Beyond Conditionality: Policy Reversals in Southern Europe in the Aftermath of the Eurozone Crisis

ABSTRACT This article proposes a framework to understand and explain the occurrence of policy reversals. We argue that the occurrence and absence of policy reversals is shaped by the constraints of responsiveness (to voters) and responsibility (vis-à-vis creditors, international institutions and financial markets). We review the literature on reversals and their implications for Southern Europe. We finally summarise the main findings of the contributions in the volume, that address when and why governments prioritise responsiveness or responsibility, as well as the economic consequences of these choices.

are expected to act prudently and consistently and to follow accepted procedural norms and practices, [. . .] living up to the commitments that have been entered into by their predecessors in office and abiding by agreements that these predecessors have made with other governments and institutions (Mair 2009, p. 12). In contrast, responsive governments 'listen to and then respond to the demands of citizens and groups' (Mair 2009, p. 12). This tension is at the core of the questions addressed in this volume because the policy reforms made during the crisis to honour commitments with international institutions and appease financial markets were often at odds with the demands of the electorate of the countries in which they were implemented. Put in slightly simplistic terms, while responsibility would lead to continuity, responsiveness would lead to reversals.
A superficial look at policy reversals after the crisis seems to suggest that measures adopted under conditionality are principally shaped by the ideology of governing parties. For example, the governments led by Mariano Rajoy in Spain (2011Spain ( -2018 and by Matteo Renzi (2014Renzi ( -2016 and Paolo Gentiloni in Italy (2016)(2017)(2018) appear to have persisted in the policy course pursued during the crisis, while the following governmentsthe socialist Sánchez I government in Spain (2018Spain ( -2019 and the populist Conte government in Italy (installed in 2018)have declared their willingness to break with austerity and water down past reforms. The left-wing Syriza government in Greece, which in August 2018 exited the third Greek economic adjustment programme, in Spring 2019 partially rolled back the VAT increases adopted during the crisis and made additional pension payments. In Portugal, the left-wing government that has held office from 2015 to 2019 is often presented as the flagship of successful reversals (Jones 2017). There is more than meets the eye, however. As contributions in this volume make clear, previous (right-wing) governments in Spain and Portugal reversed several important policies as soon as external constraints weakened, whereas there is more continuity in the current governments than what we often read in the press. 1 In this volume, we aim to study empirically the political aftermath of the eurozone crisis in Southern Europe. We pursue three objectives. A first objective of the volume is descriptive, namely to understand whether the policies, institutions, and processes introduced in the framework of conditionality have proved resilient, or whether the end of conditionality has marked a return to the statu quo ante. The second objective is explanatory; the contributions seek to explain why some changes persist and others do not, exploring possible drivers of continuity and change such as electoral demands or institutions. The third objective is evaluative. Namely, we ask what the social and economic impacts of different policy paths have been, on indicators of growth and poverty for instance. These three objectives are undeniably important from both a practical and a theoretical perspective. Yet with the exception of studies on compliance after Eastern enlargement (for example, Epstein & Sedelmeier 2009;Levitz & Pop-Eleches 2010) and on the long-term economic effects of IMF programmes (Remmer 1986;Vreeland 2005), scholars have made surprisingly few systematic efforts to understand policy reversals in the aftermath of conditionality. Instead, most of the contributions herein look at policy reversals. Some articles focus only on structural reforms in Italy, Spain and Portugal (Branco et al. 2019;Afonso & Bulfone 2019), whereas another  observes all kinds of policies triggered by conditionality in Spain and Portugal. Finally, Perez and Matsaganis (2019) analyse the social effects of reforms of the labour market passed during the crisis in both these countries and Greece.
We chose to study post-conditionality in these four countries because they were all profoundly affected by the eurozone crisis and carried out deep spending cuts and broad structural reforms. Conditionality, however, operated through different channels. In Portugal and Greece, governments committed to taking policy actions as defined in Memoranda of Understanding (MoUs) negotiated with the EU and the IMF in exchange for financial support. In Spain, MoUs only concerned the financial sector but in 2011 the Commission (mandated by the Council) started a programme of strict monitoring of fiscal consolidation measures and structural reforms, and linked its recommendations for change to the disbursement of the loan instalments . In Italy, the same monitoring programme was installed, and (as in Spain and Portugal before the bailouts) the ECB more or less explicitly made its intervention on the markets subject to specific policy reforms.
In this introduction, we first present the extant literature on reversals. We then describe the context in which South European countries were operating during the economic and financial crisis, and discuss what happened after conditionality. We finally summarise most important findings of the contributions of this volume.

What do we mean by reversals?
