Individualisation and Poverty Over Time: The Italian Case (1985–2011)

Annual percentage growth of ‘Final Consumption Expenditure’ for Italy, from 1985 to 2009 (source: ISTAT).

In this context, by using a cumulative data-set that merges ISTAT surveys, we can study social change over 27 years by considering detailed socio-economic information on 747,204 families. The average size of the samples is around 27,000 cases per year (see Appendix Table A1).

As discussed in the previous sections, levels of expenditure can be used to calculate the percentage of poor families. Different methods are applied to estimate poverty. The most common of them focus on the concept of ‘relative poverty’, defining as poor a family with a level of consumption below 60% of the median equivalised expenditure (the so-called poverty line). Equivalent consumption takes account of the size and composition characteristics of the household. In particular, we adopted the OECD modified scale (Mejer and Siermann 2000) currently used by Eurostat to evaluate poverty and based on Europeans’ incomes (see ECHP/EU-SILC). Thus, the equivalised consumption was calculated by dividing the household's total expenditure by its equivalent size. In the equivalent size, the first adult had weight 1; the second and each subsequent person aged 14 and over had weight 0.5; and each child aged below 14 had weight 0.3.^⁵

On the basis of this definition of poverty in Italy, we obtained the trend reported in Figure 2. The percentages vary from about 16% to 18.5% and are similar and fairly constant during the period considered, in particular, from the late 1990s (Brandolini 2009). An increase is present between 1996 and 1997, but it should be noted that there is a methodological interruption in the ISTAT series. In fact, in 1997, questionnaires were redesigned and some questions were modified (including some expenditure items in the diaries). Also Figure 2 shows small but significant differences between poverty based on income (ECHP/EU-SILC data) and poverty based on consumption, but it is likely that these difference are due to different types of poverty measures, as well known in the literature (Buhmann et al.1988; Cutler and Katz 1992). In particular, it seems that the poverty line calculated on consumption is relatively lower than that calculated on income (Cutler and Katz 1992). Consequently, the number of poor families is smaller. Thus, interpretation of the results will depend upon careful consideration of analysis conducted on a definition of poverty based on household expenditure and not on household income.

Percentages of poor households according to various definitions. Households at-risk-of poverty rate (Source: Eurostat – SILC): http://epp.eurostat.ec.europa.eu/.

Figure 2.

Percentages of poor households according to various definitions. Households at-risk-of poverty rate (Source: Eurostat – SILC): http://epp.eurostat.ec.europa.eu/.

As discussed, the status of poverty may be due to both structural social conditions and life course variables. which we try to grasp by looking at demographic variations in the population. Empirical evidence in Italy shows that elderly persons, large households and one-parent families tend to have greater chances of being in poverty or in vulnerability conditions (Brandolini 2009; Freguja and Pannuzi 2007; Lucchini and Sarti 2005). Thus, in order to answer our research question correctly, we had to consider in our analysis the possible confounding role of the demographic characteristics of the families. In fact, demographic changes may exert effects on poverty independently of socio-economic structures.

The form of the data-set presented a problem concerning the unit of analysis, the household (the only level at which we could estimate poverty with the data available). For this reason, we had to aggregate the individual characteristics at familial level.

We therefore decided the following operationalisation of the demographic familial indicators:

a dummy variable for the gender of the reference person;^⁶
an ordinal variable for the age of the reference person (<35 years, 35–44 years, 45–54 years, 55–64 years and >64 years);
an ordinal variable for the size of household (1 member, 2 members, 3–4 members, more than 4 members).^⁷

Moreover, among the structural conditions, we had to include the territorial cleavage together with education and occupation. In Italy, there is a marked difference between the Northern and Southern regions, whereby the latter are severely disadvantaged in social and economic assets (Biolcati-Rinaldi and Podestà 2008; Morlicchio 2012; Saraceno 2003). Likewise, poverty is more widespread in the regions of Southern Italy than in the North.

