1. Introduction

1In the last few decades, the governance of Italian capitalism has changed profoundly. The past thirty years (1990-2020) have seen many changes in economic and social policies, with processes of reform, deregulation and privatization. The Keynesian postwar model, with its emphasis on state intervention, the welfare state and more consensus-based approaches to decision-making, has been eroded. The European policies of participation in the Single Market and the single currency have undoubtedly contributed to a common and converging set of policy responses and institutional changes to basic political and economic structures (Della Sala, 2004). The member states’ forms of economic governance have thus opened up to market-based regulation and coordination. Nevertheless, Italy can still be numbered among the countries with a predominately mixed model of capitalism, where the state has a rather sizeable role in the economy (Hall, Soskice, 2001; Amable, 2003), and from the standpoint of social policies the family shoulders much of the weight of looking after citizens’ needs (Esping Andersen, 1990; Ferrera, 1996).

2Against this backdrop, the present study seeks to determine if and how Italians’ attitudes towards the role of the state and the market have changed. Specifically, it will investigate two sets of polar opposite attitudes: individualism versus collectivism, and free market liberalism versus statism. The study considers two opposing macro hypotheses. The first is that Italian citizens adopt attitudes that are in line with policy trends, while the second is that there is a tendency to develop a compensatory reaction. In addition, the study seeks to understand whether and to what extent belonging to a given social class influences these attitudes. It is posited that the connection between class and attitudes can take different shapes depending on which of the macro hypotheses holds true.

3The rationale for the study lies in the need to understand the dynamics of resource redistribution benefiting social groups who support governing coalitions. Redistribution is a significant part of the political process which is accomplished by setting limits on the range of economic action available to private actors and creating incentives and disincentives for individual behaviors. However, as policy feedback scholars argue, the existence and persistence of certain kinds of public policy also (re)shapes the social norms with which citizens are expected to comply (Van Oorschot, Meuleman, 2014). Hence the interest in determining whether and to what extent citizens’ expressed attitudes reflect the norms implicit in policies, or are a reaction to them. This paper thus focuses on the role that class continues to have on Italians’ attitudes towards state intervention. As belonging to a given social class is an indicator of the advantages and disadvantages of a free market (and in particular a free job market), it is assumed to influence personal interest in such questions. Accordingly, and despite assertions to the contrary, there are valid reasons for investigating the effect of class on attitudes towards the role of the state and the market. The study’s findings show that this effect has by no means disappeared over the thirty years we consider, although it has weakened in certain specific respects.

2. Classes and attitudes towards the role of the state and the market

4Studies of social inequalities are primarily concerned with the material resources to which the members of a given social class have access, both as an intrinsic characteristic and as a result of belonging to that class. For example, home ownership is a characteristic of a middle class life style, while at the same time being middle class increases the likelihood of home ownership (Filandri, 2015). Despite a certain inevitable degree of heterogeneity, we consider that material characteristics on the one hand shape class, and on the other reproduce the system of social stratification.

6Most modern societies are polarized around issues such as the role of the state and the market (Curtis, Andersen, 2015; Carriero, 2016). The lower classes tend to support government intervention in matters of redistribution and social security – a policy traditionally, but not exclusively, promoted by the left. By contrast, the upper classes are likely to favor market expansion and reducing redistribution – policies traditionally, but not exclusively supported by the right. People are thus “class aware”; i.e., they recognize how their class position and related inequalities affect their life chances (Curtis, Andersen, 2015). In other words, classes differ in their economic interests, and the attitudes of the members of a given class can be said to reflect these interests. This would not only explain the association – weakened, but still a reality – between class and left-wing voting, but also the intolerance towards outgroups such as immigrants, whom the working class sees as a threat to their economic wellbeing (Kalmijn, Kraaykamp, 2007). In addition to operating through the mechanism of personal interest, class influences attitudes because social interaction takes place most often within classes rather than between them (Kalmijn, 1994; Chan, Goldthorpe, 2004). As mentioned earlier, this implies that certain attitudes or lifestyles can be developed and reinforced among members of a class through social influence. Lastly, the desire for social distinction on the part of members of a given class may lead to attitude differences because people seek to distinguish themselves from other classes by expressing different values (Bourdieu, 1979; Lamont, 1992).