The literature has not as yet provided a conclusive definition of reversals, but two different types of phenomena can be distinguished. On the one hand, in the case of liberalisation efforts, scholars studying the political economy of reforms operationalise reversals as a decline in the value of pro-market indicators. For example, Merlevede (2003) codifies reversals as a downgrading of the transition indicators developed by the European Bank for Reconstruction and Development. These indicators reflect the average policy stance with respect to price liberalisation, trade and foreign exchange liberalisation, privatisation, restructuring and financial market reform. Similarly, Campos and Horváth (2012) build indexes of price and wage liberalisation; trade barriers and capital control liberalisation, and privatisation reforms; and subsequently codify reversals as a decrease in the value of these indexes. Abiad and Mody (2005), Agnello et al. (2015) and Rajan and Zingales (2001) have also operationalised reversals along these lines. On the other hand, scholars sometimes understand reversals as a policy change, whichever its direction (either less or more liberalisation), that occurs after conditionality. In that context, reversals are defined as a return to the status quo ante (Johnson 2008), heterodox deviations from IMF policy prescriptions (Pop-Eleches 2009) or as the undoing of pre-accession institutional change (Sedelmeier 2014).
Similarly, 'backsliding'a concept often used as a synonymous for reversalsis not explicitly defined in the literature. For some authors, backsliding is interpreted as a re-installment of spending levels once conditionality expires (Rickard & Caraway 2014); or a reversal of structural reforms adopted under the IMF (Duncan 2002). In this volume, we look at the reversals (or maintenance) of reforms and processes. We define policy reversals as a change to the reforms and processes towards the status quo ante prior to the crisis.

How frequently do reversals occur?
How common are reversals? The political economists cited above found that the undoing of pro-market reforms (whether or not they had been induced by conditionality) is not a rare occurrence. Rajan and Zingales (2001), for example, document a 'great reversal' of financial liberalisation in the 1930s, when free trade was replaced by a rapid return to protectionism. Similarly, in their study on financial reforms in 35 countries between 1973 and 1995, Abiad and Mody (2005) counted 27 reversals (13 per cent) out of 186 policy changes. Using the yearly transition indicators developed by the European Bank for Reconstruction and Development in 25 countries, Merlevede (2003) observes reversals in liberalisation in 8.9 per cent of the cases. Campos and Horváth (2012) study liberalisation and privatisation reform efforts for 25 Eastern European and former Soviet Union economies between 1989 and 2001 and show that reversals occur in 14-20 per cent of the cases depending on the reform indicators. Rickard and Caraway (2019) look specifically at the reversals of reforms induced by IMF conditionality in the period from 1980 to 2014, and observe that cuts to the public sector wage do not persist in the longer-term.
On the other hand, enlargement scholars observe that reversals were rare in the first decade after accession (Levitz & Pop-Eleches 2010;Epstein & Sedelmeier 2009;Sedelmeier 2012), with the exception of highly contested commitments (Sasse 2008;Grabbe 2014). In a recent paper, Sitter et al. (2016) aim to systematically map reversals of democratisation in the EU in the areas of the rule of law, corruption and equality. They show that Hungary, Poland, Romania and Bulgaria represent the most serious cases of democratic backsliding. However, other member states, such as Slovakia, Slovenia, Italy and Greece (countries with a worsening corruption problem) perform poorly; and in terms of reversals of the equality indicators, Hungary is joined by four new member states (Czech Republic, Latvia, Lithuania, Romania) and four older ones (Ireland, Italy, Spain, UK).

Why do reversals occur?
As stated above, the second objective of this volume is to explain the scope and variation of reversals. The literature suggests governments that witness the termination (or diminution) of conditionality are subject to two opposing pressureswhich largely correspond to the tensions between the 'responsibility' and 'responsiveness' of governments.
On the one hand, governments are under pressure to keep the reforms in place. First, conditionality does not completely go away in most cases (see below). Additionally, policy changes introduced during the crisis were not always entirely imposed 'from above'; indeed, there was often domestic support for changes. This is in line with research showing that, in contrast to earlier assumptions, austerity and retrenchment are not always unpopular (Giger & Nelson 2011). In some cases, governments took the opportunity of the crisis to introduce reforms that they themselves desired but had felt unable to introduce prior to the crisis (see for example Hence, decision-makers may often choose not to reverse the reforms introduced simply because they correspond to their primary preferences. Because austerity reforms are actually popular in certain constituencies (Giger & Nelson 2011;Stanley 2014), there is not always an incentive to reverse them. This is why it is necessary to temper the above statement that 'responsiveness' should lead to reversals.
