue decreases.

Second, we consider the dependent variables pro- and counteractivating frames. This measure represents the monthly aggregated share (i.e., percentage) of pro-activating and counteractivating frame elements of all frame elements. By taking the relative amount, we ensure that the substantive attention for active aging is accurately captured at the frame level and rule out that our dependent variables are affected by trivial variation across newspapers’ formats (i.e., size). The following formula was used:

Next, we discuss our newspaper-level variable. We follow Gans’s (1979) notion that conservative press resonates values such as respect for tradition and the defense of social order. We do not classify newspapers in terms of their right- or left-wing ideology. Conservative newspapers were determined as such based on descriptions of their ideological affiliation on news sites or on the basis of previous research (e.g., Bonafont and Baumgartner 2013; Van Dijk 1988). To trace the influence of newspapers’ political leaning on our dependent variable, a dummy variable was created distinguishing between conservative-leaning newspapers (1) and progressive or liberal-leaning newspapers (0) (see Table 2).

Our temporal variable is the second phase of the financial crisis. We created a dummy distinguishing between the period before (0) and after (1) 2012 (see De Graaf-Zijl et al. 2015).

We turn to the political variable. We used pert ratings of the Chapel Hill Expert Survey 2014 to determine the ideological stance on economic issues of governments in power during the research period (Bakker et al. 2015). Expert ratings were retrieved for individual political parties that were part of one or more ruling governments. To compute the ideological stance of each government that was in power at some point during the research period, the mean of the political parties that constitute each specific government was taken, with the party delivering the prime-minister party weighing twice as heavily. Governments on the economic left aim for an active role in the economy, whereas governments on the economic right emphasize a leaner governmental role in the economy (0 = extreme left, 10 = extreme right).

We now consider our policy variables. First, the variable employment protection legislation was derived from OECD and captures the year-level strictness of regulation on individual and collective dismissals, with higher values referring to higher levels of protection legislation. Second, social insurance refers to countries’ mean annual social expenditure per inhabitant in Euros. Both variables vary across time and countries, and were obtained from Eurostat.

Last, we discuss our control variables. Consistent with the idea that the issue of workforce aging becomes more pressing as more (older) people lose their jobs, we control for monthly varying unemployment rates, obtained from Eurostat. Lagged values were included, so as to account for fluctuations in unemployment in the recent past. In addition, in our models explaining frame variation, the number of articles in newspapers per month was added as control. Finally, lagged dependent variables were added to the models to account for autocorrelation. See Table 4 for the descriptive statistics.

Table 4.

Descriptive Statistics.

	Spain		Denmark		The United Kingdom		The Netherlands
	M	SD	M	SD	M	SD	M	SD
Attention	3.01	0.23	2.42	0.20	2.06	0.15	1.72	0.16
Governments’ ideology stance on economic issues (left–right)	5	0.11	5.71	0.12	5.24	0.11	6.58	0.05
EPL	2.28	0.01	2.16	0.00	1.24	0.00	2.85	0.00
Social insurance	5,287.57	32.39	13,340.45	78.55	8,361.07	35.28	10,308.61	61.00
Unemployment	15.36	0.45	4.84	0.11	4.95	0.07	4.00	0.07

Note. EPL = employment protection legislation.

Estimation Method

For each newspaper (n = 8), we aggregated the data to the monthly level, which makes 96 months × 8 newspapers = 768 observations. Newspapers covered the issue of older workers’ employability on average in 72.25 months during the research period. In months without coverage, the dependent variables were set to zero. The data can be characterized by having a time-series cross-sectional structure, with time periods nested within newspapers.