7These mechanisms (personal interest, social influence, distinction) can operate in the relationship between class and attitudes in differing ways and in varying intensities, depending on the sphere in which the attitudes considered arise. As regards the role of the state and the market, personal interest would appear to be the most obvious interest at play, and indeed much of the earlier literature (see for example Brooks, Svallfors, 2010) has often used class in this way, as an indicator that captures both income and exposure to economic risk. For the attitudes in question here, however, the other two mechanisms cannot be ruled out, given that the economic aspects (of more state, less market, for example, or more or less redistribution) are inextricably linked to moral judgements (about deservingness, reciprocity, confidence in institutions) that class can bring to light. This is what Svallfors (2006) calls the “moral economy of class”. There is a clear reference here to Thompson (1971), who presents the moral economy as being at the basis of the critical concerns for the workings of market-oriented capitalism, which treats labor as a fictitious commodity, and the influence this has on the wellbeing of individuals and society in general. Attitudes thus reflect an internalization, more or less profound, of the norm of reciprocity.

8The degree of stratification in socio-political attitudes can vary, as there may be a significant lack of uniformity among members of the same class. Even people of similar occupational status differ, and often considerably, in educational attainment, political views (Brooks, Manza, 2006; Svallfors, 2006; Manza, Crowley, 2018) and work logic (Oesch, 2006). Take, for example, the cleavage in attitudes towards the welfare state that can be observed within the higher classes. On the one hand, managers and business owners tend to support welfare spending cuts, while on the other, sociocultural professionals would like the welfare state to be expanded, or at least not curtailed (Gonthier, 2019; Carriero 2021).

3. The context: Italy in the last thirty years

9Before turning to an empirical investigation of the relationship between class and attitudes to the role of the state and the market in Italy, a few words are in order concerning the context in which these attitudes took shape. The contextual aspects we will discuss primarily involve the social and economic policies adopted in Italy in this period. Two theoretical mechanisms from the literature will help explain the importance of presenting these aspects. The first is the so-called policy feedback effect, whereby the existence and persistence of certain kinds of public policy also (re)shapes the social norms with which citizens are expected to comply (Pierson, 1993; Mettler, Soss, 2004; Van Oorschot, Meuleman, 2014). The second, or “thermostat” effect is a similar but opposite mechanism (Wlezien, 1995; Erikson et al., 2002). In this case, public opinion is seen as a thermostat that sends a signal when policy is becoming “too hot” or “too cold” for the public taste. If, for example, policies have moved very far towards liberalization, the more moderate members of the public will tend to demand a higher level of government intervention. Conversely, when the state takes a heavy-handed approach in economic matters, public opinion will react by demanding more laissez-faire. Sociologically, it is thus important to determine to what extent the attitudes expressed by citizens reflect the norms embodied in policies, or tend to oppose them.

11But where did the push for such reforms come from? To understand the changes that have taken place in Italy, we must look at what happened earlier in other countries. The answer to our question lies in the rise and spread of “neoliberal” economic (and social) ideas between the late Seventies and early Eighties in the UK and US. The keywords of this ideology are free market, competition, merit, individual, small government and so forth. In Italy, the neoliberal ideas underlying some of the reforms we have mentioned appeared on the political agenda largely during the Nineties (Regonini, 1993; Rangone, Solari, 2012). It was then that the first privatizations of state-owned enterprises got under way along with labor market reforms. According to Ferrera (2013), however, neoliberalism in Europe began to ebb early in this century, replaced or combined with other ideological visions that see the state’s role and investing in the welfare state in a more favorable light, without abandoning the pursuit of effectiveness, efficiency and fiscal stability.

