Electoral Coalitions and Policy Reversals in Portugal and Italy in the Aftermath of the Eurozone Crisis

ABSTRACT This paper links patterns of deregulatory reforms and post-crisis reversals in Italy and Portugal with the electoral constituencies of political parties. Combining insights from the social coalitions and electoral behaviour debates, we link reform outcomes to the class set-up of the electorate in the two countries by using the class schema developed by Daniel Oesch. We find that governments in both countries reversed austerity measures in order to protect core groups of voters within their electorate in spite of wide ideological differences between governing coalitions in the two countries. There is, however, some variation in the extent of the reversals across policy areas.

how responsive governments have been to the electorate in the context of the crisis and its aftermath in Portugal and Italy. To do so, we compare reforms and reversals affecting two electorally salient policy areas: pensions and employment protection legislation (EPL).
Building on work analysing the sociological base of politics, we focus less on responsiveness towards public opinion or measured voter preferences (which tend to be fairly volatile and shaped by single policies) and more on responsiveness vis-à-vis the perceived material interests of different social groups, such as production workers, managers or small business owners. For instance, we argue that in countries where constituencies with interests harmed by liberalisation are large, politicians will have strong incentives to roll over austerity measures. The social setup of parties will also shape the nature of policies implemented. In other words, we look at how the social composition of party electorates constrains and shapes government policies. Our main contribution is to use a granular socio-economic analysis of party electorates to understand the incentives for parties to engage in policy reversals. In doing so, we draw on two bodies of literature. The first is grounded in political economy and assumes that public policies can be explained by the nature of coalitions between different social groups in society (see for example Iversen & Soskice 2006;Amable & Palombarini 2009;Baccaro & Pontusson 2018). The second body of literature examines the changing nature of class and how it shapes electoral behaviour (Oesch & Rennwald 2018). We combine these two strands to connect policy developments in the aftermath of the crisis in Portugal and Italy with the social setup of party electorates.
The rest of the paper is organised as follows. In section two, we outline the theoretical framework of our analysis illustrating the link between the class composition of the electorate and policy-makers' reform strategies. In section three, we explain the rationale for the selection of our countries and policy areas of interest and illustrate our empirical approach combining European Social Survey (ESS) data with the Oesch class categories. Section four provides some background information on the pre-crisis and post-crisis political context in Italy and Portugal. This is followed by an analysis of the social composition of the electorate in Italy and Portugal, and two empirical sections accounting for the pattern of reforms and reversals in our policy areas of interest.

Electoral class coalitions and policy reversals
As mentioned above, the eurozone crisis has coincided with a shrinkage of the domestic policy space. Armingeon and Baccaro (2012, p. 264) have argued that, at the peak of the crisis, 'domestic institutions and politics, either party-or interest group-based, have ostensibly played a minor role in selecting the policy response'. The point of departure of our analysis is that the domestic policy space expanded as external pressures subsided, leaving more room for domestic politics and coalitions to shape policy outcomes.
It has been common in political economy research to link public policies to the balance of power among social groups in society. While earlier Marxist theorising saw the political process as a battle for the control of the state between the working class and the bourgeoisie (see Marx & Engels 1848, p. 15), more recent studies have adopted a more nuanced view to this approach when analysing the material base of political cleavages. Iversen and Soskice (2006), for instance, look at how electoral institutions shape the alliance patterns between low, middle and high-income groups, resulting in different levels of redistribution. The recent volume edited by Beramendi et al. (2015) provides a prominent example of the use of social alliances and electoral politics as key variables to explain social policy outcomes. Beramendi et al. (2015) assume that political parties represent the interests of different social constituencies, and that the policies the elites put in place are constrained by the interests and preferences of these constituencies. At the micro-level, the kind of job they perform and have learnt is considered an important locus of political preference formation and socialisation (Kitschelt & Rehm 2014;Oesch & Rennwald 2018). Here, we also assume that there is a structural base for voter preferences and party support. This does not mean that the collective preferences of socio-economic groups are fixed and constant, but that politicians seek to appeal to the electorate by devising policies that match their social base of support or allow them to extend it. Similar to Iversen & Soskice, we 'assume that parties represent classes, or coalition of classes, and that it is difficult for parties to commit credibly to electoral platforms that deviate from the preferences of their constituents' (Iversen & Soskice 2006, p. 167).
In order to trace the connection between policies and social bases of support, we use a more sophisticated conception of class than Iversen & Soskice. This is done by applying Oesch's (2006) new class schema to Southern Europe. Oesch argues that post-industrial economies require a new way to conceptualise class, which cannot be reduced to a dichotomous (workers and managers) or purely unidimensional hierarchy (lower; middle; upper class). Instead, according to Oesch, patterns of class support for political parties cut across two cleavages: the traditional cleavage between workers and owners of capital, but also the cleavage between different 'work logics'. The latter distinguishes between the interpersonal work logic characterising sectors such as healthcare, social care, the media or the arts, and the more hierarchical work logic of production lines and organisations not directly involved in handling clients. In jobs guided by a hierarchical work logic, authority plays a bigger role, translating into more authoritarian values than those of 'interpersonal' workers. For the purpose of class voting, Oesch proposes a schema of eight social classes based on the characteristics of the jobs occupied: production workers (assemblers, carpenters, machinery mechanics); service workers (childminders; home helpers, waiters); clerks (secretaries; bank tellers); managers (financial managers; public administrators); technical professionals (computing professionals; mechanical engineers); small business owners (farmers; hairdressers; shopkeepers; owners of small businesses); socio-cultural professionals (teachers; medical doctors; social workers) and the traditional bourgeoisie of large employers and self-employed professionals (hotel owners; lawyers).
The main hypothesis of our analysis is that the relative size of these different groups within the electorate shapes alliance patterns and influences the viability of different policy agendas (Beramendi et al. 2015, p. 24;Amable 2017). Following this logic, we posit that office-seeking politicians will have to take two factors into account in their effort to create a sufficiently large base of electoral support for their policies (Amable & Palombarini 2009): the relative size of different groups of voters within the electoratedetermining the reformist agenda they will implement to broaden their base of supportand the relative weight of different groups of voters within their own electoral basedetermining the content of measures aimed at protecting their traditional voter base. By extension, the relative weight of these groups of voters within the electorate determines the likelihood and shape of policies. If the social base of support for a deregulatory agenda is too small, or in other words if most large groups of voters within an electorate are hit disproportionately hard by austerity, governments will face strong electoral incentives to roll over liberalising reforms. We do not assume a rigid and stable connection between specific social groups and policy preferences, because party alignments vary over time. According to Häusermann and Kriesi (2015, p. 217), the self-employed, technical experts, small business entrepreneurs and managers are more clearly to the right on economic issues (market liberalism) than production workers, while sociocultural professionals are equally hostile to market liberalism. They also find, however, that class has a low impact on socio-economic preferences in Southern Europe compared to the West and North, possibly due to a traditionally clientelistic mode of voter mobilisation (see Hopkin & Mastropaolo 2001;Afonso, Zartaloudis & Papadopoulos 2015). However, the fact that preferences across social classes do not differ much does not undermine the fact that their preferences matter to politicians; in fact, political elites in a clientelistic system are probably even more incentivised to satisfy their social constituencies.
It is important to situate our approach in relation to connected work. Recent analyses often use a broader and somewhat looser definition of the base of support of economic policy agendas, looking not only at the electorate itself but also at producer and interest groups yielding power outside electoral channels (Baccaro & Pontusson 2018). In many ways, it is reasonable to assume that electoral incentives are not the only inputs that shape and influence socio-economic agendas, and that the role of organised interests is important especially in areas that are characterised by 'quiet politics' (which are technical and do not feature prominently in the 'hot' politics of electoral campaigns) (Culpepper 2010). However, we still believe that electoral considerations have played a decisive role in shaping policy-makers' reform agendas, at least in high salience policy areas like pensions and EPL reforms.