In discussing the individualisation of poverty, we said that life events can be considered as contingent situations faced differently by individuals (see the previous section). It was for this reason that we decided to add a control variable, namely the presence of at least one unemployed person within the family. This was operationalised as a dummy variable at familial level. Although unemployment is associated with macro-economic conditions, in this way we would be able to exclude a factor strongly associated with poverty that supporters of the decline of the role of social structures might consider an event linked to idiosyncratic choices in life courses.

The structural indicators that we used were:

a dummy variable for the residential area of the household (northern versus southern regions),^⁸
an ordinal variable for the educational attainment of the reference person (primary, lower secondary, upper secondary or tertiary)^⁹ ;
a categorical variable for the family's social class (manual workers, petty bourgeoisie, white collars and bourgeoisie).^¹⁰

The associations of structural variables like area, educational attainment or social class, with poverty had to be controlled for by confounding variables. It should be noted that one of these control variables was a dummy that captured the interruption in the surveys series questionnaires (it was 0 for the surveys from 1985 to 1996 and 1 for the surveys from 1997 to 2011). We consequently had to implement multivariate models in order to investigate the change over time in the association between structural variables and poverty, controlling for some life course variables. We used hierarchical regression models with families at level one and time at level two, as used by DiPrete and Grusky (1990), in order to estimate the year-to-year fluctuations. Multilevel model building strategies can be either top-down or bottom-up. We adopted a bottom-up approach where the various models were developed incrementally (Hox 2002). In the first model, we considered the area of residence as the main independent variable: since in Italy macro-economic territorial differences are cleavages important in determining social stratification (Morlicchio 2012), we wanted first to evaluate the association between area and poverty across time. In the second model, our focus shifted to the educational level, while in the third model the attention was on the social class. We chose this sequence on theoretical bases because the level of education in terms of the life course influences the occupational group (which defines the social class), and not vice versa.

The following basic equations define the analytical basic model where the structural regressors are treated as ‘random effects’ and all control variables are treated as ‘fixed effects’:

P_{i t} ~ B i n (1, π_{i t})

π_{i t} = \frac{\exp (η_{i t})}{1 + \exp (η_{i t})}

η_{i t} = β_{0 t} + \sum_{j = 1}^{J} β_{j t} x_{j i t} + \sum_{k = 1}^{K} γ_{k} x_{k i}

β_{0 t} = β_{0} + u_{0 t}

β_{j t} = β_{j} + u_{j t} for j = 1, \dots, J

where: i represents households and t the temporal context (the year of the survey); P is the dependent variable, the chance of a household being below the poverty line; J denotes the number of the structural indicators: area of residence, educational level and social class (J = 1 for model 1, J = 3 for model 2, J = 6 for model 3); K denotes the number of the control variables included in the model: series of the survey^¹¹ , gender, age, size of household, presence of at least one unemployed person (K = 10 in all the models);

γ_{k}

denotes the ‘fixed slopes’ associated with the control variables;

β_{0 t}

denotes the ‘random intercept’ and

β_{j t}

denotes the ‘random slopes’ associated with the structural variables.

Moreover, the letter u denotes the residuals at the second level. With the aim of estimating the variations across time by means of the second level residuals, we consider the random slope for each of the main independent variables in each of the three models. Beta and gamma coefficients of the regression are the values we wanted to estimate.^¹² We used the PQL2 procedure (penalised quasi-likelihood, second order estimation) as suggested by Rasbash et al. (2009) and Rodriguez and Goldman (2001).^¹³

4. Results of the analysis

The purpose of the analysis was to describe the year-to-year fluctuations in the associations between poverty and the traditional social structural indicators using hierarchical regression models with household at the first level nested in units of time at the second level, controlled for some potential confounding variables. The results of the multilevel models are set out in Table 1.

TABLE 1.