12The second contextual factor to be taken into account is the political parties’ stance regarding the role of the state and the market. Where the parties stand on these issues is important inasmuch as they (like other intermediaries such as unions and employers’ associations) act as “institutional translators” (Edlund, 2007) in framing a political question in the public debate. Parties can thus influence citizens’ attitudes towards the state-vs-market question both by directing more or less attention to it than others (immigration, for instance) and by taking explicit positions for or against policies that put greater reliance on the market as a mechanism for regulating the economy and public services.

4. Research questions and hypotheses

16The second research question concerns the influence of social class. To define social class concisely but meaningfully, we adopt a four-category schema which is widely used in Italian empirical research: upper class, employed middle class, self-employed middle class, and lower class. It is known that the middle class can be dichotomized on the basis of employment status: on the one hand there are non-manual workers employed in public or private knowledge and service industries. On the other, there are self-employed workers and owners of small and medium businesses (Ranci Ortigosa, 2012; Bagnasco, 2016). As small businesses represent a sizable slice of the Italian economy, their workers and owners are the core of the self-employed middle class. However, they are a far less homogeneous group than the employed middle class in terms of income, household wealth, educational attainment and living conditions (Negri, Filandri, 2010; Dagnes et al., 2018). The four-class schema, based on an aggregation of the European Socio-economic Classification or ESeC (see the following section for further details) combines the advantage of parsimony, avoiding the fragmentation that prevents a clear understanding of ongoing processes, with a precise identification of the main social groups in Italian society.

17Given the personal interest mechanism, social class can be associated with different attitudes, with higher classes being more in favor of the market than lower classes. By contrast with post-modernist arguments, we thus expect that class will still be a significant factor in shaping attitudes (H2).

18The contextual factors outlined earlier can also attenuate class differences. If there have been no fundamental shifts in social and economic policies during the thirty years considered in the study, and no sharp contrasts in the major parties’ positions, then the effect of class, though continuing to operate through the personal interest mechanism, could become less intense over time. Both the policy feedback mechanism and the activity of Italy’s “institutional translators” contribute to smoothing out large differences between classes in questions concerning the role of the state and the market (H2a). The thermostat effect, whereby individuals faced with macroeconomic hardship or more neoliberal polices tend to call for more government action, can also attenuate class cleavages. However, it should be borne in mind that the two alternative mechanisms – policy feedback and the thermostat effect – can operate in different classes, the first in the better-off classes, and the second among the less fortunate. In this latter case, the classes would move farther apart, reinforcing social stratification (H2b).

19The third research question concerns changes in within-class differences over time. Here, we can expect the personal interest mechanism to increase uniformity within groups. For example, all the members of the classes that are more exposed to economic risk should lean more towards a collectivist-statist attitude as a means of protecting themselves from this risk. Over time, this could lead to less divergent attitudes within groups (H3a), strengthening class identity. However, if individuals change their attitudes on the basis of the thermostat effect as well as personal interest, and if the two micro and macro mechanisms interact in the same class, we might see an increase in within-class differences (H3b). This could also mean that social classes are less able to represent the social structure. In other words, over and above the mean effect, greater variability would lead us to conclude that there is a tendency towards individualization in the classes, i.e., that the members of a class are less and less likely to identify with it.

5. Data and method

20The study is based on data from four Italian waves of the EVS: 1990, 1999, 2009, 2018. Five items from the same battery were repeatedly employed to tap citizens’ attitude to the role of the state and the market in economic and social matters. Items were on self-anchoring 10-point scales whose end points were statements of opposite semantic orientation, representing the individualist-liberal position and the collectivist-statist position (table 1).

Table 1. Items (1 to 10 scale) tapping attitudes to the role of the state and the market in the EVS surveys

Item	Individualist-liberal position (1)	Collectivist-statist position (10)
State responsibility	Individuals should take more responsibility for providing for themselves.	The state should take more responsibility to ensure that everyone is provided for.
Unemployment	People who are unemployed should have to take any job available or lose their unemployment benefits.	People who are unemployed should have the right to refuse a job they do not want.
Competition	Competition is good.	Competition is harmful.
Redistribution	There should be greater incentives for individual effort.	Incomes should be made more equal.
Government ownership	Private ownership of business and industry should be increased.	Government ownership of business and industry should be increased.

