Introduction
Italy’s labour market policy has historically prioritised strong job security for permanent employees and passive support measures that favour those already employed. This ‘insider-oriented’ approach has created a marginalised group of ‘outsiders’ who must depend on limited active labour market policies (ALMPs), including direct job creation, training, and job search assistance (Ciarini & Rizza, Reference Ciarini and Rizza2021; Ferragina & Filetti, Reference Ferragina and Filetti2022). Nevertheless, among ALMP, one area of intervention stands out as an exception. Employment subsidies have consistently been the most prominent ALMP, averaging 0.21 per cent of GDP between 2009 and 2022, well above the European average (0.13 per cent). This contrasts sharply with the 0.13 percent allocated to training during the same period, the residuality of direct job creation programmes, and the mere 0.02 per cent dedicated to public employment services.Footnote 1
Employment subsidies are demand-side labour market instruments that reduce employers’ labour costs by providing financial incentives to hire or retain workers. They typically take the form of exemptions or reductions in employers’ social security contributions, tax credits for new hires, or cash transfers. In Italy, prominent examples include non-targeted contribution rebates for open-ended contracts, hiring incentives for young workers or residents in Southern regions, and subsidies tied to the conversion of fixed-term contracts into permanent ones. Unlike training or employment services, these measures do not directly aim to enhance workers’ skills or employability but instead seek to stimulate labour demand by lowering the cost of employment.
Scholars have highlighted Italy as a clear example of a growth model in which employers’ competitive strategies favour low-skilled, low-wage, and low-quality jobs, thereby weakening demand for enabling ALMPs (Baccaro & Bulfone, Reference Baccaro, Bulfone, Baccaro, Blyth and Pontusson2022). Italy’s economy is characterised by a predominance of small and medium-sized enterprises (SMEs), many of which are concentrated in low-tech, traditional industries. These firms often lack the resources, scale, and incentives to invest in comprehensive training programmes or advanced workforce development initiatives (Burroni et al., Reference Burroni, Pavolini, Raspanti, Hassel and Palier2025). As a result, there is limited employer demand for robust skill formation systems, which has contributed to the underdevelopment of ALMPs focused on long-term human capital development and may have incentivised investments in employment subsidies.
This contribution offers an alternative, yet complementary, explanation for the dominance of employment subsidies in Italy’s ALMP regime. Drawing on political economy literature, it highlights the role of ideas as a powerful force shaping labour market policy implementation (Béland, Reference Béland2005; Carstensen & Schmidt, Reference Carstensen and Schmidt2016; Schmidt, Reference Schmidt2008; Hauptmeier & Heery, Reference Hauptmeier and Heery2014). We argue that the public discourse on the challenges facing the Italian labour market is primarily shaped by a narrative assertively promoted by the business community: the high cost of labour. This account has overshadowed an alternative explanation that attributes the situation to the limited innovative capacity of Italian firms (Evangelista & Pacelli, Reference Evangelista and Pacelli2025; Burroni et al., Reference Burroni, Gherardini and Scalise2019; Trigilia, Reference Trigilia2024). Policy choices have disproportionately favoured employment subsidies, which align with cost-reduction strategies, while more transformative active labour market measures aimed at enhancing skills and innovation remain significantly underutilised.
The study employs a multi-step analytical approach combining qualitative and computational text analysis to examine how ideas shape Italy’s subsidy-oriented pattern of labour market policy. Qualitative methods uncover the framing strategies of key actors, while computational techniques identify their dissemination and prominence, providing complementary insights into the ideational foundations of policy outcomes. National news media and parliamentary debates constitute the primary research material as they offer a window into public debate where actors can articulate their positions and attempt to legitimise their policy preferences, while also providing a valuable basis for tracing the salience of these positions. The analysis draws on four original textual datasets covering the period 2014–2020: the Labour Market Developments Dataset, the Labour Cost Dataset, the Active Labour Market Policy Dataset, and the Parliament Dataset. Together, these sources enable a systematic examination of how employers’ narratives on labour costs emerged, diffused across the media, and were incorporated into parliamentary debate. Special focus was directed toward social partners, whose interventions are pivotal in shaping how labour market dynamics are understood and debated in the public and policy spheres.
This article addresses several significant gaps in the discussion on the Italian ALMP regime and offers a key original contribution to the current debate. First, employment subsidies have predominantly been analysed in terms of their impact on firms’ economic and financial performance (Santarelli & Zaninotto, Reference Santarelli and Zaninotto2007). This includes their role in promoting investment, boosting productivity, and fostering the creation of new firms, particularly in underdeveloped regions in the Mezzogiorno (De Blasio & Lotti, Reference De Blasio and Lotti2008). However, little attention has been given to examining the role of employment subsidies within the broader ALMP supply in Italy. This is quite surprising given that employment subsidies are among the few nationwide ALMPs. ALMP governance in Italy is decentralised, with regional governments having autonomy over funding and delivering labour market services and training, while the central government plays a (weak) coordinating role (Giuliani & Raspanti, Reference Giuliani and Raspanti2022). In contrast, employment subsidies are funded through the annual Budget law and are directly managed by national governments.
Second, this paper contributes to the literature by emphasising the role of narratives in developing ALMP supply. While the relationship between narratives and social policy development has been explored internationally, the focus has primarily centred on activation policies (Lødemel & Moreira, Reference Lødemel and Moreira2014; Weishaupt, Reference Weishaupt2011), excluding employment subsidies from the analysis. Building on Béland (Reference Béland2005, Reference Béland2024), Schmidt (Reference Schmidt2008), and Carstensen and Schmidt (Reference Carstensen and Schmidt2016), we illustrate how discursive mechanisms have generated intense pressure to prioritise employment subsidies as the main instrument of ALMP.