Case selection
In this section, we justify the rationale for the selection of our case studies and policy areas of interest. The rationale for comparing Italy and Portugal is that these two countries were deeply affected by the eurozone crisis. While Portugal was under an explicit adjustment programme by a Troika of lenders -International Monetary Fund (IMF), European Central Bank (ECB), European Commission (EC)between 2011 and 2014, Italy was not. Instead, it found itself in a situation of 'implicit conditionality' where the implementation of reforms was an implicit condition for the ECB to continue buying government bonds (Sacchi 2015). Above all, Portugal and Italy provide an interesting variation in the relative strength of progressive and conservative forces. In fact, while the revolutionary experience has tilted the balance of the political spectrum in Portugal towards the left (Jalali 2007, p. 73), in the post-war period Italy has mostly been ruled by conservative parties. This long-term pattern is echoed in the partisan affiliation of the current governments, an alliance between the Socialists and the radical left in Portugal and a coalition between the radical right Lega and the catch-all populist M5S in Italy. This partisan difference provides an ideal focal point to study the variation in incentives to roll back similar austerity measures across the political spectrum.
Moving to the choice of our policy areas of interest, pensions and EPL were chosen because they are highly visible policy areas that underwent radical deregulatory reforms during the crisis. This makes them crucial most-likely cases to test our argument about policy reversals. High electoral salience motivates the decision to select EPL and pensions rather than other less visible policy areas that also underwent radical reforms like industrial relations (Bulfone & Afonso 2019).