21The five items represent five dependent variables, treated as cardinal variables and analyzed by distinct linear regression. To permit comparison with other investigations (see e.g. Maraffi, 2020), we created an index (average score) that can serve as a synthetic measure of the individual attitude to government intervention, although Cronbach’s alpha is poor (0,47).

• Upper class: includes ESeC classes 1 and 2 (variable present in the dataset for the 2009 and 2018 waves) correspondingly roughly to the categories of employers/managers of establishments with 10 or more employees and professional workers, whether employed or self-employed, used in the 1990 and 1999 waves.

• Employed middle class: ESeC classes 3 and 6 in the 2009 and 2018 waves; middle and junior level non-manual office workers, foremen and supervisors, and members of armed forces in the 1990 and 1999 waves.

• Self-employed middle class: ESeC classes 4 and 5 in the 2009 and 2018 waves; employers/managers of establishments with less than 10 employees, self-employed workers, farmers on own account in the 1990 and 1999 waves.

• Lower class: ESeC classes 7, 8 and 9 in the 2009 and 2018; manual workers with various levels of skill in the 1990 and 1999 waves.

Table 2. Sample characteristics, 1990, 1999, 2009, 2018

	1990			1999				2009				2018
	N.	Mean	SD		N	Mean	SD		N.	Mean	SD		N.	Mean	SD
Attitudes
State responsibility	1260	5.42	3.00		1289	5.48	2.68		1150	5.70	2.63		1772	5.42	2.72
Unemployment	1269	3.53	2.69		1278	3.35	2.42		1136	3.44	2.46		1761	3.36	2.48
Competition	1233	4.11	2.69		1267	4.11	2.49		1124	4.40	2.60		1763	4.11	2.38
Redistribution	1262	5.11	2.94		1277	4.83	2.72		1137	4.96	2.79		1770	5.43	2.80
Government ownership	1191	4.18	2.45		1207	4.00	2.22		1035	4.65	2.38		1648	5.14	2.47
Index	1293	4.51	1.65		1301	4.38	1.51		1164	4.65	1.52		1792	4.69	1.40
Social class
Upper class	1302	0.28	0.45		1321	0.36	0.48		1173	0.29	0.45		1794	0.23	0.42
Employed middle class	1302	0.24	0.43		1321	0.21	0.40		1173	0.22	0.41		1794	0.17	0.38
Self-employed middle class	1302	0.08	0.27		1321	0.08	0.28		1173	0.16	0.37		1794	0.17	0.38
Lower class	1302	0.40	0.49		1321	0.35	0.48		1173	0.33	0.47		1794	0.43	0.50
Age group
Up to 34	1302	0.29	0.45		1321	0.30	0.46		1173	0.19	0.40		1794	0.16	0.36
35-44	1302	0.20	0.40		1321	0.20	0.40		1173	0.21	0.40		1794	0.19	0.39
45-54	1302	0.19	0.39		1321	0.17	0.38		1173	0.19	0.39		1794	0.22	0.41
55-64	1302	0.14	0.34		1321	0.16	0.37		1173	0.16	0.36		1794	0.17	0.38
65 and over	1302	0.19	0.39		1321	0.17	0.37		1173	0.25	0.44		1794	0.27	0.44
Gender
Female	1302	0.41	0.49		1321	0.42	0.49		1173	0.47	0.50		1794	0.41	0.49
Male	1302	0.59	0.49		1321	0.58	0.49		1173	0.53	0.50		1.794	0.59	0.49
Area of residence
Northwest	1301	0.31	0.46		1321	0.30	0.46		1173	0.29	0.45		1794	0.29	0.45
Northeast	1301	0.21	0.41		1321	0.22	0.41		1173	0.21	0.41		1794	0.21	0.41
Center	1301	0.20	0.40		1321	0.19	0.40		1173	0.18	0.39		1794	0.22	0.41
South	1301	0.18	0.39		1321	0.19	0.40		1173	0.21	0.41		1794	0.20	0.40
Islands	1301	0.10	0.30		1321	0.10	0.29		1173	0.11	0.31		1794	0.09	0.28

Note: Statistics calculated for cases with valid values for social class.