Finally, from a methodological perspective, this paper advances the use of mixed methods by incorporating a quantitative approach to text analysis. In doing so, it challenges the dominance of purely qualitative interpretive methods and contributes to the development of a more balanced and robust methodological toolkit for labour market policy analysis.
This paper is organised as follows: ‘How ideas and narratives shape ideational power and discourse in public policies’ outlines the theoretical framework, emphasising key studies on the role of ideas in public policy. ‘Employment subsidies and employers’ preferences’ reviews research exploring employers’ preferences for employment subsidies. ‘The Italian way to ALMP: a historical view on employment subsidies’ examines the long-term characteristics of the Italian labour market and provides a historical overview of employment subsidies. ‘Research design and methods’ describes the research design and methodology. ‘Results’ presents the results of the text analysis. ‘Discussion’ discusses the findings. Finally, Section ‘Conclusions’ reflects on our contribution and limits.
How ideas and narratives shape ideational power and discourse in public policies
Our analysis draws on the literature on ideational power, which emphasises how collective actors use ideas and narratives to frame policy problems, define what counts as legitimate solutions, and influence political debate (Blyth, Reference Blyth2002; Gofas & Hay, Reference Gofas and Hay2010; Moschella et al., Reference Moschella, Quaglia and Spendzharova2024). Narratives – the coherent stories actors tell about social and economic challenges – are particularly central to shaping policy legitimacy and the boundaries of feasible reform (Radaelli, Reference Radaelli1999; Hay, Reference Hay2002). Another crucial dimension is the relationship between ideational power and the capacity to control discourse (Schmidt, Reference Schmidt2008). Discourse can be understood as the structured process of communication through which ideas, meanings, and interpretations are shared, contested, and institutionalised. It encompasses both the substantive content of ideas – such as policy narratives and frames – and the interactive processes through which these ideas are conveyed, negotiated, and legitimised within the public sphere or among decision-makers. Actors wielding ideational power shape public perceptions by strategically influencing the narratives surrounding specific issues to emphasise policy problems while downplaying others. Through strategic discursive processes, actors legitimise or delegitimise policy proposals in the eyes of various constituencies by constructing policy narratives – stories that provide a coherent explanation of policy issues and justify proposed solutions (Béland, Reference Béland2024).
In the domain of welfare-state reform, empirical studies show how dominant framings influence both policy orientation and reform space. For example, Béland (Reference Béland2005, Reference Béland2024) demonstrates that framings of welfare recipients as ‘deserving’ or ‘undeserving’ strongly affect the political viability of reform agendas, shifting policy preferences toward market-based, conditional models of social provision. In the Italian context, Jessoula & Natili (Reference Jessoula and Natili2024) argue that retrenchment of pension and labour-market protections and the subsequent ‘expansionary turn’ in family and antipoverty policy reflect bigger changes in institutional framing and ideological repertoires rather than purely fiscal constraints. Turning to ALMPs, research on Italy reveals essential insights. The OECD’s 2019 country study on Italy’s active labour market policies outlines how structural weaknesses – fragmentation, limited funding, uneven regional implementation – are embedded in policy narratives about activation, flexibilisation, and employer cost burden. Natali’s (Reference Natali, Blum, Kuhlmann and Schubert2009) work emphasises that Italy’s enduring dualism – insiders vs outsiders, heavy transfers vs light services – has been sustained by underlying dominant ideas about protection, labour markets, and family responsibility.
Applying these insights to the Italian ALMP field, we argue that the persistent reliance on employment subsidies should be read not simply as an outcome of institutional inertia or economic constraints, but as part of a dominant policy narrative. Employers’ organisations in Italy have deployed a framing that positions excessive labour costs as the main obstacle to employment growth. While this framing does not automatically determine policy choices, it appears to have helped legitimise a policy orientation favouring cost-reduction instruments – such as subsidies for employment – over more transformative or enabling ALMPs (for example, training, human-capital development, innovation-oriented interventions). The limited articulation of alternative narratives has further reinforced this alignment.
From this theoretical-empirical vantage point, we formulate the following hypothesis:
Hp1: The predominance of employment subsidies in Italy’s ALMP regime may be associated with a dominant narrative that frames excessive labour costs as the principal obstacle to employment.
The persistence of employment subsidies in Italy’s ALMP toolkit, beyond economic or institutional factors, aligns with a broader narrative emphasising competitiveness through wage moderation and cost containment. Based on this, we formulate the following hypothesis:
Hp2: Employers’ organisations have been the leading promoters of this narrative, exercising ideational power by influencing public discourse on labour-market challenges.
Italy’s persistent reliance on employment subsidies, irrespective of the political orientation of incumbent governments, suggests that this policy approach reflects shared assumptions rather than partisan agendas. Once a dominant narrative becomes embedded in the public sphere, it tends to diffuse into the political arena, shaping how problems are framed and which solutions appear legitimate. This expectation informs the third hypothesis:
Hp3: This employer-driven narrative is reflected in parliamentary debates, helping to legitimise policy preferences that prioritise cost-reduction measures over human-capital development and innovation-oriented interventions.
The three hypotheses are sequential and interdependent, with each building on the preceding one. Together, they outline how a dominant policy narrative emerges and diffuses across public discourse and political decision-making, and how this process sustains the centrality of employment subsidies within Italy’s ALMP regime.
Employment subsidies and employers’ preferences
Employment subsidies are employer-oriented ALMPs that provide financial incentives to employers to reduce labour costs. Labour policy scholars often distinguish employment subsidies from other active programmes, particularly in terms of their effects. Rather than being tailored to the needs of the labour supply, through initiatives aimed at enhancing the skills of the unemployed or improving job placement opportunities, these measures often appear to be applied in a biased manner. They tend to be predominantly oriented toward addressing labour demand, particularly benefiting low-productivity enterprises by reducing hiring costs (Robertshaw, Reference Robertshaw2024).