Moreover, policy changes may have altered domestic power relationships so that reversals become less likely. For instance, labour market reforms introduced during the crisis might have weakened the power resources available to those wishing to come back to the status quo ante, such as the trade unions (Cioffi & Dubin 2016). A final factor favouring the status quo is the simple fact that, in policy-making, stability is easier than changegiven the existence of veto players and positive feedback (Tsebelis 1995;Pierson 1995). Even if policy reforms adopted by previous governments may not align with the preferences of new governments, they may prioritise other policy domains. Moreover, new policies may create powerful new constituencies of support opposing their reversal, because societal and economic actors adapt to new regulations and usually dislike policy volatility as it creates uncertainty.
However, governments are also pressured to come back to the status quo ante. First, not all policies were intended to be permanentsome were clearly meant to be temporary from the outset. Second, some policies are disliked by incumbentsbecause of their party ideology and/or because they proved costly in terms of votes (Afonso, Zartaloudis & Papadopoulos 2015). In particular, the reversal of a reform that was politically costly might pay off electorally. Reversals of unpopular reforms can occur even when the changes induced by conditionality correspond to the incumbent's preferences. There, the success of radical and/or populist parties may put some pressure on governments to reverse their discourse and policies (Mudde 2013, but see Rooduijn, de Lange & van der Brug 2014). Moreover, governments that had used conditionality to pass potentially unpopular measures cannot easily defend their maintenance, even if they were their own measures (Johnson 2008). In that line, Alesina, Ardagna and Trebbi (2006) show that reversals are more likely to occur close to elections.
Finally, the failure, or limited success of reforms might pressure governments to reverse them. For example, Chen (2009) argues that the failure of the privatisation of a given service often leads to a reversal of the delegation. Agnello et al. (2015) show that episodes of negative growth are significantly associated with reversals of structural reforms. Given these two conflicting pressures between change and stability, responsibility and responsiveness, we expect contradictory (or compensatory) movements; towards change and towards persistence.

Under which conditions do policy reversals occur?
Given the two contrasting pressures towards reversal and stability, it is crucial to understand why and when reversals occur. In the literature, we find that variation across countries and areas is explained by (1) the costs and benefits of reversals, (2) the number and preferences of veto players, (3) government composition and partisanship, and (4) the success of social learning and persuasion that had occured during conditionality.
As regards the first point, many scholars indeed argue that cost-benefit considerations are important in explaining stability after conditionality. Enlargement scholars assume that governments weigh the costs of compliance with the EU against the costs of the potential sanctions and the domestic costs of change (Johnson 2008). In a similar vein, political economists looking at liberalisation and privatisation reversals demonstrate that a change in material conditions leads to a re-calculation of incentives for stability or change. Dewatripont and Roland (1995), for example, argue that unfavourable changes in economic conditions lead to reform reversals (see also Agnello et al. 2015). Abiad and Mody (2005) show that banking crises foster reversals of liberalisation processes. Campos and Horváth (2012) find that different factors matter differently for different reforms, so that (1) foreign direct investment makes reversals of privatisation less likely; (2) a deterioration in the terms of trade has a positive impact on the probability of external liberalisation reversals; and (3) labour strikes increase the likelihood of reversals of price liberalisation. Moreover, IMF scholars and experts argue that reversals are more likely in areas where the domestic costs of reversals are low and/or when they bring high electoral benefitssuch as in the case of public sector wages and employment (Independent Evaluation Office 2003; Rickard & Caraway 2018).
Second, scholars have argued and demonstrated that reversals are less likely if the agreement of multiple actors ('veto players') is necessary to return to the status quo ex ante. In that vein, Sedelmeier (2012) shows that gender equality institutions that had been 'locked-in' during the accession process survive despite government disapproval. Similarly, Gehlbach and Malesky (2010) demonstrate that reform reversals are negatively correlated with the presence of multiple veto players.
Third, research shows that partisan preferences also matter for reversals insofar as governments or a particular political persuasion will seek to undo the reforms implemented by governments of an opposite political persuasion. For example, Pop-Eleches (2009) argues that, in Latin America and Eastern Europe, reversals during or after conditionality occur when a left-wing government follows a right-wing one (see also Campos & Horváth 2012). Sedelmeier (2012) show that, in the absence of lock-in, reversals of gender equality institutions are explained by a change in government orientation.
Finally, variations in reversals have been explained by social learning. Scholars studying enlargement have shown that the surprising lack of reversals in the first decade after enlargement is due to the transfer of norms through social learning and persuasion that had occurred earlier (see the contributions to Epstein & Sedelmeier 2008;Grabbe 2014). Consequently, reform reversals are dependent on whether or not this background of social learning had been successful (Sasse 2008).