24In addition to the graphs, we also present the kappa index (Hout et al., 1995), which expresses the overall extent of class differences in attitudes. Kappa is the standard deviation of the class effects obtained from the regression or, equivalently, of the estimated class means. As kappa increases, then, so does the extent of the differences between all classes, and not just the gap between the upper and lower classes.

6. Results

26By introducing the social class variable, we can determine whether and how social stratification is reflected in attitudes for or against state intervention. Starting from Figure 2, we see that the lower class’s average estimated attitude differs significantly from those of the other three classes. In all waves, the most disadvantaged class tends to be less oriented towards individualistic-liberal positions. Looking at the variations in detail, however, we see that the lower class’s average first drops, then rises, and finally returns in 2018 to the 1990 level. By contrast, all other classes rise slowly towards the collectivist-socialist pole. The end result is that the variability in the differences between all classes (kappa) is higher in 1990 and 2008, and lower in 1999 and 2018 (table 3). The upper-lower class cleavage in the four waves is 0.7, 0.4, 0.5 and 0.4. Thus, though the effect of class persists as posited by (H2), differences between groups are also less pronounced. All classes call for more government action, though there is still a higher proportion of free market liberals among the three higher classes. The hypothesis that the effect of social class is maintained over time (H2a) but with a tendency towards convergence on the collectivist-statist pole appears to be confirmed. In other words, a thermostat effect operates for all classes with the exception of the lowest.

Table 3. Kappa index for each attitude and the additive index, 1990-2018

Year	Index	State responsibility	Unemployment	Competition	Redistribution	Government ownership
1990	0.31	0.47	0.27	0.41	0.53	0.30
1999	0.17	0.26	0.26	0.15	0.40	0.31
2009	0.28	0.41	0.16	0.40	0.43	0.26
2018	0.17	0.26	0.26	0.22	0.32	0.12

27Analyzing the relationship between class and the additive index enables us to determine whether there has been an overall trend towards a specific pole on the part of Italy’s classes. However, if we turn to the difference in the individual items, we see that attitudes vary aspect by aspect. Starting from the aspects that remain more stable over time, figure 3 shows that social class plays a major part in determining the attitude towards the state’s responsibility for providing for each of its citizens. In particular, though variations can be seen, they show that the lower class always calls for more government action than the upper class. The initial cleavage shown in 1990 tends to narrow in 1999, widen again in 2009 and narrow once more in 2018. This is confirmed by the kappa index, which is a summary measure of the overall extent of class differences: the index shows the highest values in 1990 and 2008, and the lowest in 1999 and 2018 (table 3). The pattern may be explained by the fact that the 2009 wave took place in the midst of a severe economic crisis, when lower class individuals feel a stronger need for government protection. It is also interesting to note that the lower class is similar to the employed middle class in the oldest survey and in the latest (in 1990 the difference between the two classes is still statistically significant).

28Similar considerations apply to attitudes towards competition (fig. 4). Here again, the lower class stands apart from the others, particularly by comparison with the upper class (with the exception of 1999 if we look at the confidence intervals). The mean attitudes of the employed middle class are comparable to those of the upper class, while the self-employed middle class’s attitude differs from that of the other classes only in 1990 (and from the lower class in 2008). As the kappa index shows, overall class cleavage fluctuates over time: wider in 1990 and 2008, narrower in 1999 and 2018.

29Attitudes about whether competition is good or harmful, point estimates and 90% confidence intervals by social class and year, Italy 1990-2018

30A class effect is also apparent in attitudes to the conditionality of unemployment benefits (fig. 5), though its operation is opposite to those we have seen so far. Members of the lower class tend to be more in favor of conditionality, as is the self-employed middle class. This is most evident in 1999 and in 2018. By contrast, upper class and employed middle class individuals are on average more inclined to think that the unemployed should be able to refuse a job without losing their benefits. This reversal of expectations is probably due both to education (between-class differences are smaller when we control for educational level) and to the ambivalence of the item’s phrasing (see above). The kappa values for this item indicate that the overall class cleavage is roughly constant over time.