There are two main types of subsidies: wage subsidies, such as short-time work schemes, are designed to support employees at risk of involuntary job loss (Neumark, Reference Neumark2013). Hiring subsidies, on the other hand, aim to lower companies’ recruitment costs. In this paper, the term ‘employment subsidy’ will specifically refer to hiring subsidies. These subsidies provide fiscal incentives for hiring through social contribution rebates, fiscal credits, or cash transfers, effectively transferring public resources from the state to employers (Cronert, Reference Cronert2018). Non-targeted employment subsidies, which apply to all new hires regardless of their characteristics or employment status, are advantageous for employers but can result in higher deadweight and displacement effects (Brown & Koettl, Reference Brown and Koettl2015). The former occurs when a subsidy benefits employers who would have hired the individual even in the absence of the subsidy. The latter effect refers to the situation in which the employment subsidy reduces employment opportunities for existing workers. Both effects can undermine the goals of ALMPs by failing to generate additional employment or by negatively impacting the employment prospects of current workers.
Cronert (Reference Cronert2019) categorises employment subsidies as ‘joint’ ALMP programmes, produced in collaboration between the state and employers, distinguishing them from ‘unilateral’ ALMP programmes, such as training and labour market services, which are solely state-provided. From this perspective, it is crucial to outline employers’ preferences regarding ALMPs. Although businesses frequently oppose passive labour policies, viewing them as drivers of higher labour costs and reduced market flexibility, they generally favour ALMPs (Cronert, Reference Cronert2018). These policies are seen as beneficial for increasing firm competitiveness and enhancing workforce productivity (Gordon, Reference Gordon and Oude Nijhuis2019; Bender, Reference Bender2023; Colombo et al., Reference Colombo, Natali, Pavolini, Clegg and Durazzi2023). Employment subsidies are a typical example of this alignment, as they directly reduce labour costs for businesses while promoting employment.
The extent of business support for ALMPs can vary significantly across different contexts. As Clasen and colleagues argue, employment subsidies can have very different objectives and yield divergent outcomes, depending on their targeting and the specific groups they are intended to support (Clasen et al., Reference Clasen, Clegg and Goerne2016). As a result, they may be backed by very different support coalitions. In Italy, the prominence of employment subsidies within ALMPs takes the form of demand-led interventions, i.e. mainly contribution relief and tax credits for employers without introducing specific selective criteria (van der Aa & van Berkel, Reference van der Aa and van Berkel2014). As a result, they leave the structure of labour demand and the characteristics of the labour supply essentially unchanged. This contrasts with more targeted, demand-oriented interventions – such as tax breaks or subsidies linked to employee training, technological innovation, or research and development – which could incentivise businesses to invest in human capital and drive long-term economic transformation. However, such measures remain significantly underutilised. Rather than being strategically aligned with skill formation and workforce development, these subsidies are frequently used to incentivise hiring by reducing employers’ immediate financial burden. This approach tends to prioritise short-term cost savings over long-term investments in training and job-search assistance. As a result, while employment subsidies may temporarily boost employment figures, their impact on labour productivity and the overall competitiveness of the workforce may be limited (Ardito et al., Reference Ardito, Berton and Pacelli2023).
The Italian way to ALMP: a historical view on employment subsidies
The overall expenditure level in ALMP in Italy is traditionally among the lowest in Europe (Bonoli, Reference Bonoli2013; Rizza et al., Reference Rizza, De Luigi, Giullari and Trigilia2023). Figure 1 compares the ALMP average expenditure in Italy between 2009 and 2022 with the EU27 average. The European average exceeds the Italian average across all actions except for expenditure on employment subsidies. Italy spends only 0.366 per cent of GDP on ALMP compared to an average of 0.499 per cent. The Italian bias towards employment subsidies is evident in the composition of ALMP expenditure. Employment subsidies accounted for 0.210 per cent of GDP, nearly twice the European average (0.129 per cent), representing 65.7 per cent of total ALMP expenditure. The contribution of other ALMPs in Italy is significantly marginal in absolute and relative terms. This implies that ALMP supports employment mainly through a temporary reduction in labour costs.
Expenditure by ALMP programme (% GDP), average value 2009–2022.
Source: Authors’ elaboration on DG-Employment data (lmp_expsumm database).
The prominence of employment subsidies dates back to the late 1970s, when governments had to address the consequences of the 1979 oil shock. Massive downsizing and restructuring were managed with the utilisation of short-time work schemes, generous early retirement programmes, and means-tested disability pensions in Southern Italy (Ascoli & Pavolini, Reference Ascoli and Pavolini2015). Employment subsidies became the most critical ALMP to support labour demand for the most marginal groups in the labour market, i.e. young jobseekers, women, people in southern regions, and long-term unemployed.Footnote 2
During the 1980s and 1990s, successive governments repeatedly relied on contribution rebates and tax credits to counter rising unemployment, thereby entrenching subsidies as a permanent feature of Italian labour policy. In the early 1990s, Italy entered a recession following a gradual decline in GDP growth from 1989 onwards, which led to higher unemployment and inflation (Gualmini & Rizza, Reference Gualmini and Rizza2011). Against this backdrop, the government again turned to employment subsidies to address the labour market crisis. The 1991 Budget law introduced a structural tax credit for employers hiring long-term unemployed or beneficiaries of short-term work schemes, a measure that was repeatedly renewed until 2015. The 1999 Budget law targeted Southern unemployment by granting a three-year exemption from social contributions for employers hiring on open-ended contracts.