Estimates of regression coefficient (logit) and standard errors on the chances of a family being poor (OECD modified)

		Model 1.1	Model 1.2	Model 1.3*
	Intercept	–2.590 (0.039)	–3.781 (0.031)	–3.994 (0.058)
Series of survey	From 1997 to 2011	–0.007 (0.030)	0.171 (0.015)	0.164 (0.023)
	From 1985 to 1996	0	0	0
Gender	Female	0.234 (0.009)	0.162 (0.009)	0.222 (0.015)
	Male	0	0	0
Age (years)	<35 years	–0.253 (0.013)	0.298 (0.014)	0.153 (0.017)
	35–44 years	–0.221 (0.012)	0.229 (0.012)	0.096 (0.015)
	45–54 years	–0.225 (0.011)	0.008 (0.012)	–0.029 (0.014)
	55–64 years	0	0	0
	>64 years	0.830 (0.010)	0.621 (0.010)	0.216 (0.018)
Size of household	1	0.356 (0.011)	0.362 (0.011)	–0.333 (0.014)
	2	0.101 (0.009)	0.085 (0.009)	–0.284 (0.014)
	3–4	0	0	0
	5+	0.636 (0.012)	0.537 (0.012)	0.577 (0.013)
Unemployment	At least one member unemployed	0.552 (0.010)	0.462 (0.010)	0.311 (0.013)
	No unemployed	0	0	0
Structural variables
Geographic area	Southern Italy and Islands	1.199 (0.031)⁺	1.183 (0.031)⁺	1.213 (0.033)⁺
	Central and northern Italy	0	0	0
Educational attainment	Primary	–	1.585 (0.024)⁺	0.948 (0.033)⁺
	Lower secondary	–	0.915 (0.023)⁺	0.560 (0.022)⁺
	Upper secondary or tertiary	–	0	0
Social class	Manual workers	–	–	1.087 (0.057)⁺
	Petty bourgeoisie	–	–	0.700 (0.044)⁺
	White collars	–	–	0.372 (0.036)⁺
	Bourgeoisie	–	–	0
Variation year-to-year (random part)	–Intercept	0.031 (0.009)	0.019 (0.006)	0.071 (0.023)
	–Southern Italy and Islands	0.025 (0.007)	0.029 (0.008)	0.026 (0.008)
	–Primary		0.013 (0.004)	0.023 (0.008)
	–Lower secondary		0.011 (0.004)	0.008 (0.004)
	–Manual workers		–	0.024 (0.010)
	–Petty bourgeoisie		–	0.038 (0.014)
	–White collars		–	0.075 (0.024)
VPC at second level^a		0.009	0.006	0.021
	Valid cases	747204	747204	509342

		Model 1.1	Model 1.2	Model 1.3*
	Intercept	–2.590 (0.039)	–3.781 (0.031)	–3.994 (0.058)
Series of survey	From 1997 to 2011	–0.007 (0.030)	0.171 (0.015)	0.164 (0.023)
	From 1985 to 1996	0	0	0
Gender	Female	0.234 (0.009)	0.162 (0.009)	0.222 (0.015)
	Male	0	0	0
Age (years)	<35 years	–0.253 (0.013)	0.298 (0.014)	0.153 (0.017)
	35–44 years	–0.221 (0.012)	0.229 (0.012)	0.096 (0.015)
	45–54 years	–0.225 (0.011)	0.008 (0.012)	–0.029 (0.014)
	55–64 years	0	0	0
	>64 years	0.830 (0.010)	0.621 (0.010)	0.216 (0.018)
Size of household	1	0.356 (0.011)	0.362 (0.011)	–0.333 (0.014)
	2	0.101 (0.009)	0.085 (0.009)	–0.284 (0.014)
	3–4	0	0	0
	5+	0.636 (0.012)	0.537 (0.012)	0.577 (0.013)
Unemployment	At least one member unemployed	0.552 (0.010)	0.462 (0.010)	0.311 (0.013)
	No unemployed	0	0	0
Structural variables
Geographic area	Southern Italy and Islands	1.199 (0.031)⁺	1.183 (0.031)⁺	1.213 (0.033)⁺
	Central and northern Italy	0	0	0
Educational attainment	Primary	–	1.585 (0.024)⁺	0.948 (0.033)⁺
	Lower secondary	–	0.915 (0.023)⁺	0.560 (0.022)⁺
	Upper secondary or tertiary	–	0	0
Social class	Manual workers	–	–	1.087 (0.057)⁺
	Petty bourgeoisie	–	–	0.700 (0.044)⁺
	White collars	–	–	0.372 (0.036)⁺
	Bourgeoisie	–	–	0
Variation year-to-year (random part)	–Intercept	0.031 (0.009)	0.019 (0.006)	0.071 (0.023)
	–Southern Italy and Islands	0.025 (0.007)	0.029 (0.008)	0.026 (0.008)
	–Primary		0.013 (0.004)	0.023 (0.008)
	–Lower secondary		0.011 (0.004)	0.008 (0.004)
	–Manual workers		–	0.024 (0.010)
	–Petty bourgeoisie		–	0.038 (0.014)
	–White collars		–	0.075 (0.024)
VPC at second level^a		0.009	0.006	0.021
	Valid cases	747204	747204	509342