Conditionality in Southern Europe: policy-making in straitjackets
One of the main lessons of the eurozone crisis, and especially its manifestations in countries where 'adjustment programmes' under direct conditionality were implemented, has been the extent to which European integration can constrain domestic policy-making in unprecedented ways. This can be explained by the fact that the institutional setup of the eurozone has changed the power balance between EU institutions and member states, giving rise to a new type of 'policy intrusiveness' in economic and social governance (Theodoropoulou 2014). Largely due to the power given to the European Central Bank as potential lender of last resort (De Grauwe 2012), the crisis gave European institutions an unprecedented degree of potential leverage in domestic economic policies when Southern countries found themselves in a liquidity crisis. This was a result of the transfer of monetary policy competences to the EU level, combined with a loss of investors' confidence in the ability of South European governments to service their debts (Afonso 2019, p. 4) Indeed, prior to the single currency, European governments controlled the currency in which this debt was issued, thereby providing an implicit guarantee that money could be printed by national central banks to pay bondholders (De Grauwe 2012, p. 4). However, since the creation of the eurozone, member states issue bonds in a currency they can no longer control: the European Central Bank (ECB) is independent and beyond the sole control of any single member state. Within this new architecture, when markets lose confidence in a government's ability to repay its debts, investors can sell its bondspushing borrowing costs upwardsto buy bonds of a more credible country (e.g. Germany) denominated in the same currency (De Grauwe 2011). This ability of creditors to sell and buy government bonds within the same currency potentially allows for the skyrocketing of borrowing costs in one member state and their rapid decline in another, without the ability for the former to issue currency to pay the higher yields, thereby resulting in a liquidity crisis (Matthijs & Blyth 2015). This mechanism was essentially hidden prior to the eurozone crisis because of the convergence of interest rates within eurozone countries at the onset of the introduction of the Euro. In this context, the EU's influence on domestic social and economic policies was fairly small: as there was no difficulty for governments to borrow on financial markets, the ECB did not have a political instrument it could use to pressure governments (Blavoukos & Pagoulatos 2008).
However, when investors lost confidencenotably as a result of revelations about the manipulation of public account data in Greece in 2009interest on government bonds in countries such as Greece, Portugal, and Ireland skyrocketed, making it impossible for them to roll over their debt. Consequently, those countries turned to international financial institutions for rescue packages, the ECB, International Monetary Fund (IMF) and EU. Bond yields on Spanish and Italian debt also increased due to fears of contagion. Because it was now the only actor able to issue currency, countries in fiscal distress depended heavily on the European Central Bank to purchase government bonds (in the secondary market only, because it is not allowed to buy bonds directly) and give buyers of Greek, Portuguese or Spanish debt a guarantee that they would be able to resell it. This is what the ECB started doing through a series of successive programmesthe Security markets programme (SMP), Outright Monetary Transactions (OMT), and expanded asset purchase programme (APS). The ECB's intervention has been considered central in reducing yields on government bonds, especially after its President Mario Draghi said that the ECB would do 'whatever it takes' to save the Euro, and somewhat reluctantly agreed to act as lender of last resort (Hall 2012, p. 364).
Given the greater fiscal vulnerability of national governments within the eurozone and their dependence on ECB intervention to borrow at sustainable rates, the bargaining position of EU supranational institutions became much stronger in the conditionality exchange with member states, allowing the EU to impose a number of conditions in exchange for access to fiscal relief. In Italy, for example, the ECBbecause it is the only actor that could credibly prevent a liquidity crisiswas able to impose explicit or implicit conditions for domestic reforms to the countries from which it bought bonds (Sacchi 2015). In countries which asked for a bailout (such as Greece, Portugal and Ireland), the payment of the various instalments were conditional upon compliance with a precise programme of reforms agreed on with the troika (ECB, European Commission and International Monetary Fund). In Spain, the Memorandum of Understanding (MoU) only covered the financial sector, but the European Commission started a parallel programme of strict monitoring and surveillance of fiscal consolidation measures and structural reformslinking compliance with the programme to the disbursement of the loan. 2 Finally, in all cases discipline was ensured by the market, which 'punished' the governments that deviated from policy prescriptions. Hence, policy activism at the domestic level was a way for governments to signal their willingness to carry out reform to financial markets and international financial institutions, and make sure that fiscal relief would endure.
Even though policy prescriptions were imposed through different channels, they shared common assumptions.The first assumption was that business confidence and growth depended on reducing public and private debt. Hence, one common policy prescription was to cut public spending and incentivise banks, companies and individuals to deleverage (Greer 2014). The second assumption was that consumers would be better served if competition was increased, and if public ownership was reduced. Hence, the privatisation of public enterprises, liberalisation of protected professions and reduction of firms' excessive profit ('rents') were prescribed in all countries . Lastly, 'internal devaluation' became an overarching policy goal. This was fostered by a series of reforms allowing for a quicker downward adjustment of wages, such as a loosening of employment protection, the introduction of more flexibility within collective labour agreements, or the dismantling of collective bargaining (Afonso 2019). Under the auspices of the crisis and tight monitoring by international financial institutions, South European governments engaged in a massive programme of reforms in a wide array of domains, most importantly in the labour market.