In the 2000s, employment subsidies increasingly became a standard anti-crisis tool. The 2001 Budget law introduced a temporary subsidy for firms hiring individuals aged twenty-five or over on open-ended contracts who had been unemployed for at least two years, with additional incentives for employers in the South. In 2014, a centre-left government approved a comprehensive labour market reform, the ‘Jobs Act’. Among other provisions, the reform modified rules on unfair dismissal for newly hired open-ended workers in firms with more than fifteen employees. The 2015 Budget law introduced a very generous non-targeted employment subsidy for all new open-ended hires, regardless of firm size. The government thus sought to stimulate demand for stable employment with social contribution relief (Ardito et al., Reference Ardito, Berton and Pacelli2023). The subsidy was renewed in 2016, albeit at a lower level of generosity. As shown in Figure 2, public expenditure on employment subsidies grew to an unprecedented level since the end of the 1990s. According to European Commission data, Italy spent 21.3 billion euros in subsidies from 2015 to 2018, primarily driven by non-targeted hiring subsidies introduced in 2015 and 2016. SMEs benefited the most, with higher conversion rates from fixed-term to permanent contracts than large firms (Boeri & Garibaldi, Reference Boeri and Garibaldi2019). However, many of the newly created open-ended jobs were concentrated in non-innovative sectors and ended once the subsidy expired (Ardito et al., Reference Ardito, Berton and Pacelli2023), particularly in low-wage and low-productivity firms (Patriarca et al., Reference Patriarca, Santoni and Scarlato2023). In 2016, a new government took office, and the subsidy was not renewed.
Expenditure in employment subsidies, Italy.
Source: Authors’ elaboration on DG-Employment data.
From 2016 onwards, successive governments continued to rely on employment subsidies regardless of political orientation. More recently, a new measure was introduced in 2020 for open-ended hires in Southern regions and a three-year subsidy for employers hiring young people under thirty-six on permanent contracts. In 2023, the right-wing government of Giorgia Meloni introduced a subsidy for firms hiring unemployment assistance beneficiaries. It expanded tax reduction for new open-ended hires by all employers and the self-employed.
Overall, the historical trajectory of employment subsidies reveals a consistent pattern: governments have used these measures less as activation tools and more as mechanisms of labour cost containment.
Research design and methods
This study adopts a multi-step analytical approach (summarised in Table 1) to clarify the role of ideational factors in shaping Italy’s subsidy-oriented pattern of ALMP expenditure. We understand ideas as operating across multiple levels, from the formation of meanings to their diffusion and normalisation in public debate, culminating in their articulation within the parliamentary arena.
Summary of the analytical process and datasets used in the study
National news media and parliamentary debates constitute the primary research material, covering the period between 2014 and 2020.Footnote 3 Although this selection is based on dataset availability, it remains conceptually sound. ALMP gained prominence in public discourse from 2014 onwards due to the importance of labour market reform in the centre-left government’s agenda, known as the ‘Jobs Act’.
The study integrates qualitative and computational text analysis, which serve complementary purposes: the former helps explore meaning and framing construction, whereas the latter facilitates tracking their dissemination and prominence. Specifically, qualitative study was used to reconstruct the narratives and interpretive frames through which key social partners diagnosed the ongoing labour market stagnation (Step 1); subsequent, computational text analysis helped quantify and map their diffusion across broader media datasets (Steps 2 and 3); finally, a qualitative reading of parliamentary debates assessed their resonance and integration into political settings (Step 4).
Step 1 focused on employers’ organisations’ perspectives on Italian labour market developments. The goal of this analysis was to identify common themes and patterns in the discourse of business leaders about the reasons behind the sluggish growth of the Italian labour market, as well as the interventions aimed at promoting employment growth and reducing unemployment. To this end, we analysed eighty national media articles drawn from NexisUni between 2014 and 2020 (Labour market developments dataset). The focus on national newspapers – including Il Corriere della Sera, Il Sole 24 Ore, La Stampa, and the most crucial national press agency (Ansa) – is justified because employment subsidies are financed at the national level, primarily through annual Budget laws. Articles were filtered using key keywords such as ‘aumento dell’occupazione’ (employment growth) and ‘riduzione della disoccupazione’ (unemployment reduction), as well as mentions of prominent employers’ organisations. The results of this first step underscored the relevance of (perceived) excessive labour costs in hindering employment growth. In the next step, we have decided to focus on the public debate over labour costs.
Building on the centrality of the labour costs narrative identified in this stage, in the second step, we investigated its diffusion into public debate. Using the online repository MyDesk24, we retrieved articles from Il Sole 24 Ore, the journal financed by the leading Italian employers’ organisation: Confindustria. The newspaper was selected as a key outlet reflecting employers’ perspectives and operating as a privileged channel for business frames to reach the public. We retrieved articles containing the key phrase ‘costo del lavoro’ (labour cost), together with at least one additional keyword, including ‘sussidi’ (subsidies), and ‘sgravi’ (tax relief), or references to major social partners, including Confindustria and the main trade union, the CGIL. This procedure yielded a corpus of 627 articles (Labour Cost Dataset). For greater precision, the corpus was then filtered to retain only the 896 sentences containing the key phrase ‘costo del lavoro’.
In Step 3, we widened the lens to the general news media again to track ALMP-related topics, thereby allowing us to verify whether narratives initially rooted in employers’ discourse had permeated this broader discursive environment beyond sector-specific channels. NexisUni searches yielded 1,413 articles (Active Labour Market Policy Dataset) mentioning ‘politiche attive’ (ALMP) and major social partners, which were then filtered to isolate the 1,535 sentences containing the key phrase. BERTopic, a topic modelling approach that leverages advanced text embedding techniques and clustering algorithms (Grootendorst, Reference Grootendorst2022), was used to analyse the texts from these two latter steps to systematically identify recurrent themes (details on the hyperparameter selection process and evaluation metrics are provided in the Appendix).