Empirical approach and data
In the empirical section, we proceed in three steps to link the electoral size of different class groups with patterns of policy reversal and continuity. First, we compare the size of different electoral class groups in Italy and Portugal to that of other European countries, identifying similarities and differences. Second, we describe the 'social base' of the main parties in Italy and Portugal by providing a breakdown of the distribution of socio-economic groups across party electorates. Third, we link the patterns of reforms and reversals in our policy areas of interest to the social base of Italian and Portuguese parties.
We draw on data from the ESS, a widely used source of data on political attitudes and socio-economic conditions in European countries. This data source has limitations, however, when it comes to the countries under scrutiny: while Portugal has participated in all the eight waves of the ESS between 2002 and 2016, Italy has only taken part in four of them (in 2002, 2004, 2012 and 2016). Due to the small size of some voters' groups, we decided to merge all the waves of the ESS to present the class breakdown of parties. This produces a dataset with reliable estimates of the proportion of each class within each party electorate. However, merging all waves also has the significant drawback of hiding change in the social composition of party electorates over time. In an online Appendix (available at https://doi.org/10.1080/13608746.2019.1644809), we provide estimates of proportions for each ESS wave for each country, with corresponding confidence intervals. This analysis shows that the class setup of party electorates is fairly stable over time.
We applied the Oesch class schema to the data using the 'Oesch' package for Stata developed by Kaiser (2018). This schema is ideal to deal with post-industrial labour markets, and also has the advantage of being readily and easily applicable to a number of data sources including the ESS. Because some parties only have a We recoded party choice in the last election available in the ESS into a smaller number of categories. 1 In Italy, the Partito Democratico (PD -Democratic Party) and its forerunners the Left-Wing Democratici di Sinistra (DS -Left-Wing Democrats) and Margherita (Daisy) as centre-left; Forza Italia (FI -Go Italy) and its forerunner Popolo delle Libertà (PdL -People of Freedoms) as centre-right; the Lega and its forerunner Lega Nord (LN -Northern League) 2 along with Alleanza Nazionale (AN -National Alliance) and Setting aside individuals for whom one or the other dimension was missing, we obtained a sample of N = 3,271 observations in Italy and N = 11,614 in Portugal for all waves combined. We present a breakdown of party electorates by class for each country using post-stratification weights. As data for Italy in the ESS is less completein terms of wavesand does not cover the most recent 2018 election, we complement it with secondary data from a postelection survey by Ipsos (2018). In the next section, we provide an overview of the political and institutional context in which our analysis is embedded.

Political context: governments and coalitions in Italy and Portugal
Italy: from fragile bipolarism to fragmentation and instability Since the early 1990s, Italy has experienced a prolonged period of political instability with governments routinely failing to create a sufficiently large base of support for their reforms (Amable & Palombarini 2009;Amable, Guillaud & Palombarini 2011). Between 1994 and 2008, two coalitions competed for office: the Centre-Left (built around the PD and other minor centrist and leftist parties) and the Centre-Right (centring on Berlusconi's PdL/FI and the Lega). While the third-way pro-European Centre-Left relied on support from unionised workers of large firms, clerks, and public sector socio-cultural professionals, Silvio Berlusconi's Centre-Right was backed by the self-employed, small business owners, shopkeepers and their employees, mostly based in Northern Italy, as well as by Southern voters depending on state transfers in the form of employment in an oversized public sector (Barta 2018;Diamanti 2013). Both coalitions were inherently weak, albeit for different reasons. The Centre-Left electorate was internally coherent in its preferences, with production workers agreeing to limit their wage demands in exchange for high employment protection for labour insiders, and Northern public sector workers accepting the constraints deriving from the process of European integration because of their universalistic values. However, these two groups represent a minority within the electorate and the Centre-Left failed to win relevant support from other groups, managing to win the elections only twice, in 1996 and (by a razor-thin margin) in 2006. The electorally hegemonic Centre-Right coalition was instead paralysed by the incompatibility of the preferences of its two core constituencies: small business owners and shopkeepers from the North and low-income voters from the South. While on the one hand, small business owners, the self-employed and shopkeepers support budget cuts, lower taxes, a lenient attitude towards tax evasion and various anticompetitive protections from the state, fragile voters instead support generous transfers of public money to the Southern regions (Amable, Guillaud & Palombarini 2011;Barta 2018). The only way to reconcile these incoherent demands was through generous public spending.
By reducing the resources political actors could allocate to their voters, the crisis further complicated the formation of stable electoral coalitions. When Italy was hit by the crisis, the Centre-Right government was paralysed by the internal infighting between its Northern and Southern constituencies, and was thus forced out of office and replaced in late 2011 by the technocratic executive of Mario Monti (Bulfone & Tassinari 2019). Monti was supported by a very large majority headed by the PD and Berlusconi's PdL along with other minor centrist liststhe NL being the only large party in opposition (Bosco & McDonnell 2013). Given the internal distributional conflicts pervading both the centre-left and the centre-right, 'externalising' policymaking to a technocratic government was an expedient way to manage electoral blame.
The Monti government implemented reforms of the pension system and EPL that hit unionised production workers particularly hard, making them turn their back to the PD (Table 1). For its part the Centre-Right suffered an erosion of its support among small business owners, shopkeepers and Southern voters due to its perceived incapacity to react to the crisis. The contemporary weakening of both the main coalitions favoured the rise of the catch-all populist M5S, which won more than 25 per cent of the votes at the 2013 elections. While the PD won the elections by a narrow margin over the Centre-Right, both coalitions suffered large losses among their core constituencies: business owners and self-employed for the Centre-Right, and socio-cultural professionals for the Centre-Left (Diamanti 2013). Hence, the 2013 elections aggravated Italy's political crisis. According to Amable and Palombarini (2009), it is precisely during these phases of transition that political leaders have more leeway to shape innovative reformist strategies. This is precisely what the leader of the PD Matteo Renzi, in power between 2014 and 2016, tried to do with his reformist agenda 3 . Renzi severed the ties between the PD, trade unions and production workers, and instead promoted a centrist neo-liberal agenda aimed at winning votes from skilled labour market outsiders, the self-employed and small business owners (Picot & Tassinari 2015;Bulfone & Tassinari 2019).
This strategy was deemed to be an electoral failure, as Renzi antagonised Southern voters and production workers and failed to win votes among shopkeepers, the self-employed and small business owners, who remained loyal to the Centre-Right (Ipsos 2018, pp. 8-12). As a result, a growingly unpopular Renzi was forced to resign in 2016 being replaced by another PD-led government with Paolo Gentiloni as Prime Minister. In the 2018 elections the PD allied with small centrist parties fell to 22 per cent of the votes, finishing a distant third to the Centre-Right at 37 per cent (with the Lega replacing Berlusconi's FI as the largest party), and the M5S at 32.9 per cent, thus testifying the failure of Renzi's centrist strategy (Ipsos 2018).