One way to get a sense of the volume of reforms undertaken in South European countries is to chart the total number of labour market reforms implemented since the crisis, an area for which we actually possess systematic data collected by the European Commission in its Labref database. This indicator is of course imprecise because it does not measure the importance of reforms, but it at least provides a proxy for policy activity. Figure 1 maps these reforms between 2008 and 2013 using data from the Labref database, mapping reforms in nine policy areas and compiled by the European Commission. South European countries (together with Belgium) clearly stand out as the ones with the largest number of reforms, including retrenchment reforms. For the four countries analysed in this volume (Portugal, Italy, Greece, Spain), retrenchment (coded as either a decrease in generosity or protection) accounted for 84 per cent of all reforms in wage-setting, and 65 per cent of those concerning job protection.
Of course, the number of reforms alone does not provide information on the content and impact of the changes that have taken place in this period. While the contributions in this volume document in detail the substance of reform efforts in a wide array of domains, a partial but telling view of the substantial direction of reforms is given by the OECD Employment Protection Legislation (EPL) index, a widely used, but admittedly imperfect, indicator which seeks to provide a quantitative measurement of the degree of restrictions on hiring and firing. As rigidities in the labour market were considered one of the major areas to reform, this is a relevant indicator as it gives a sense of the direction of change. Figure 2 charts change in the EPL index in European OECD countries between 2008 and 2013. While South European countries clearly displayed the highest levels of protection at the beginning of the crisis, they are also the ones that have made the most extensive changes, bringing them on a par with countries like France, Germany or the Netherlands. Portugal and Greece are the countries that have seen the greatest decline in employment protection, in particular for permanent contracts.
As shown by the cursory data presented above, fiscal vulnerability and tight conditionality have led to an unprecedented period of policy activism, which In the 2013-2018 period, eurozone bailouts programmes ended and the economic situation improved. Table 1 summarises the period that we study for each country. For the bailed out countries (Portugal, Spain and Greece), we consider the end of the bailout programmes. In Italy, there was a marked decline in pressure after the Mario Draghi 'whatever it takes' speech in July 2012, and that it is why Branco et al. (2019) consider this date as the end of conditionality in Italy.

After conditionality
Governments of Spain, Italy, Portugal and Greece are still externally constrained in the post-conditionality period for four main reasons. First, the countries that were bailed out are still subject to inspection by lenders during the reimbursement period. Second, countries in the eurozone benefiting from the ECB's quantitative  Note: for the bailed-out countries (Portugal, Spain and Greece), we consider the end of the bailout programmes. In Italy, we consider the 'whatever it takes' speech in July 2012. Source: own elaboration easing are still vulnerable to the bank's preferences: the mechanisms of 'implicit conditionality'that is 'conditionality based on an implicit understanding of the stakes and sanctions involved, underlain by some measure of power asymmetry' (Sacchi 2018, p. 1)are still at play. Third, EU rules under the Stability and Growth Pact were reinforced in the aftermath of the crisis to extend EU powers over domestic budgets. Sanctions for noncompliance are more encompassing and easier to pass (Bauer & Becker 2014). In November 2018, for instance, the European Commission warned the Italian government led by the Five Star Movement and the Lega that an excessive debt procedure was warranted, judging that its projected budget would exceed the three per cent deficit limit provided for by the Stability and Growth Pact (Juncker 2018). This led to a standoff which remains unresolved at the time of writing in July 2019. A fourth pressure is that there are still many good reasons for governments to be concerned about their credibility at the international level. When dealing with their peers and EU institutions, credibility is important to gain leeway to shape both national policies and also future EU rules (Sacchi 2018). And, of course, credibility also matters for markets and rating agencies (Tavares 2004;Matthijs & McNamara 2015). These factors would lead governments to persist in the course initiated during the crisis, and behave 'responsibly' in the sense outlined by Mair (2009).
However, the external constraints after a bailout or crisis are obviously weaker than before. Investors might look less closely at policy developments in a given country, and the IMF does not have the same 'sticks' and 'carrots' after a bailout to make governments comply with its policy prescriptions. Consequently, the available evidence suggests that the IMF's surveillance activities are not very effective in countries where the IMF cannot withhold payments (Lombardi & Woods 2008). In the European Union, monitoring agents such as the European Commission have been more hesitant to apply sanctions to governments that fail to comply with EU rules since the crisis. On the one hand, the European Commission understands the economic and political consequences of a loose application of the rules (Schmidt 2016). On the other, the Commission is cautious about going after member states for fear of damaging its own credibility if these proceedings fail (Grabbe 2014). Finally, as times goes by and uncertainty decreases, the efficiency of austerity andto a lesser extentof structural reforms has been questioned, challenging the discourse of non-alternatives and necessity that had prevailed at the peak of the crisis (Bailey & Parks 2018).