At the end of the third step, we stopped interrogating our datasets and moved on to analysing parliamentary debates. In Step 4, we investigated whether and how previously identified discursive patterns appeared in parliamentary debates. Drawing on the ParlaMint corpus (Ogrodniczuk et al., Reference Ogrodniczuk, Osenova, Erjavec, Fišer, Ljubešić, Çöltekin, Kopp and Meden2022; Erjavec et al., Reference Erjavec, Ogrodniczuk, Osenova, Ljubešić, Simov, Pančur and Fišer2023) – a collection of speeches from over twenty national and regional European parliaments (Parliament Dataset) – 2,400 sentence segments were selected with the same set of keywords used in the previous steps (e.g. sussidi, incentivi, costo del lavoro, occupazione). Returning to a qualitative reading at this stage was necessary as it allowed us to trace how business-centred ideas were taken up, reframed, or contested within formal political discussions.
Results
Analysis of public discourse on Italian labour market policy instruments indicates that high labour costs are consistently framed as a key obstacle to job creation. Both unions and employers’ organisations stress the importance of fiscal measures to increase employment levels. For example, according to the secretary of one of the central unions, ‘If open-ended contracts cost 10–15 per cent less, I don’t know how many companies would choose them’ (a_2).Footnote 4 Representatives from employers’ organisations have consistently expressed the need for labour cost reduction via fiscal reforms, the simplification of recruitment-related administrative procedures, tax cuts, and deductions from compulsory employers’ social contribution (a_3). The same organisations have also criticised previous government measures targeting employees, such as the Eighty Euro Tax Rebate, which provided an automatic tax rebate of eighty euros to employees, freelancers, independent contractors, and project-based workers, arguing that lowering the tax wedge for employers would have had a more substantial impact (a_3). Our analysis shows that references to the utilisation of other supply-side measures are marginal in the debate.
Employment subsidies entered public discourse only after the 2015 labour market reform, as the debate focused on their attractiveness in fostering the adoption of the new open-ended contract. The subsidy was positively welcomed by all employers’ associations (a_5) (a_6) (a_8). Evaluating the employment outcomes of the reform, the leading employers’ organisation requested the continuation of subsidies by reporting that employment growth during the first nine months of 2015 was driven by tax incentives for new hires (a_14). Caution has been voiced against reversing the subsidy as the policy has positively affected employment (a_16, a_18). However, comments on the subsidy implementation were framed into a call for broader reforms. A local branch of an employers’ organisation warned that ‘if we do not address procedures and taxes, businesses […] will struggle in a context that hinders rather than supports their activities’ (a_9). This prevailing narrative emphasises that, while employment incentives are valuable, they must be part of a comprehensive strategy that includes tax cuts and bureaucratic simplifications to foster a business-friendly environment. The employers’ organisation of agricultural firms shared similar views, emphasising the importance of reducing tax burdens on agrarian enterprises to facilitate investment in innovation and employment growth in Italy’s agri-food sector (a_13).
As the discussion about interventions to improve labour market performance concentrated on labour costs, we decided to examine how this theme is framed within the broader public debate. Figure 3 summarises the results of the quantitative analysis conducted with the BERTopic method on the Labour Cost Dataset in Step 2. As expected, arguments about labour costs relate to fiscal reform and national productivity strategies. Key priorities are (1) Fiscal measures and tax reductions, suggesting that budgetary relief is perceived as essential for profitability and growth. Similarly, the theme (2) Productivity and Competitiveness emphasises labour cost reduction to strengthen Italian firms’ competitiveness in international markets. Confindustria frequently argues that reducing the tax wedge on labour is vital for alleviating business financial pressures and restoring competitiveness (b_119, b_217, b_225, b_686). In their view, employers’ high social contributions have eroded the cost competitiveness of Italian firms (b_85). By drawing comparisons with other countries, employers’ organisations highlight how a more sustainable labour cost structure is not only feasible but necessary for improving export performance and competitiveness (b_360). Employers’ organisations’ discourse on labour costs reflects a deliberate effort to frame high labour costs as a burden to productivity increases. Accordingly, cost-reduction measures – such as tax cuts and fiscal incentives – are essential to align Italy’s economic structure with its more competitive European counterparts.
BERTopic output on ‘labour costs’. Outliers are excluded. Refer to the Online Supplementary Material for the complete output.
The fourth priority by relevance is (4) Long-term Employment Contracts and Tax Incentives. Interestingly, it aligns the discussion of the relationship between open-ended and fixed-term contracts with the call for fiscal measures, highlighting a strategic preference for tax incentives as the principal form of intervention in ALMPs to increase employment. Again, employment subsidies come into play when the discussion focuses on the introduction of the new open-ended contract in 2015. Employers call for stabilising labour costs through the employment incentive and for open-ended contracts to be more convenient than fixed-term contracts (b_727). Employment subsidies could be viewed as a special type of fiscal measure, as they are, in effect, tax reductions on employers’ social security contributions.
Following the analysis of the discourse on labour cost reduction in Step 3, we moved to the debate about ALMP to discern the extent to which subsidies are prioritised (or not) within the discussion of labour market policies. Figure 4 shows the results of the quantitative analysis using the BERTopic model. As anticipated, employment subsidies are absent from the debate. Indeed, social partners’ focus is on vocational training. This appears paradoxical, given the vast amount of public spending on employment subsidies compared to the significantly lower investment in public employment services, job search assistance, and training, as shown in the previous section. This lack of attention indicates a tacit understanding among stakeholders that employment subsidies do not fully align with the primary goals of ALMPs, namely, improving employability through measures such as training, skill-building, job matching, and job-search support. Indeed, these subsidies are frequently applied in ways that stray from these objectives, and their limited role in policy discussions suggests a shared awareness that they are not chiefly intended to support the most disadvantaged workers.