Portugal: growing polarisation and the unification of the left
The post-authoritarian period (after 1974) in Portugal has been characterised by a competition for power between the centre-left (PS) and the centre-right bloc (PSD and CDS-PP in its different incarnations). The PCP was also a relevant player due to its strength in the rural south of the country and its strong ties with the  (2019) and Natili and Jessoula (2019) trade union movement (Watson 2015, p. 90). However, the PS never opened to the PCP, choosing instead to cooperate with the centre-right PSD leading to the de facto exclusion of the radical left from power (Jalali 2007, p. 196). The PS and the PSD 'consistently cooperated in an effort to destroy the communists' organisational base within the labour movement and competed over the same group of centre-right voters' (Watson 2015, p. 45), leading to a weak system of protection for workers. This strategy notably involved the creation of a new trade union, the UGT (União Geral de Trabalhadores -General Union of Workers) in direct competition with the Communist-linked CGTP (Confederação Geral dos Trabalhadores Portugueses -General Confederation of Portuguese Workers) (Jalali 2007, p. 90).
Since the 1980s, the weakening of the PCP and the 'unfreezing' of its electorate opened up competition for working class votes between the PS and the PSD. This in turn generated a convergence in party programmes coupled with a significant expansion of social protection for workers (Watson 2015, p. 164). Moreover, it weakened the class connotation of the PS and PSD electorates, with sociological variables becoming less important in explaining party choice (Jalali 2007, p. 106). As a consequence, Portuguese governments have enjoyed a substantial margin of manoeuvre in their economic policies also due to the generally low levels of political engagement from citizens and the weak social rooting of political parties (Magalhães 2005). Essentially, in the absence of credible challengers, electoral sanctions for unpopular policies mainly took the form of voter movements between the two centrist parties. Competition around the centre allowed for an expansion of social protection without significant fiscal imbalances, at least during the period of high growth between 1986 and 1992 (Amaral 2010, p. 35). After 2000, a policy of disinflation made inevitable by Portugal's membership in the European Union led to a phase of low growth and fiscal consolidation (Amaral 2010, p. 43).
The agenda of policy reforms in Portugal in the 2011-14 period was essentially guided by the memorandum of understanding (MoU) signed in early 2011 between the Portuguese government and the Troika (see Standring 2017 andBranco et al. 2019) (Table 2). While the bailout was requested by the PS government of José Socrates, the programme was implemented by the PSD government of Pedro Passos Coelho in coalition with the CDS-PP. In 2015, the alliance between the PSD and CDS-PP won the largest share of the popular vote in the elections but failed to secure a vote of confidence in parliament. This led to the formation of a PS-led government supported by the PCP and the BE (radical left) coming to power on a pledge to reverse a number of policies implemented during the adjustment programme and 'turn the page of austerity' to relaunch the economy (Costa 2015). This alliance, nicknamed the Geringonça (contraption) because of the uneasy fit between its constituent parties, formed the basis of a government coalition between the PS and the radical left for the first time in Portuguese history. The PS government depends on the external support of PCP and BE based on