The particular temporal and spatial structuring of the data set requires special attention for autocorrelation and panel differences (i.e., newspapers). First, we have to assure that the dependent series are stationary, meaning that the means of each newspaper series are unaffected by a change in time origin. Augmented Dickey–Fuller tests confirm that the null hypothesis of nonstationarity can be rejected. Second, it is recommended to check for heterogeneity (Kittel 1999; Wilson and Butler 2007). Fixed-effect analyses including all our independent variables suggest (marginally) significant fixed effects for our dependent variables, indicating newspaper-specific heterogeneity. In addition, fixed- and random-effect models where compared using the Hausman test, which examines the null hypothesis that differences in coefficients are not systematic. Results fail to reject the null hypothesis for the pro- and counteractivating frames, indicating that differences in coefficients are not systematic. We are, however, able to reject this null hypothesis for the dependent variable attention. Inspections of the produced results by both the fixed and random models indicate that the differences are not substantial. As random-effect models are more efficient, and because we are not mainly interested in cross-sectional variation (i.e., time-invariant variables), we decided not to include fixed effects. Third, it is possible that series differ in their absolute level of change. Based on the error structure of the fixed-effect analysis, panel-level heteroscedasticity was determined for both our models predicting attention and frame variation, respectively, Waldattention χ2(8) = 1,228.66, p < .001; Waldpro-activating χ2(8) = 82.98, p < .001; Waldcounteractivating χ2(8) = 47.57, p < .001. Fourth, we have to consider the possibility of contemporaneous correlation across panels. To investigate whether monthly scores correlate across newspapers, Breusch–Pagan Lagrange multiplier (LM) tests were executed. We find (marginally) significant dependence between panels for our dependent variable attention, χ2(28) = 41.256, p = .051, predicting pro-activating, χ2(28) = 43.291, p < .05, but not for our model counteractivating, χ2(28) = 32.137, p = .269, framing. This indicates that newspapers undergo slightly similar developments over time. Both the structure of the data (moderate N of newspapers, moderate T of months) and the existence of modest heterogeneity, heteroscedasticity, and autocorrelation make ordinary least squares with panel-corrected standard errors the appropriate estimation method (Beck and Katz 1995). We used a lagged dependent variable to account for autocorrelation. We can write the model as follows:

whereby is the value of the newspaper i on time t on the dependent variables, c is the constant, the value of newspaper i on time t on an independent variable, the disturbance term, is the value of the error term one month earlier, is the autoregressive parameter that corrects for autocorrelation in our model predicting attention for the issue, and is the part of the error term that cannot be explained by the previous part of the error term. Given the large differences between the selected countries, our design can be characterized by a most-different-systems design, whereby we expect that the hypotheses generally apply to all countries.[3]

Results

Figure 1 displays the absolute attention for the issue across countries, while Figure 2 shows the prominence of the pro- and counteractivating frame across time. The explanatory analyses are discussed. Table 5 shows that the models explain between 8.6 percent (visibility), 8.9 percent (counteractivating), and 16.9 percent (pro-activating) of the variation in the dependent variables.

Figure 1.

Absolute issue attention across countries.

Figure 2.

Pro-activating and contra-activating frames across time.

Table 5.

Explaining Variation in Attention for Workforce Aging and the Pro-activating and Counteractivating Frame.

	Attention		Pro-activating Frame		Counteractivating Frame
	B	SE	B	SE	B	SE
Lagged dependent variable	0.131	0.040**	0.057	0.038[†]	0.031	0.042
Conservative (vs. progressive) newspapers	−0.460	0.183*	−2.593	1.895	3.327	1.829[†]
Financial crisis (second phase)	−0.147	0.247	−4.020	2.416[†]	5.556	2.324*
Governments’ ideology stance on economic issues (left–right)	−0.130	0.068[†]	−0.057	0.785	−1.392	0.738[†]
EPL	−0.114	0.148	4.461	1.654**	2.380	1.676
Social insurance	0.000	0.000[†]	0.000	0.000	−0.001	0.000
Unemployment	0.502	0.612	−1.822	6.509	8.616	6.213
Long-term unemployment	−0.395	0.617	3.204	6.571	−8.086	6.273
No. of articles			4.851	0.539***	2.257	0.523***
Constant	1.687	0.582**	2.460	6.277	16.029	6.004
R [2]	.089		.169		.087

Note. The data predict changes in attention, pro-activating, and contra-activating framing. Cell entries are unstandardized coefficients and standard errors; No. of articles is centered around its grand mean. EPL = employment protection legislation.

p < .10. *p < .05. **p < .01. ***p < .001.

Newspapers’ Political Leaning

We hypothesized that compared with progressive newspapers, conservative newspapers will address workforce aging less often in pro-activating (H1a) terms and more often in counteractivating terms (H1b). Contrary to what was expected, we do not find an effect of newspaper type on the prevalence of the pro-activating frame. However, and in line with H1b, we find a marginally significant and positive effect of conservative-leaning newspapers on the counteractivating frame (B = 3.33, SE = 1.83, p = .067). This indicates that the share of counteractivating frames is 3.33 percent higher in conservative-leaning newspapers compared with progressive-leaning newspapers, keeping other factors constant.

Temporal Influence

It was expected that the attention to the issue of workforce aging increased during the second phase of the financial crisis (H2). Table 5 shows that the crisis did not affect attention to the issue; hence, we reject H2. Second, we expected that during the second phase of the financial crisis, the counteractivating frame would gain momentum (H3). Table 5 displays a significant positive effect of the financial crisis on the incidence of the counteractivating frame (B = 5.56, SE = 2.32, p < .05). This indicates that compared with the pre- and initial crisis period, the proportion of the counteractivating frame increased with 5.56 percent during the second phase of the financial crisis. We accept H3. In addition, Table 5 displays a negative marginally significant effect of the second phase of the crisis on the incidence of pro-activating frames (B = −4.02, SE = 2.42, p = .096), indicating that during this time frame, the pro-activating frame faded from the news agenda.