BERTopic output on ‘active labour market policy’.
Even when training is mentioned in public discourse on ALMPs, it is often addressed superficially, without engaging with fundamental questions such as the objectives of training programmes, the target groups, or the expected outcomes. The emphasis on training often appears more symbolic than substantive – a kind of ‘ritual talk’ tied to the European Social Fund’s role in financing training and vocational education at the regional level. This helps explain the frequent references to ‘Regional agreements’ seen in Figure 4. In our view, this pattern once again underscores the marginal role actually assigned to training and other active labour measures, as reflected in the predominance of business incentives within the Italian ALMP framework. References to public employment services emerge in the dataset as the Jobs Act reform modified the governance of labour market policy by introducing the National Agency for Active Labour Policies (ANPAL).
These results could be interpreted as a tacit admission by stakeholders that employment subsidies are not within the scope of ALMPs.
The policy frameworks promoted by employers’ organisations enter the parliamentary arena, shaping policymakers’ discussions of labour market interventions. Drawing from the ParlaMint corpus, the analysis reveals how high labour costs, echoing concerns raised by employers, are regarded as significant barriers to job creation and economic growth.
A common understanding emerges across the political spectrum: addressing high labour costs is critical for the Italian economy. Both right- and left-wing parties agree that reducing the tax burden on businesses is essential. This is evident in various proposals for tax reforms aimed at lowering employer contributions and promoting stable employment contracts (c_5052, c_9705). There is also widespread recognition of the need to simplify and enhance access to fiscal incentives for employers (c_8969).
All parties emphasise the immediate benefits of tax incentives, although their arguments differ slightly in focus. Right-wing parties argue that these policies could significantly stimulate the economy by easing tax burdens on SMEs, leading to job creation and GDP growth (c_42). Right-wing parties further emphasise the creation of youth employment incentives, proposing fiscal reliefs such as reductions in the regional tax on productive activities and the personal income tax to encourage job creation, particularly for those under thirty-five (c_2394, c_6117). This approach aligns with the broader right-wing narrative that lowering labour costs is paramount for fostering a competitive economic environment, creating a direct connection between tax relief measures and employment incentives (c_6137).
On the left and centre-left, there is also a focus on labour costs and fiscal reforms, but with a greater emphasis on labour protections and inclusivity. Centre-left factions advocate reducing the tax wedge for employees (c_2596) and promoting permanent contracts and enhancing the stability of existing employment incentives (c_9705). Their proposals include targeted tax cuts for businesses and workers, employment subsidies targeting young and women (c_28863), and public-private investment strategies to support long-term job creation and economic growth (c_18782, c_19927). Additionally, left-wing parties emphasise the importance of investing in human capital, advocating education and training programmes to enhance labour productivity while supporting innovative industries (c_20173, c_18117). Reducing the tax wedge is thus framed as an essential reform to secure Italy’s economic growth and development (c_23252).
Discussion
The findings provide strong support for Hp1, which posited that the predominance of employment subsidies in Italy’s ALMP regime is associated with a dominant narrative framing excessive labour costs as the principal obstacle to employment growth. Across media coverage, employer statements, and parliamentary debates, labour costs consistently emerge as the primary diagnostic frame through which labour market stagnation is interpreted. This framing marginalises alternative explanations – most notably those pointing to low innovation capacity, weak skill formation systems, or structural deficiencies in labour demand. Importantly, this narrative is closely aligned with policy outcomes. Employment subsidies, indeed, function as de facto reductions in employers’ social contributions and fit within a labour-cost-reduction framework. By contrast, enabling ALMPs such as training, job-matching services, or innovation-oriented measures appears conceptually disconnected from the dominant diagnosis of the problem. The findings thus suggest that the Italian ALMP mix cannot be fully understood without accounting for the ideational environment that privileges cost-reduction solutions over transformative interventions. Empirical results indicate that, despite economic constraints, firm characteristics undoubtedly matter, and the labour-cost narrative plays a constitutive role in defining which policy responses are perceived as appropriate, legitimate, and urgent.
The evidence also provides strong support to Hp2, indicating that employers’ organisations have been the primary promoters of the labour-cost narrative and have exercised ideational power by shaping public discourse on labour-market challenges. Employers’ representatives consistently articulate labour costs as the central barrier to job creation, while simultaneously framing fiscal incentives and contribution relief as the most effective policy responses. This framing is reproduced and amplified through influential media outlets, particularly Il Sole 24 Ore, which functions as a key transmission belt between business interests and broader public debate. From the perspective of ideational power, these findings illustrate how employers exert influence not only through lobbying or institutional access but also through discursive leadership. By repeatedly advancing a coherent and internally consistent narrative, employers’ organisations delimit the range of conceivable policy alternatives. In Schmidt’s (Reference Schmidt2008) terms, they exercise power in ideas by shaping how labour-market problems are understood, and power through discourse by normalising cost-reduction strategies as common-sense solutions. The relative absence of competing narratives further reinforces employers’ discursive advantage. Even when such themes appear, they tend to remain secondary or symbolic rather than constituting a sustained counter-frame.
In Hp3, we have assumed that the employer-driven labour-cost narrative would be reflected in parliamentary debates, thereby contributing to the political legitimation of employment subsidies. The analysis of parliamentary discourse confirms this expectation. References to high labour costs and the need to reduce the tax wedge recur across parties and governments, regardless of ideological orientation. Employment subsidies are routinely discussed as pragmatic tools to stimulate hiring, often detached from broader considerations regarding their effectiveness as activation measures or their long-term developmental impact. This cross-party convergence suggests that the labour-cost narrative has become deeply embedded within the political sphere, operating as a shared cognitive frame rather than a contested political position. As a result, weak labour demand is treated with technical instruments to correct cost distortions rather than as policy choices with distributive and structural implications. This helps explain their persistence over time and their resilience to changes in government composition.