This relaxing of conditionality begs the question of whether, since 2014, governments have been able or willing to reverse the policies adopted under conditionalityand hence to what extent they have managed to balance the need to be responsive and responsible. This tension has pervaded European politics for decades, but the financial crisis and the wave of austerity reforms brought about during this period put it front and centre (Mair 2011).

Reversals in Spain, Portugal, Italy and Greece
So what happened in Spain, Portugal, Italy and Greece after conditionality? In Spain, the centre-right PP (Partido Popular -Popular Party) majority government led by Mariano Rajoy (2011)(2012)(2013)(2014)(2015)(2016) in power at the end of the macroeconomic adjustment programme, reversed some of the measures as soon as the programme ended, often to the benefit of public servants. These reversals were generally adopted a few months before an election, and were always partial and conditional to additional revenues so as to be budgetary neutral. The courts also reversed a series of bills when the programme terminatedsuch as some measures on collective bargaining or the calculation of pay for unfair dismissals (Branco et al. 2019). After the PP lost its majority in June 2016, the second Rajoy government (2016-2018) had to negotiate with nationalist parties, trading reversals (the re-introduction of budgetary transfer to the regions, for example) in exchange for support for its state budget. On its own will, however, the second Rajoy government also passed a few measures that further (partially) reversed what had been done under conditionality (such as the freezing of the minimum wage and civil service salary cuts).
In May 2018 a motion of no confidence was presented by the PSOE against the government of Mariano Rajoy. The ensuing vote resulted in the downfall of the PP government and the election of Pedro Sánchez as prime minister of a new socialist executive. Despite promises during the campaign to reverse what had been done under conditionality, few reversals followedpartly because it was difficult to find majorities to pass budgets and legislation. Examples of reversals include the abolition of thehardly implemented but very symbolicrestriction of health care provided to immigrants, the re-introduction of social security contributions for nonprofessional carers and the removal of the additional tax on electricity (thus lowering the electricity bill of families by two per cent and of industry by up to 40 per cent). Like its predecessor, the government had to trade some reversals in exchange for parliamentary support (for example, the delegation of the competencies to regulate civil servants' working hours back to the regions and the postponement of the pension reforms). However, in Spain today most retrenchment measures and structural reforms remain. Spending cuts in education and pensions, for example, have not been reversed; and the new labour code has been left untouched by governments. In fact, Moury, Cardoso and Gago (2019) calculate that only 37 per cent of the major reforms have been reversed by the Spanish governments.
When it comes to the evaluation of the reforms passed under conditionality, it must be finally noted that the picture is not as rosy in Spain as sometimes depicted. Perez and Matsaganis (2019) show that while exports and the economy grew in the aftermath of reforms, net job creation during the recovery (2013-2017) compensated for only half of the job destruction during the crisis. Worse, the share of part time employment has increased, with 62 per cent of part-time workers actually preferring to work full time.
Portugal has witnessed more reversals than its neighbour:  found that more than half (55 per cent) of the measures have been reversed. As in Spain, the centre-right government of Passos Coelho (2011)(2012)(2013)(2014)(2015) reversed some reforms as soon as the adjustment programme ended, e.g. the freezing of the minimum wage and the cuts in (the lowest) public servants' salaries. Another similarity with Spain is the activism in the post-conditionality period of the Constitutional Court, which reversed two measures permanently decreasing pensions (Branco et al. 2019).
The large majority of reversals, however, were undertaken by the Costa governmenta socialist minority government that had the explicit support of radical left parties (2015-2019). For example, during the socialists' first year in power, the pre-bailout public sector salaries, working hours and promotions were re-established, the privatisation of urban transport companies and the national airline were reversed, four public holidays were reintroduced, the minimum income was increased and pensions, child and unemployment benefits were extended (Lisi 2016;. During the legislature, important new reversals followed, such as the ending of the freeze on public sector recruitment and the abolition of an extraordinary tax on the highest pensions. All these reversals had been part of the programme that the government and its supporting parties had agreed upon. Remarkably, these reversals were all compatible with low budget deficits because of the extra revenues triggered by the decline in interest rates and new taxes, but also thanks to cuts in investment and in the financing of welfare state services (Fernandes, Magalhães & Santana-Pereira 2018).