Political Influence

It was projected that the counteractivating frame would proliferate under the rule of governments with a left-leaning ideology regarding economic issues, aiming for active intervention in the economy, compared with governments with a right-leaning ideology regarding economic issues, aiming for a leaner welfare state (H4). The results reveal a marginally significant negative effect of the governments’ economic ideology stance (B = −1.39, SE = 0.74, p = .059). This result indicates that, as was expected, under the rule of governments on the economic right, the proportion of the counteractivating frame decreased. The effect is rather small: The prominence of the counteractivating frame decreased with 1.39 percent—keeping other factors constant.

Policy Influence

Last, we hypothesized that the level of social insurance would positively affect the incidence of the pro-activating frame (H5). The results presented in Table 5 do not support this assumption. We do find a marginally significant effect of the level of social insurance on attention for the issue; however, as the effect size is small, this variable does not contribute substantially to attention for the issue (B = 0.0001, SE = 0.000, p = .071). In addition, we expected that the level of EPL would positively affect the incidence of pro-activating frames in the news (H6). This appears to be the case: The level of EPL has a positive effect on the proportion of the pro-activating frame (B = 4.66, SE = 1.80, p < .05). This finding indicates that for every unit increase in the level of EPL, the share of the pro-activating frame increases with 4.66 percent, keeping other factors constant. We accept H6.

Discussion

To sustain an aging workforce, European policymakers have stressed the importance of abolishing counteractivating measures promoting accommodation and protection while stimulating pro-activating measures that support an active and prolonged working life (Hofäcker 2015). This study aims to investigate how policy responses toward the issue of workforce aging are framed in European news media, and to account for dynamics in these representations.

In the following, we will elaborate on the cross-country differences. In Spain the counteractivating frame was most prominent and coincided with a high proliferation of the pro-activating frame, pointing to a relatively heated competitive frame environment in this country. Because in Spain pensions and subsidies are crucial in supporting entire families, the introduction of activation policies that may comprise social benefits has substantial consequences. Due to the significant consequences of activating policy reforms for individual careers and welfare, it is understandable that in the Spanish context pro-activating frames frequently clash with counteractivating frames that highlight the need for maintaining protective, accommodative measures.

The pro-activating frame was most prominent in the Netherlands, reflecting the necessity of pro-activating reforms to improve low reemployment opportunities among older workers in this country (Euwals et al. 2009). Dutch employers have not yet adopted an activating mind-set, as they hold negative stereotypes about the productivity of older workers and perceive them as more costly, partly due to accommodative strategies for dealing with this group (Van Dalen et al. 2010a, 2010b). These results have the potential to be generalized to other countries that are similar in terms of key labor market policies; countries such as Greece and Italy (much alike Spain) and Germany and Belgium (much alike the Netherlands) are both positioned high on employment protection policies (European Commission 2015; Eurostat 2017).

At the other end of the continuum, in Denmark, and especially in the United Kingdom, the policy debate was comparably less elaborate and less competitive. These results may be explained by the fact that the share of employed older workers is already relatively high in the United Kingdom and especially in Denmark (OECD 2012, 2015). Employers in these countries seem to have adopted a relatively activating mind-set and are generally committed to using older workers’ potential (OECD 2015; Van Dalen et al. 2009). Consequently, activating reforms may have been easier in these countries (see Vickerstaff et al. 2003), resulting in less heated news debates. The framing of workforce aging might be similar in other countries that combine low levels of employment protection with high social insurance (much like Denmark), for example, Finland and Austria. Countries that combine low levels of employment protection with low social insurance (like the United Kingdom) are, for example, Ireland and Hungary (European Commission 2015; Eurostat 2017).

There are some important causal mechanisms driving the effects we found. Theoretically, we assumed that both inter- and extra-media factors determine news media’s attention for, and framing of, the policy debate on workforce aging. First, and regarding the intermedia antecedent, the results show that the framing of the issue is driven by the ideology of the news organization. It was argued that conservative forces may see activation as a threat to the traditionally protected employment position of older workers (Häusermann 2006, 2012). As was projected, readers of conservative-leaning newspapers are exposed to a less elaborate debate about the issue, in which the counteractivating frame more strongly proliferates. This offers support for the idea that the ideological stance of the editorial board determines to a large extent what is deemed newsworthy, and how issues should be framed (Reese 2001).