Class structure of electoral coalitions in Italy and Portugal
From a comparative perspective, the first thing to bear in mind about the class structure of countries in Southern Europe is that it tends to have a greater proportion of production workers (with distinct preferences for state intervention) and small business owners, shopkeepers and the self-employed (which often require state 'protection' for instance in the form of a lenient attitude towards tax evasion); this contrasts with the higher proportion of service workers and managers in Northern Europe, which can constitute a larger base for pro-market policies. Figure 1 shows a breakdown of the class structure of selected European countries as measured by the Oesch class schema applied to round 8 of the ESS (2016; n = 39,983). This is important to situate the two countries under analysis within the European context. One can immediately detect the relatively large size of the constituency of (traditionally leftwing) production workers in our two countries compared to Northern Europe: production workers in Italy and Portugal represent twice the share found in the United Kingdom or the Netherlands and are the largest group in the schema, whereas service workers are the largest group in all other countries but Switzerland. Another important feature is the importance of small business owners (shopkeepers and owners of microbusinesses with fewer than ten employees): Italy and Portugal have the largest constituencies of this group of all the countries presented here. This is in line with Beramendi et al. (2015) who argue that production workers and small business owners are, more than elsewhere, pivotal groups of voters in Southern Europe. Let us first take a more detailed look at the class composition of party electorates in Italy (Table 3). As mentioned above, the two largest constituencies in the electorate are production workers and small business owners, the former representing the largest constituency of the PD and the M5S, and the latter constituting the largest social constituency of FI and the radical right (Lega, AN and FdI). Taken together, these two categories represent more than half of the electorate of FI and the radical right when merging all the ESS waves. Unfortunately, the available ESS data does not allow us to account for the recent profound transformation of the party system, with the rise of the Lega and the decline of FI. Overall, while we would usually assume right-wing parties to represent higher-income earners and advocate more free-market policies, the nature of party alignments in the Italian context partially contradicts these assumptions. What is particularly striking about the Italian case, at least when compared to other South European countries, is the underperformance of the PD among low-income voters. This is a consequence of the relative weakness of the left in Southern Italy, where most low-income voters are located. It is also due to the pervasiveness of the clientelistic networks created by the Democrazia Cristiana (Christian Democracy -DC) in the post-war Mezzogiorno and later inherited by the PdL/FI (Hopkin & Ignazi 2008). When the crisis weakened these networks, voters turned to the M5S rather than to the left (Ipsos 2018).
Our analysis of the class structure of the Italian electorate finds confirmation in an Ipsos study (2018) on the electoral base of support of the different coalitions at the 2018 elections that uses a different class schema. The PD emerges as the main party among high-income earners and retired workers, winning 22.5 per cent of the votes among managers and 27.6 per cent among the retired, while underperforming among production workers (11.3 per cent) and unemployed (ten per cent) (Ipsos 2018, pp. 8-12). This result is the mirror image of that of the Lega, strong among production workers (23.8 per cent), shopkeepers and small business owners (23.6 per cent). As a coalition, the Centre-Right -FI, Lega and FdIwas also strong among the unemployed, winning an aggregate 40 per cent of the votes (ibidem). Despite winning at least 30 per cent of the votes across all social classes except retired workers, the M5S was particularly strong among unemployed (37 per cent), production workers (37 per cent), housewives (36 per cent) and clerks (36 per cent). Furthermore, the M5S won 41 per cent of the votes among public sector employees and 35 per cent of the votes in the 18-34 age group (Ipsos 2018). On the other hand, the Lega performed particularly well in the 50-64 age group, obtaining 19.7 per cent of the votes. From this analysis, it clearly emerges that the M5S and the Lega outperformed all other competitors among the three pivotal electoral constituencies: production workers, Southern voters and small business owners (ibidem). Although they retained a large electoral base in the North, the M5S became hegemonic among fragile voters and young labour market outsiders in the South, attracted by the flagship economic proposal of the citizens' income ('reddito di cittadinanza') and the promise to repeal Renzi's EPL reform (see below). The Lega regained a strong base of support among shopkeepers and small business owners from the North-East and Lombardy (but this time also from Central Italy), and became the largest party among production workers. The electoral manifesto of the Lega centred on four main promises: a crackdown on illegal immigration, the introduction of a flat tax regime favourable to small businesses, a more lenient attitude to tax evasion, and the repeal of the pension reform implemented by the Monti government (Lega 2018). Turning now to Portugal, we can see both a number of similarities and also notable differences in the class setup of the electorate as a whole, and of party constituencies in particular, indicating once again a small base of social support for austerity and deregulation (Table 4). First, all parties, even on the right of the political spectrum, depend heavily on the support of production workers, and it is a very strong anchor for both the PS andeven more sothe PCP. This category alone represents a third of the electorate, and about 40 per cent of the PS and PCP electorates. Taken together, production and service workers represent nearly two thirds of the electorate of the PS and PCP, while the backbone of the centre-right is constituted by small business owners and shopkeepers, followed by production workers, just as in Italy. In general, it is the weakly differentiated electorate of the PS and the PSD that is striking: while the share of small business owners is predictably larger within the PSD, the socioeconomic profile of their electorates is actually quite similar. Interestingly, although the BE's agenda is often compared to that of the PCP, it has the least working-class profile of all Portuguese parties, andsomewhat similarly to Green or social-democratic parties in Northern Europehas an overrepresentation of socio-cultural professionals and intermediary professions (clerks). The PCP, in contrast, has a more traditionally working-class profile anchored among production workers.

Italy: reforms, social blocs, continuity and policy reversals
In this section we analyse reforms and post-crisis reversals affecting our two policy areas of interestpensions and EPLin Italy. We will show how the class set-up of the electorate presented in the previous sections helps make sense of patterns of reforms and partial reversals.