31If we look at how class is associated with the last two attitudes regarding income redistribution and increasing government ownership (fig. 6 and 7), we see different situations although both have shifted on average towards the collectivist-statist pole. In the first case, regarding the need to make incomes more equal, there is a clear and constant gap between the lower class on one hand and the upper and employed middle classes on the other throughout the period in question, though this gap narrows in the last survey wave. Over time, the self-employed middle class and the lower class move in parallel but remain widely separated, returning to roughly the same positions occupied at the beginning in 1990. Interestingly, in 2018 members of the three highest classes have virtually the same view of income equalization measures. Overall, the differences between the four classes have lessened over time, and indeed, the kappa index went from 0.53 to 0.32 between the first and last surveys.

32Attitudes about private or government ownership of business and indus-try, point estimates and 90% confidence intervals by social class and year, Italy 1990-2018

33The attitude towards government ownership was the only item where social class no longer tends to have an effect of any kind. In all waves with the exception of 1990, the differences between classes are never significant. In this case, we see all classes moving to converge on the midpoint of the scale, towards the statist pole.

34Lastly, we come to within-class differences in attitudes, as an indicator of internal cohesion and group identity (fig. 8). In examining the standard deviations for all items, we see a clear reduction in the dispersion around mean attitudes between 1990 and 2018 only for the employed middle class and the lower class. These two groups’ greater within-class similarity is consistent with the idea that the members of each of these classes share a more similar occupational status which reinforces their class identity. Indeed, as the Italian literature (Bagnasco, 2016; Ranci Ortigosa, 2012) has shown, the upper class and the self-employed middle class are the least uniform groups. It would thus seem that hypothesis H3a is only partially confirmed. In any case, it cannot be said with confidence that processes of individualization are under way for the upper and self-employed middle classes, given that the dispersion around mean attitudes has increased over time for some items and decreased for others.

7. Conclusions

35The study investigated the relationship between social class and attitudes towards the state and the market in Italy from 1990 to 2018. The findings show that the influence of class has continued to be significant over the period, though it has weakened overall as a result of convergence towards collectivist-statist values. Detailed analysis of five different aspects indicated that there are differences in the class effect’s intensity as well as its direction. Members of the higher classes are farthest from the collectivist-statist pole in their attitudes towards the government’s responsibility for providing for all citizens, towards competition, and towards income equalization. For these three aspects, class cleavages have persisted, but are attenuated by a shift-by the higher classes in particular-towards the less individualistic-liberal pole. This trend can be interpreted as reflecting a thermostat effect, i.e., a reaction to the Italian policies of recent decades that have curtailed the role of the state and given freer rein to the market and competition. A similar mechanism may be at work in attitudes towards government ownership of business and industry. In this case, class cleavages have even disappeared, as all groups now look more favorably on government ownership. The same cannot be said of opinions about unemployment benefits. Though no significant differences have emerged over time, the lower class and the self-employed middle class take the hardest line: the unemployed should have to take any available job or lose their unemployment benefits.

36As regards within-class differences as an indicator of the strength of group identity, the findings show that they tend to decrease for all items only in the lower class and, to a slightly lesser extent, in the employed middle class. For the remaining two classes, within-class differences oscillate for some items and are stable for others.

38Some limitations of the study should be acknowledged. First, the battery of EVS items used to operationalize attitudes towards the role of the state and the market has shown that the relationship between class and attitudes changes according to the item under consideration, especially for the item about unemployment benefits. Consequently, proposing a simple additive indicator is not sufficient, useful though it may be in providing a general sense of how attitudes are trending in the population. More thought should be devoted to how the battery could be extended to produce a more homogeneous scale. Second, our analysis of the relationship between class and attitudes is limited by the small sample size, especially in the first survey waves, and by the lack of information which is crucial in this area of investigation, viz., whether respondents work in the public or private sector. Obviously, public and private employees’ attitudes towards the role of the state and the market can differ significantly, even within the same class. Such information could be made readily available in the next EVS waves.