Second, an important conclusion of this study is that extra-media-level forces matter for media’s framing of the policy debate on workforce aging. This confirms the validity of including the wider political and policy context to the study of policy frame contests (Carragee and Roefs 2004). To start, the time frame under investigation proved to matter for the framing of the issue. The study concludes that during the second phase of the economic crisis, when the employment of older workers was hit hardest, the counteractivating frame gained momentum. It was anticipated that the availability of sources promoting the counteractivating frame would increase during economic heavy weather, as increasing unemployment rates and deteriorating labor market conditions for older workers appeal to counterargumentation of protection and accommodation (see Conen 2013). The results support the idea that the intensified opposition against reforms during a crucial moment in the financial crisis was the driving force behind the change in framing.

In addition, governments’ ideological stance proved an important driver of variation in news coverage on the issue. In line with expectations, under the rule of governments on the economic left, the prominence of counterframes on the news agenda increased. The agenda of solidarity and equality of left-wing political parties resonates well with the counteractivating argumentation that older workers should receive the same benefits as their parents did, instead of working longer while contributing more to comparable or even smaller pensions (Auer and Fortuny 2000). As a consequence, the availability of official sources propagating the counteractivating frame likely increased under the rule of left-wing governments. As official (governmental) sources occupy an institutionalized position on the news agenda, they are relatively successful in channeling their perspective on issues to the public (Livingston and Bennett 2003; Shoemaker and Reese 1996).

Last, and due to our comparative approach, the study was able to identify the policy context as an antecedent of media’s framing of the issue. The findings show that in the context of high social insurance, visibility of the issue increased. This indicates that in policy contexts characterized by solidarity, where governments spend taxes to protect the welfare of (un)employed workers, social issues—like older workers’ employability—are logically more in the forefront. Second, and in line with our expectations, the results reveal that the level of EPL positively affected the incidence of pro-activating frames. This offers support for the idea that the costs associated with EPL and its potential negative consequences for the activation of older workers increased the pool of available sources advocating pro-activating change. Journalists might also have taken an active stance here. Triggered by the problems that exist in relation to current employment policies, journalist might have actively framed the issue in pro-activating terms. Yet, the study finds no influence of the level of social insurance on the dynamics of competitive framing. This might be because a sufficient level of social security also has positive effects on activation, such as job mobility (T. Andersen et al. 2008), and the general level of welfare. It is, therefore, conceivable that journalists and key sources perceive high levels of social insurance as less problematic.

The results reveal that pro- and counteractivating frames were not affected in an equal manner, and thus evolve in part separately from each other with consequences for the extent that competing frames proliferate in conjunction on the news agenda. A potential explanation for this is that the promotion of a certain frame may trigger counterresponses, as journalists aim to provide a balanced perspective on the news. This indicates that the degree of competitiveness between frames is also a dynamic process.

We belief the reported findings have generalizability potential to other comparable policy domains that are being fundamentally reviewed, and where the battle between “old” and “new” perspectives creates tensions. In the debate about flexible employment, for example, strong opposing and supporting arguments prevail (Hoekstra et al. 2016). The causal mechanisms identified in the current study may help scholars to identify which perspective is more supportable under certain circumstances. Yet, it should be noted that the highly complex interplay between governments, key (political) actors, news media, and specific policy issue arenas make the identification of generalizable framing patterns a challenging endeavor. More comparative studies, as the one presented here, are needed to improve our confidence in the generalizability of the here identified causal mechanisms to other policy domains.

The limitations of this study should be discussed. To start, as indicated by the marginally significant effects, systematic differences across the selected countries exist, but they are relatively limited in size. We encourage future studies to investigate whether our findings hold across other countries and continents. Second, and following the reasoning that governments are an important force driving the framing and attention for policy issues (Livingston and Bennett 2003; Paletz and Entman 1981), we included governments’ stance on economic issues as our main political variable. It should be noted, however, that the comparability of governments is limited as the countries under study have different systems (i.e., the United Kingdom has a majority system, whereas the Netherlands has a coalition government). Future studies may include variables tapping into the broader political context by including the composition of parties in the parliament. Last, an interesting question not addressed in this study relates to the extent to which activation reforms trigger intergenerational conflict.

In sum, this study has isolated the factors that explain the relative success and failure of competitive framing in the debate about activating policy reforms, of which the outcomes are vital in the quest for public support. We have demonstrated that framing of an issue that is the same across all European countries (i.e., policy responses to workforce aging) is covered differently across countries, and that news media ideologies, time frames, governments’ political leaning, and national employment policies are useful in explaining dynamics in competing frame perspectives. These insights may contribute to news representations that promote an activating approach on workforce aging in European news, which is needed to financially and socially sustain older workers’ employability.