As in Spain, post-conditionality economic indicators are mixed (Perez & Matsaganis 2019). While exports have grown more than the EU average in the post-programme period and although the poverty rate is below its 2008 level, cumulative growth in the 2013-2016 was only half the Spanish level. Another similarity with Portugal's neighbour is that new job creation compensated for only half of the job losses during the crisis; and part time employmentwhich was almost nonexistent in 2008reached nine per cent in 2017 (one third of which is involuntary). Despite this relative lack of success of the internal devaluation strategies, most of the labour markets reforms were kept, with the exception of measures on the extension of collective agreements (that was requested by employers) and on extra working hours in the private sectorwhich had the agreement of employers and the smaller trade unions (Afonso & Bulfone 2019).
In Italy, successive governments led by Enrico Letta, Matteo Renzi and Paolo Gentiloni continued on the austerity course, despite a clear attempt to balance austerity reforms with a form of social compensation, notably to reduce dualisation (Picot & Tassinari 2017). This was clearly one of the objectives of the Jobs Act, an important reform of the labour market and one of Matteo Renzi's flagship projects. More generally, policy reforms pursued in Italy in the aftermath of the crisis proved less radical than in Spain, Greece or Portugal. An example of this is the internal devaluation measurement, as shown by Perez and Matsaganis (2019). They calculate, for example, that real compensation of employees per hour worked has fallen in Italy much less than in bailed out countries since 2013. Similarly, unit labour costs have continued to rise in Italy since the pre-crisis, in line with the Euro Area, but have decreased in Greece, Portugal and Spain. This did not prevent the new government formed by the Movimento Cinque Stelle and the Lega, which took power in 2018, to actively decry the reforms passed under austerity and announce reversals. Hence, as Afonso (2019) show in this volume, Matteo Salvini promised to 'shred' the pension reform passed during the Monti government, while Luigi di Maio, leader of the Cinque Stelle Movement, promised to 'get rid of the Jobs Act'. The extent to which these declarations will be followed by radical policy reversals is unclear however.
For Greece, reversals were not considered systematically by the contributions in this volume as the bailout only came to a close in August 2018. Nevertheless, Perez and Matsaganis (2019) analyse the effects of internal devaluation. By any account, the changes in socio-economic governance in Greece were the most farreaching of the countries in the sample analysed here. The adjustment programme put in place in Greece basically marked the complete dismantling of the collective bargaining system, with a dramatic collapse in collective bargaining coverage and deep spending cuts, coupled with a mass increase in unemployment. Perez and Matsaganis notably show the dramatic rise in poverty rates in Greece, reaching an astounding 44 per cent in 2013. Those reforms, however, did not produce the desired effects: in Greece, in the last 4-5 years, although labour cost has decreased, economic recovery is non-existent, export performance is disappointing, the newly created jobs did not make up for the looses and anchored poverty has risen further (to 46.3 per cent in 2017).
What do we learn in this volume that we did not know before?
The contributions in this volume draw a broad and comprehensive picture of the number and extent of policy reversals in the aftermath of the eurozone crisis, and outline the mechanisms through which they occur. The main contributions can be summarised in the light of the three objectives outlined at the start of this article, namely description, explanation and evaluation.
First, given the two conflicting pressures between responsibility and responsiveness, we expected contradictory (or compensatory) movements; towards change and towards persistence. This is confirmed by the articles in this volume, which show that reversals are a common occurrence in the aftermath of the crisis; but that governments are selective in their choices. The contributions of Branco et al. (2019), Afonso and Bulfone (2019) and  show that reversals are observed across all policy areas, by both the left and the right, and in all countries. For example, Branco et al.
(2019) count a total of 17 reversals of structural reforms: five in Spain, and six each in Portugal and Italy.  calculate that 46 per cent of the major reforms introduced in Spain and Portugal during the crisis have been reversed. However, there are substantial differences across countries and different types of reforms. First, there were almost twice as many reversals in Portugal as in Spain. Second, more than 80 per cent of the structural reforms were kept, while all reforms designed to be temporary were predictably reversed. Insofar as a number of these latter reforms were freezese.g in pensions and salariesthey still had long-lasting budgetary effects as increases were lower than they would otherwise have been. This means that, even in cases like Portugal where governments seem to have reversed everything that had been done previously, bailouts have had a durable impact on the social and economic institutions.
Turning to the explanatory side, the contributions explain the political mechanisms which enable or hinder reversals. In line with our theoretical framework, contributions in this volume show that the distribution of costs and benefits matters to explain policy reversals. Moury, Cardoso and Gago (2019), Branco et al. (2019) and Afonso and Bulfone (2019) describe the costs associated with reversals in terms of lack of budgetary revenues, damaged credibility or the possibility of sanctions by the EC. They show that, given these costs, governments have a choice to make and are more likely to reverse reforms when there is a strong possibility they will bring an electoral payoff. Indeed, Moury, Cardoso and Gago (2019) demonstrate that reversals become more likely when the benefits of these reversals are concentrated and visible, in contrast to cases where the benefits are diffuse. Similarly, Afonso and Bulfone (2019) show that the nature and occurence of reversals is also shaped by the social constituencies that constitute the support base of parties in power. For instance, pensions are a particularly important issue for the Portuguese Communist Party and the Italian Lega because of the sizeable share of older voters within their electorate; in contrast, labour regulations are more salient to the Portuguese Bloco and the 5-star Movement, which have a younger electoral clientele. Related to this, the research by Moury, Cardoso and Gago (2019) additionally stresses the power of vested interests: after conditionality, the latter lobby to regain their rents and manage to do so when these are not very visible.