Pensions
Italy became involved in the Eurocrisis when it was ruled by the Centre-Right coalition of Silvio Berlusconi. In August 2011, at the peak of the crisis, the Berlusconi government received a letter from the ECB which implicitly conditioned the purchase of Italian sovereign bonds to the implementation of a series of deregulatory reforms. Commenting on the pension system, the letter asked the government to make the eligibility criteria for retirement pensions more stringent and to realign the retirement age for women in the private sector to that established for the public sector (Draghi & Trichet 2011). In October, the Berlusconi government attempted to pass a pension reform restricting the eligibility criteria for seniority pensions but had to back down due to the firm opposition from the LN. The LN opposed the reform because most seniority pensions go to manufacturing workers from Northern Italy, one of their reference groups of voters (Sacchi 2015, p. 83). Unable to implement other austerity measures due to coalitional infightings, and with the spread between Italian and German bonds spiralling out of control, the Centre-Right government was replaced in November 2011 by a technocratic executive guided by Mario Monti.
In December that same year the Monti government passed a reform of the pension system as part of the 'Save Italy' decree. The pension reform, named after the Minister of Labour, Elsa Fornero, equalised the pensionable age of genders and across professions at 66 years and 7 months, among the highest in the EU. Furthermore, 'seniority pensions were abolished and a new early retirement option was introduced with much tighter conditions' (Natili & Jessoula 2019, p. 350). The reform was implemented unilaterally without prior negotiations with the social partners and under strong pressure from the bond markets. This unilateral policy style was meant to obtain financial relief from the ECB on the bond market by signalling the government's commitment to austerity (Sacchi 2015). Although the LN voted against the reform, it could do little to prevent its implementation as it was now in opposition.
In the period between Monti's resignation in 2012 and 2018, Italy was ruled by three coalitions guided by the PD. As the PD was among the main supporters of the Monti government, none of these coalitions reversed the Fornero pension reform. During the 2018 electoral campaign, the leader of the Lega, Matteo Salvini, attacked the Fornero pension reform stating: 'After 41 years of employment, if you wish to, you have earned the sacrosanct right to get your money back . . . The Fornero reform is wrong and it has to be repealed.' (Repubblica tv 2018). The Lega's electoral manifesto called for the reintroduction of seniority pensions for workers with more than 40 years of contributions and the introduction of Quota 100 (Quota 100) to allow retirement upon reaching 100 as the sum of age and years of contributions (Lega Nord 2018, p. 5). When in power, the Lega earmarked €3.97 billion for the pension reform in the 2019 Budget Law promising to implement the Quota 100 scheme. This would therefore correspond to a reversal of the Fornero reform. Coherently with our framework, the Lega made the repeal of the Fornero pension reform one of its main political priorities because the reform hit one of its core groups of voters. In fact, many of the 660,000 workers that would benefit from Quota 100 are male workers from Northern regions like Lombardy, Veneto, Piedmont and Emilia-Romagna who started working early with a stable contract allowing them to reach service seniority at a relatively young age (Barbieri 2018). 4 An analysis of voters' age distribution confirms the salience of pensions amongst the Lega electorate. In fact, the Lega obtained 19.7 per cent of the votes in the 50-64 age group (against an overall result of 17.8 per cent), that is to say among the voters set to directly benefit from Quota 100.

Employment protection legislation
Throughout the 1990s and early 2000s, a series of labour market reforms liberalised the use of flexible employment contracts while maintaining a high level of employment protection for workers on open-ended contracts. This 'deregulation at the margins' led to the widespread diffusion of non-standard employment among young workers. Workers on open-ended contracts also benefit from very generous Short-Time Work (STW) schemes 5 , while unemployment benefits and activation policies for labour market outsiders were nearly absent (Sacchi & Roh 2016). The two main EPL reforms implemented during the crisis, the Fornero labour market reform implemented by the Monti government and the Jobs Act, both aimed to tackle the dual nature of the labour market. Implemented in June 2012, the Fornero labour market reform lowered the cost of dismissal for workers on open-ended contracts and reduced the possibility of reinstatement in case of discriminatory dismissal, thereby watering down the very symbolic Article 18 of the Workers' Statute. At the same time, the reform strengthened the employment benefits for fixed-term contracts (Picot & Tassinari 2017).
The PD-led grand coalition government of Matteo Renzi implemented a second EPL reform, the Jobs Act. Following in the path of the Fornero reform, the Jobs Act further reduces employment protection for workers on openended contract by introducing a new open-ended contract with dismissal protection related to seniority. The new contract excludes the possibility of reinstatement for economic dismissals and, in case of unfair dismissals and establishes a monetary compensation between four and 24 salaries (depending on seniority). The Jobs Act also strengthens the unemployment insurance scheme for temporary contracts and limits the use of the Cassa Integrazione Straordinaria (the structural STW scheme) (Picot & Tassinari 2017). The Jobs Act represents a sharp break with the previous reformist strategy of the Centre-Left in that, for the first time, the PD targeted labour market insiders leading to an open confrontation with the main Italian trade union the Confederazione Generale Italiana del Lavoro (Italian General Confederation of Labour -CGIL). Renzi explicitly accused trade unions of defending the worker aristocracy at the expense of fragile young workers. Despite this pro-outsider rhetoric, however, the Jobs Act came under heavy criticism from associations representing young labour market outsiders (as well as from the traditional trade unions) (Picot & Tassinari 2015).
In electoral terms, the Jobs Act can be seen as an attempt to build a new socio-economic base of support for the PD (Vesan 2017;Picot & Tassinari 2017), a centrist pro-European bloc formed by skilled labour market outsiders, large firms, small business owners and self-employed professionals (akin to Amable & Palombarini's bloc bourgeois). The contours of this strategy appear clear when looking at other elements of the reformist agenda of the Renzi government. For instance, in 2015 the executive raised the limit on cash transactions to 3,000 euros, rolling over a previous measure by the Monti government that had fixed the limit at 1,000 euros. This move by the government was perceived as a way of courting socio-economic groups responsible for a large share of tax evasion in Italy like 'small-business owners and the self-employed (including many doctors and lawyers), who have traditionally voted for the right.' (The Economist 2015). However, the reformist strategy did not pay off in electoral terms, as the PD came a distant third in the 2018 elections.
During the 2018 electoral campaign, the prime ministerial candidate of the M5S Luigi Di Maio promised to 'Abolish the Jobs Act' (La Repubblica 2018). Once in power, the M5S was the main sponsor of the Decreto Dignita' (Dignity Decree) that limited the scope of application of the Jobs Act. Among other things, the Dignity Decree raised the monetary compensation for unjustified dismissals or dismissals for economic reasons to a minimum of six salaries and a maximum of 36 (from four and 24). The Dignity Decree also reduced the maximum number of temporary contracts an employer can sign with an employee before having to offer a permanent contract from five to four. A further reversal of the Jobs Act occurred in September 2018 when the government reintroduced the structural STW scheme for firms facing the prospect of a closure (Branco et al. 2019).
Despite restricting the use of temporary contracts, the Dignity Decree clearly fell short of reversing the Jobs Act. Most notably, it did not reintroduce the obligation of reinstatement in case of economic dismissals granted prior to the crisis by Article 18. The M5S and the Lega voted against an amendment of the radical left asking for the reintroduction of Article 18 for all workers (La Stampa 2018). Hence, the reversal was less radical than in the case of the pension reform. The limited nature of this second reversal is due to firm opposition from the Lega to measures that would go against the interests of business owners and shopkeepers by making the labour market more rigid (Bulfone & Tassinari 2019).