Our analysis also highlights a difference in how employment subsidies feature in parliamentary debates compared to the broader public debate. Schmidt’s (Reference Schmidt2008) distinction between the coordinative and communicative spheres of policy supports our argument that there is consensus on subsidies as the main instrument to counter weak employment growth. In the coordinative sphere, parliamentary debates represent one of the technical and decision-making domains where policymakers and stakeholders engage with the specifics of policy instruments. Conversely, in the communicative sphere, which encompasses broader public debate and media discourse, policy issues are typically framed in wider and more contested terms, as political actors seek to set the agenda and advance competing solutions to salient social problems. The limited salience of employment subsidies in the communicative sphere does not signal their political marginality, but rather their discursive normalisation. Once cost reduction becomes the dominant diagnosis, subsidies can be treated as an almost self-evident policy response, requiring little public justification beyond generic references to job creation and competitiveness.
Taken together, these findings contribute to broader debates on active labour market policy by highlighting the role of ideas in shaping not only policy implementation. While existing literature has extensively examined the institutional and economic determinants of ALMPs, this study shows how discursive dynamics can systematically privilege specific instruments over others, even when their effectiveness is contested.
Conclusions
The policy mix of Italy’s ALMP reflects a deep-rooted prioritisation of employment subsidies as the cornerstone of active labour market interventions. This policy choice – historically entrenched and widely supported across the political spectrum – has sidelined more inclusive and transformative measures, such as training programmes and employment services, by helping employers pursue low-road strategies that reduce hiring costs. Scholars demonstrated that the Italian economy is dominated by small firms that exhibit limited orientation toward innovation and minimal inclination to invest in the skills and development of their workforce. This lack of investment in human capital keeps productivity and wages low, undermining Italy’s ability to pursue a growth strategy based on sustainable and transformative economic policies (Baccaro & Bulfone, Reference Baccaro, Bulfone, Baccaro, Blyth and Pontusson2022; Burroni et al., Reference Burroni, Pavolini, Raspanti, Hassel and Palier2025). Recently, the weakness of a ‘state-led’ growth strategy – emphasising government consumption through targeted budgetary investments – has further entrenched this cycle, making short-term fixes like business incentives and tax rebates the default approach to competitiveness through labour market policies.
This contribution proposes a complementary explanation for the dominance of employment subsidies in the supply of ALMP. Our analysis demonstrates that employers’ organisations’ have played a crucial role in shaping this policy landscape by promoting a narrative framing ‘excessive labour costs’ as the central problem in the Italian labour market. Employers’ associations have repeatedly argued that excessive labour costs deter hiring, particularly for low-skilled workers. Through this sustained narrative, employers’ organisations have exercised a form of ideational power that has consistently framed high labour costs – driven by (supposed) high levels of taxation and social contributions – as the primary obstacle to employment growth. As a result, employment subsidies have become the default policy instrument, prioritising cost reduction over alternative interventions rather than as part of a broader strategy to address Italy’s structural labour market challenges.
This policy orientation has been reinforced by the gradual adoption of a pro-business stance by parties across the political spectrum, reflecting the long-term shift in policy priorities towards an employer-friendly approach to labour market regulation that has begun in the 1980s (Rangone & Solari, Reference Rangone and Solari2012). We then argue that high levels of expenditure on employment subsidies are complementary to employers’ strategies, privileging a low-road to competitiveness. Consequently, interventions aimed at reducing labour costs – such as tax reliefs and cuts to employers’ social contributions – are frequently presented as the most effective instruments for boosting employment.
Our findings also contribute to the emerging debate on corporate welfare and its implications for state-business relations (Farnsworth, Reference Farnsworth2012; Bulfone et al., Reference Bulfone, Ergen and Kalaitzake2023; Bürbaumer & Pinsard, Reference Bürbaumer and Pinsard2025). This literature highlights how public interventions increasingly take the form of unconditional transfers to firms, often provided with weak or no conditionality regarding investment, employment, or innovation outcomes. In this perspective, Italy’s long-standing reliance on what we have called ‘demand-led employment subsidies’, which rely on contribution relief and tax credits for employers without introducing specific selective criteria (van der Aa & van Berkel, Reference van der Aa and van Berkel2014), can be understood as a particular manifestation of corporate welfare: state-financed support directed at employers without corresponding obligations to generate collective or developmental benefits. The persistence of such subsidies indicates a policy orientation that privileges employers’ short-term cost concerns over broader productivity and equity goals. By uncovering the narratives that justify and normalise these unconditional transfers, our analysis extends the corporate welfare debate beyond fiscal or institutional explanations. It introduces a discursive perspective that highlights how ideational power sustains employer-oriented state intervention and helps explain why such transfers persist despite their limited developmental returns.
Our study is not without limitations, which primarily stem from the nature of the datasets and the methodologies applied. These constraints should be acknowledged to contextualisse the findings and delineate the boundaries of their interpretive validity.
One limitation arises from the use of BERTopic for topic modelling, particularly in the second and third datasets. While BERTopic is a powerful tool for uncovering themes in large textual datasets, its effectiveness is generally enhanced when applied to corpora with substantial text volume. In this study, despite efforts to curate extensive datasets, the number of articles and segments was relatively modest in some cases, particularly in the Labour Cost Dataset. As a result, some of the identified topics showed overlap, making it difficult to distinguish between closely related themes. Nonetheless, the method proved valuable in identifying broad clusters of discourse and capturing dominant narratives, even if the granularity of thematic differentiation was occasionally constrained.