Second, left-wing governments are more likely to reverse policies given the pro-market policy bias of the reforms passed during the crisis (Branco et al. 2019;. However, this does not mean that rightwing governments do not reverse austerity reforms when it serves political purposes or protects their own social constituencies, as the examples in the articles make clear. Third, contributions of this volume clearly illustrate that veto players matter.  and Branco et al. (2019) show how non-governing parties extracted reversals as concessions for their support for the budgets proposed by the Rajoy, Sánchez and Costa minority governments. Similarly, Branco et al. (2019) demonstrate that the Judicial Courts and the Regions were both veto players, and as active as governments in reversing structural reforms.
The evaluative focus of the volumean assessment of the effects of reforms and reversals using a number of economic and social indicatorsis most directly tackled by the contribution from Perez and Matsaganis. They show that, as expected, the reforms produce internal devaluation in labour costs (in all but Italy). Yet they also show that the countries where the strategy of internal devaluation was pursued most aggressively were also those where the figures for job creation and recovery were the least impressive. While there are obvious endogenous mechanisms at playcountries with the weakest economic fundamentals were also those where reforms were the most brutalthe evidence they present makes for a critical assessment of the social effects of the reforms carried out during the crisis. In particular, they show that job creation has not offset previous losses, and that in-work-poverty remains close to its crisis peak in Spain and Italy and has risen even further in Greece. On the other hand, Italy, where labour market reforms took longest to be implemented and the pace of fiscal consolidation has also been more limitedlooks better than is often assumed.
The paper by  demonstrates that, as a result of most governments' commitment to European rules, reversals were often compensated for by other taxes or spending cuts that were less visible. For example, during the Costa Government, investment was reduced and overall spending in welfare state services in the first years remained as low as during the crisis, despite the increase in civil servants' salaries and reduction of working hours. This has triggered accusations of 'electioneering' by many national commentators.
Lastly, we do not find any systematic evidence for social learning in this volume. However, the fact that the PSOE government led by Pedro Sánchez declared several times its intention to follow the path of the Portuguese government shows the importance of learning or informational spill-over. In a similar vein, the fact that the European Commission softened its criticism of reversals in Portugal after seeing that they did not prevent growth or employment also provides evidence of a learning process .

Conclusion
In the introduction of this article, we argued that the politics of policy reversals in the aftermath of the eurozone crisis can be understood in light of the tensions between responsiveness and responsibility faced by governments in Europe and beyond. As already discussed by a number of authors (Armingeon & Baccaro 2012;Mair 2011;Streeck 2011), governments in today's economy have to cater to different audiences with clashing demands. On the one hand, the European Union and the (financial) markets on which governments depend for borrowing demand sound finances, wage restraint and policies in line with higher returns. On the other hand, domestic democratic constituencies demand protection and safety. This volume on policy reversals has addressed when and why governments prioritise the former or the latter, as well as the consequences of these choices.
Contributions in this volume show that, in their attempts to reconcile the tensions between these two objectives, governments targeted reversals of the most salient policies and those which concentrated benefits on their core constituencies. However, most governments managed to do so while respecting European rules. At times, this was done at the cost of policy goals that could be postponed but with negative long-term consequences. For instance, the Portuguese government has been able to reduce its deficits while restoring benefits that had been cut during the crisis, but this was done at the cost of significant cuts in future investments. The recent Italian case shows that governments that wanted to ignore European rules had to backtrack due to the pressure from markets (as in the case of the current Italian government when its budget projections were not considered realistic by EU institutions). Hence, juggling between responsibility and responsiveness, between continuity and reversal, is always a delicate balancing act.
The assessment produced in this volume is obviously subject to a number of limitations. One issue is timing: while we have a reasonable timespan to assess the policy course of the coalition that accessed power in Portugal in 2015, Italy and Spain only had a substantial alternation in power in 2018. This limits our ability to assess the role of alternation in power in policy reversals. While we provide the most up-to-date assessment of recent reforms in these countries, a longer time horizon will be necessary to truly assess the long-term institutional impact of the crisis on policies, institutions and outcomes in Southern Europe, and to determine whether the crisis was really a critical juncture in these countries' models of governance and political economy.

Disclosure statement
No potential conflict of interest was reported by the authors.