*

Conditioned model: household with at least one member employed.

+

Random slope.

a

The dichotomous dependent variable could be considered a continuous variable truncated by an arbitrary threshold (the poverty line). Thus, following the suggestions of Snijders and Bosker (1999), the variation partition coefficient (VPC) at the second level is: $ρ_{t} = Ω_{t} / (Ω_{t} + π^{2} / 3) = Ω_{t} / (Ω_{t} + 3.29)$ ⁠, where $Ω_{t}$ is the variation year-to-year.

The estimates in Model 1.1 are significant in the expected direction. The coefficients indicate a greater chance of being poor for households in southern regions of Italy, ceteris paribus demographic characteristics and the presence of unemployed persons in the family. The average estimates added to residuals of the temporal contexts are shown in Figure 3. It should be borne in mind that we were interested in describing the association between structural variables and poverty in the temporal dimension. The variation partition coefficient (VPC) or intra-year correlation is very low, about 1%. This value confirms a strong stability across time. However, close consideration of the trend in Figure 3 shows a slight increase in the risk of poverty for southern families in the early 1990s. The economic crisis that involved many Western countries in that period seems to have disadvantaged Italy's southern regions in particular.

Model 1.1. Estimates (logit) of the geographical area, residuals of the year-to-year fluctuations added to average coefficients. Reference category: Northern and Central Italy.

Figure 3.

Model 1.1. Estimates (logit) of the geographical area, residuals of the year-to-year fluctuations added to average coefficients. Reference category: Northern and Central Italy.

The focus of Model 1.2 is on educational attainment, controlled for demographic variables, series of the survey, unemployment and geographical area (which is a structural but antecedent condition). On average, where the reference person has a primary level of education, households are more likely to be in poverty (the coefficient is 1.585). The heterogeneity explained at the time level is still very low (0.6%). Figure 4 shows residuals added to average coefficients. More specifically, one notes an increase in the early 1990s in the association between poverty and education for those households in which the reference person has a lower secondary education with respect to the reference category (households with upper secondary or tertiary educations), probably due to the consequences of the economic crisis. Moreover, Figure 4 shows a slight decrease in the association in recent years for persons with a primary education. This may be due to credential inflation: that is, the presence of younger generations with a larger number of graduates, and the progressive disappearance of older cohorts with low levels of education (Bernardi 2003).

Model 1.2. Estimates (logit) of educational attainment, residuals of the year-to-year fluctuations added to average coefficients. Reference category: Upper secondary or tertiary.

Figure 4.

Model 1.2. Estimates (logit) of educational attainment, residuals of the year-to-year fluctuations added to average coefficients. Reference category: Upper secondary or tertiary.

The last Model 1.3 focuses on the association between poverty and social class. The estimates are significant in the expected direction: households in the manual working class are more likely to be poor, also controlling for educational attainment and other confounding variables. Also the intra-unit correlation remains very low, about 2.1%. Figure 5 clearly shows that the association is stable during the 27 years considered.^¹⁴ A discontinuity is apparent from 1996 to 1997, but is likely due to the different questionnaire used in the survey (in this case, the dummy variable does not seem able to capture the bias). However, if we consider the graph closely, the early 1990s show a slight increase in the likelihood of being poor for the manual working class compared with other classes.