Additionally, reliance on media articles in the first three datasets introduces biases inherent to journalistic framing. Media outlets often prioritise sensational or commercially driven narratives that might not fully capture the breadth of business or policy perspectives. This limitation could have influenced the representation of business leaders’ views on employment and labour costs. Similarly, the Parliament Dataset, derived from parliamentary debates, while offering a granular lens into political discourse, may reflect rhetorical strategies and positional framing rather than genuine policy intentions or priorities.
The study’s mixed-methods approach, while offering a comprehensive framework, also entails inherent trade-offs. The integration of qualitative and computational methods required balancing depth with breadth, which may have led to some loss of nuance in interpreting the finer details of individual texts and in assessing the broader alignment between business and political discourses.
Finally, the paper does not include party politics in its analytical framework. Although ALMP remains under-debated by all Italian parties (Giuliani & Raspanti, Reference Giuliani and Raspanti2022), left-wing parties and unions have shifted their preferences regarding social policies during the analysed period (Natili and Puricelli, Reference Natili and Puricelli2023; Giuliani, Reference Giuliani2024). Parties’ and unions’ realignment on social issues may influence their preferences regarding employment subsidies. However, we do not focus directly on party politics. Including it would be beyond the scope of our contribution.
Despite these limitations, our analysis remains robust in illuminating key intersections between business leaders’ and policymakers’ narratives on employment and ALMP. Future studies could address these constraints by employing larger datasets, incorporating more diverse sources, or expanding the temporal scope to account for evolving dynamics in labour market discourse.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0047279426101378
Competing interests
The authors declare none.
Ethical considerations
All analysed materials are publicly available sources. This study does not involve human participants or confidential data and therefore does not raise specific ethical concerns. No ethical approval was required according to institutional or national guidelines.
Appendix A. Normative references
List of employment subsidies in Italy since the 1970s
Source: Authors’ elaboration.
Appendix B. Methodology
Hyperparameter selection for topic modelling
To determine the most effective configuration for topic modelling, a detailed evaluation was conducted using BERTopic with two embedding models: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 and sentence-transformers/paraphrase-multilingual-mpnet-base-v2.Footnote 5 Both embedding models were tested on the Active Labour Market Policy and Labour Cost Datasets to identify the optimal combination of min_cluster_size and min_samples hyperparameters. The former model consistently delivered the best performance across both datasets.
-
• min_cluster_size sets the minimum number of documents required to form a cluster.
-
• min_samples determines the number of samples in a neighbourhood for a point to be considered as a core point.
A thorough grid search approach enabled the identification of optimal hyperparameter combinations that maximised coherence while minimising noise and redundancy in the generated topics.
The following evaluation metrics were used:
-
• number_uncategorized: The number of documents that could not be assigned to any topic cluster.
-
• coherence_cv, coherence_umass, coherence_npmi: These metrics assess the quality of topic coherence, with the latter focusing on normalised pointwise mutual information (NPMI) to measure semantic similarity.
-
• topic_diversity: Measures the degree of differentiation in the vocabulary across topics to ensure that topics are distinct.
-
• inverted_rbo: A ranked-biased overlap (RBO) measure, inverted to penalise overlap between topics and encourage distinct clusters.
The grid search results were analysed (see Tables B1 and B2), and configurations with the highest coherence_npmi and topic_diversity scores were prioritised. Parallel coordinates plots (see Figures B1 and B2) were used to compare the configurations and identify the most promising hyperparameters combinations.
For the Active Labour Market Policy Dataset, an additional step was undertaken to manually merge similar topics. Similarity was assessed using a combination of approaches, including shared keywords and visual tools such as 2D representations of document embeddings. These visualisations, which reduce high-dimensional embeddings into two dimensions, facilitated the identification of clusters with conceptual overlap. This manual adjustment ensured that the final topics were distinct and more representative of the underlying discourse.
A final qualitative assessment was conducted on the top three configurations for each dataset to evaluate the relevance and quality of the topics. The rows highlighted in bold represent the final configurations selected for the models.
All models were trained on the Kaggle platform using a dual T4 GPU setup (T4X2), providing the computational capacity necessary for the extensive grid search and evaluations. This setup enabled efficient exploration of hyperparameter configurations and ensured optimal topic modelling performance across the datasets.
BERTopic full output with keywords
Hyperparameter combinations and their impact on metrics (Active Labour Market Policy Dataset).
Hyperparameter combinations and their impact on metrics (Labour Cost Dataset).
Top three configuration results for the active labour market policy dataset (ordered by coherence NPMI)
Top three configuration results for the labour cost dataset (ordered by coherence NPMI)
BERTopic output: Labour cost dataset
BERTopic output: Active labour market policy dataset
aMMR (Maximal Marginal Relevance) balances relevance and diversity in selecting representative words for topics.
bToo generic to be categorised precisely.
BERTopic full output with keywords
Roberto Rizza is Full Professor of Sociology of Economic and Labour Processes, currently serving as the Director of the PhD Program in Political and Social Sciences at the University of Bologna. His research explores the interplay between labour market policies and social inclusion, with a focus on policy implementation, work-career trajectories, and gender disparities.
Dario Raspanti is Assistant Professor at the University of Florence. His research interests include labour market policy, industrial relations, age management toward older workers, and the study of policy implementation from a street-level perspective.
Francesco Albanese is a PhD Candidate in Political and Social Sciences at the University of Bologna, pursuing a Joint PhD with Radboud University (Nijmegen) in collaboration with the Environmental Governance and Politics group. His research focuses on energy and environmental politics, harnessing Natural Language Processing techniques to perform empirical analyses of political discourse.
Open access