Portugal: electoral coalitions and policy reforms
This section analyses reforms and post-crisis reversals in pensions and EPL in Portugal, linking the social class setup of parties presented above with concrete legislative changes in these two areas.

Pension reforms
Adjustments in the pension system had started under the PS government of José Socrates a few years before the eurozone crisis. The first budget submitted by its government in 2005 led to a convergence between the many public sector employment pension regimes and the general social security scheme. This was achieved by increasing the retirement age and the minimum level of contribution for special regimes, and moving public servants away to less generous ones. A new calculation system was also later introduced, setting the level of benefits based on the entire contributory career rather than the best years, and making the access to early retirement more difficult. On the eve of the bailout, an extraordinary surcharge of ten per cent (Contribuição Extraordinária de Solidariedade -CES) was introduced on pensions over 5,000 EUR a month (Martins 2015).
The adjustment programme signed between the Troika of international lenders and the Portuguese government in 2011 led to significant cuts in public pensions (in real terms, through a pension freeze). New taxes were introduced, starting with taxes affecting those at the top of the income distribution and in a second phase affecting those with lower incomes. At the same time, the PSD government of Pedro Passos Coelho also increased the lowest pensions at the bottom of the spectrum. In 2012, the CES mentioned above was expanded to apply at a rate of 25 per cent for pensions over 5,030 EUR and 50 per cent over 7,545 EUR (which represents more than eight times the average wage). Holiday and Christmas bonuses, a widespread method to top up wages and pensions, were cut for pensions over 1,100 EUR, and reduced for pensions between 600 and 1,100 EUR. Later that year, this was blocked by the Constitutional Court.
In 2013, the CES was again further extended in its application to pensions over 1,350 EUR (with rates between 3.5 per cent and ten per cent). Another attempt to cut holiday and Christmas subsidies in a way that would comply with constitutional principles failed, and was scrapped by the Constitutional Tribunal (Martins 2015). In August, the government negotiated an agreement with the unions on a legal reform converging the conditions of the pension scheme for public servants (Caixa Geral de Aposentações) and the general social security regime applying to private sector employees. In the same year, the legal retirement age was increased to 66 years (OECD 2017). Lower pensions continued to be indexed to inflation while the others were frozen. A new formula was adopted to automatically increase the retirement age in line with life expectancy, potentially raising it to 68.8 by 2060. Another attempt to tax pensions over 1,000 EUR was blocked by the Constitutional Court, meaning further reforms would be at an impasse in the absence of an agreement with the PS (Expresso 2014). With elections approaching, the government rolled back a number of the austerity reforms introduced earlier, targeting the solidarity contribution to high pensions over 4,611 EUR, and easing access to early retirement for individuals over the age of 60 and with 40 years of contributions (Martins 2015).
After the accession to power of the Geringonça, 'unfreezing' pensions was a central point in the political agreements signed in 2015 between the PS, PCP and BE to ensure parliamentary support (PS & PCP 2015). Accordingly, one of the measures introduced by the left-wing government was an unfreezing of pensions and their adjustment to inflation, and a generalised increase in their amount (Ionline 2018). The 2017 and 2018 budget laws provided for an across the board ten EUR increase in pensions, and there were discussions on the possibility of rolling back restrictions on early retirement, allowing workers with enough years of contributions to retire before the legal age. The restoration of the purchasing power of pensioners was a strong priority for the government and its allies, particularly the PCP. According to the ESS data used above, we can establish that 60 per cent of PS voters and 57 per cent of PCP voters are over the age of 55, compared with a mere 30 per cent among BE voters. Overall, we can highlight the concentration of austerity measures through taxes targeting higher-income pensioners, while lower incomes were spared. This policy was also pursued by the PSD government of Pedro Passos Coelho. As mentioned above, production workers are the largest constituency of all parties. This relates to the contribution by Perez and Matsaganis (2019) in this issue, which shows that Portugal is the only case in Southern Europe in which the implementation of austerity led to a reduction in inequality.

Employment protection legislation
If one believes in the accuracy of the OECD EPL index, Portugal was an outlier prior to the crisis due to its high level of protection for permanent contracts, whereas protection for temporary contracts was in line with the OECD average. The main objective of the reforms implemented by the Passos Coelho government in its revision of the Labour Code in 2012 was to bring protection for permanent contracts in line with the OECD average, thereby allowing for quicker downward wage adjustments. A major revision of the Labour Code (Law 23/2012) was adopted in 2012, providing for a 50 per cent decrease in compensation for extra hours, a reduction in severance payments, a loosening of the conditions for 'fair dismissal', fewer holidays and the introduction of an 'hours bank' extending by two hours the maximum duration of the normal working day, which could be compensated later on ((Diario da Republica 2012)). Severance payments were a particularly important element of the reform in this domain. They were cut from 30 days per year of seniority to 12, thus reducing the cost of dismissal of olderand more expensiveworkers.
While the reversal of EPL reforms passed under the MoU featured prominently in the electoral manifesto of the PS and other left parties, reforms have been more limited than in the field of pensions. Eventually, the Geringonça government proposed a package of legislation to reform employment protection in June 2018 that largely formalised an agreement reached with employers and (the smaller) trade unions (Martins 2018a;Da Silva 2018). One point of tension between the PS and the BE was the increase in the maximum probation period for new contracts from 90 to 180 days. Importantly, the reduction in severance payments introduced by the 2012 Labour Code revision was preserved (Publico 2018). In the end, the BE and PCP did not vote in favour of the package of legislation, which was enabled instead by the abstention of the PSD similarly to what had been observed at the start of the crisis (Martins 2018b). Whereas the BE was particularly vocal in opposing these reforms, they seemed to be less of a priority for the PCP. This can be understood in the light of the social setup of the electorate of these parties: the BE has the highest share of voters without a permanent contract. Hence, while the legislative package led to a partial (and modest) reversal, it preserved many measures adopted during the crisis to increase labour market flexibility (Moury, Cardoso & Gago 2019;Branco et al. 2019).

Conclusion
In this article, we have provided a granular socio-economic analysis of the electorate of two South European countries -Italy and Portugallinking it to patterns of deregulatory reforms and post-crisis reversals in pensions and EPL. This allowed us to combine the insights from two prominent debates respectively linking reformist outcomes to social alliances (Iversen & Soskice 2006;Amable & Palombarini 2009;Beramendi et al. 2015;Baccaro & Pontusson 2018;Oesch & Rennwald 2018) and electoral class behaviour (Beramendi et al. 2015;Oesch & Rennwald 2018).
The theoretical framework we employed is based on the assumption that politicians seeking to broaden their electoral base of supportand therefore win or retain officewill shape their reform agendas around the preferences of the main groups of voters belonging to their electoral base, or of those that they are willing to attract within specific class groups. In doing so, they will be influenced by two factors: the size of specific groups of voters within the electorate, and the weight of specific groups of voters within their own electoral base. Applying this scheme to post-crisis reversals of austerity measures in Southern Europe, we argued that policy reversals would in principle be more frequent in countries in which pivotal groups of voters are hostile to liberalisation. Our analysis of the class set-up of Italy and Portugal showed that, in both countries, governments that came into office after the crisis faced strong incentives to reverse deregulatory reforms of EPL and pensions. This is due to the weight of constituencies like production workers, perspective retirees and young workers within their electorate. And indeed, both Portugal's left-wing coalition and the alliance between the Lega and the M5S partially reversed deregulatory reforms implemented by previous governments. However, contrary to our expectations, there was variation in the extent of the reversals across policy areas, with reversals being more radical in the field of pensions than in EPL. This might be due to the fact that the reversal of an EPL deregulatory reform entails an immediate cost for employers, while the cost of reversing a pension reform is more long-term and weights on the general public budget. For this reason, employers might be more willing to lobby against a reversal in the field of EPL than in pensions. This in turns weakens to some extent the impact of electoral considerations on executives' reformist strategies in EPL. Future research should look at variation in the relative weight of voters' preferences, interests of producer groups and exogenous pressure in shaping executives' reformist agendas across different policy areas.
The findings presented here should also be refined, and extended, by applying a similar theoretical framework to the study of reforms and reversals in other South European countries. For instance, Spain's class profile seems to be more in line with that of other continental economies, with fewer production workers, more service workers and (slightly fewer) small business owners and shopkeepers than Italy and Portugal. Scholars should assess the impact of this different class configuration on the patterns of reform and reversal there. Another promising research avenue is to look more systematically at cross-country variation in the ideological preferences of pivotal groups of votersand their impact on governments' reformist strategies. For instance, while rural and low-income voters historically developed strong ties with the left in Spain and Portugal (Hopkin & Mastropaolo 2001), their counterparts in Southern Italy had always supported conservative forces (the DC and later the Centre-Right), before recently turning to the M5S. This difference in electoral preferences of low-income voters has the important effect of limiting the scope to implement a left-wing anti-austerity reformist agenda in Italy. Further research should extend this analysis to other pivotal groups of voters in countries with fragmented electorates like Greece and France. Portugal: waves 1 and 2 relate to 1999; waves 3 and 4 relate to 2005; wave 5 relates to 2009; wave 6 and 7 relate to 2011; wave 8 relates to 2015. 2. The Lega Nord was rebranded Lega in 2018 to appeal to a nationwide audience. 3. Renzi led a coalition between the PD and some smaller centrist and conservative parties. He succeeded Enrico Letta, also from the PD, who had formed a large coalition government after the 2013 elections. The change in government was the result of an internal coup within the PD. 4. While Lombardy, Veneto and Piedmont are traditional electoral strongholds, the Lega also broadened its base of support in the former left-leaning Emilia-Romagna reaching 19 per cent of the votes. 5. STW schemes subsidise a temporary reduction in working hours in a way to allow firms in crisis to maintain an existing employer/employee relationship.