Summarising, the analysis has two main results. The first is the strong stability of the association between poverty and structural variables over time, from 1985 to 2011. Hierarchical models show that the variability explained by the temporal contexts is very low. Also on considering the periods of economic crisis, the associations are stable; rather, they show a slight increase of association in these periods (in particular in the early 1990s). However, the current crisis that began in 2008, at least until 2011, has not yet shown evident changes.

Model 1.3. Estimates (logit) of social class, residuals of the year-to-year fluctuations added to average coefficients. Reference category: Bourgeoisie.

Figure 5.

https://doi.org/10.1016/j.rssm.2013.05.001

Model 1.3. Estimates (logit) of social class, residuals of the year-to-year fluctuations added to average coefficients. Reference category: Bourgeoisie.

5. Conclusions

In this article, we have dealt with the debate on individualisation and poverty. We have seen that scholars for and against individualisation theory share the same methodological strategies that can be summarised in two different streams: highlighting individual rather than structural aspects of social action and controlling for individual rather than structural variables. Both these methodological strategies deal with time. In fact, most of the studies mentioned above use not poverty status but poverty duration as their dependent variable, applying a dynamic approach to the study of the phenomenon; they usually use panel data and consider quite short observation periods. These methodological strategies enable us to gain an understanding of the interrelationships between life course, structural variables and poverty, but they do not tell us whether or not the relationships they illuminate hold over extended periods of time. From this point of view, it is extremely important to consider another dimension of time: that which concerns the individual-level relationship between structures and poverty and its change over long time periods. In this article, we switched to focus on repeated cross-sectional surveys in order to give us wide observation windows. Such data make it possible to assess the core of individualisation theory, which understands contemporary social change as the progressive weakening of the influence of social structures on individual behaviour.

We focused our analysis on Italy from 1985 to 2011 and were able to corroborate our hypothesis that the association of poverty with structural variables (area of residence, education, occupational class), controlling for life course variables, has remained stable over time. Moreover, if we consider the periods of economic crisis, the results show a slight increase of association between poverty and structural variables in these periods (in particular in the early 1990s). However, the current crisis that began in 2008, at least up until 2011, has not yet shown evident changes.

According to our perspective, these results corroborate stratification theory but they do not entail that individualisation theory should be completely abandoned. The two theories should continue their dialogue in order to reshape and improve each other. In the studies that we have classified in the second methodological strategy, we already see the interrelationships between life course and structural variables at work; interestingly, social determinants seem to mediate the role of family events but not those of jobs and careers. Even within the strategy we implemented, a more sophisticated control for life course variables ought to be developed (Mayer 2009).

A second general remark concerns the research design. We have focused on a single nation (Italy) but the results could be different in other countries with different levels of economic condition and diverse welfare regimes, as other scholars have already highlighted (Layte and Whelan 2002: 224). This calls for adopting a comparative perspective in the OECD framework. Scholars should be aware that the methodological strategy here implemented puts heavy constraints on data availability that standard harmonised data-sets are not always able to fulfil: it requires datasets where reliable poverty variables and harmonised structural variables are available over a long time period. In some cases, a comparative project may require a stream of collaboration between national research groups.

Footnotes

1.

Translated by the authors.

2.

The different poverty patterns were derived from a latent class analysis of the complete ECHP.

3.

In all these papers, the authors have analysed how structural variables mediate the effects of life course variables, but it is also possible to find in the literature examples of the reverse approach, i.e. controlling for individual variables (Layte and Whelan 2002: 228–230).

4.

The main advantage of the diary technique of data collection is that it furnishes reliable information on low-level expenses or daily consumption activities (Fortini 2000).

5.

See also http://www.oecd.org/redirect/dataoecd/61/52/35411111.pdf. Used in Italy is another scale of equivalence, called Carbonaro (1985). In this alternative formula, the weights are based only on the size of household. For 1 member the weight is 0.6, for 2 members it is 1, for 3, 1.33, for 4, 1.63, for 5, 1.9, for 6, 2.16 and for 7 or more it is 2.4. We also considered this scale to test the robustness of analysis. The results can be obtained from the authors.

6.

The reference person is the respondent of the familial section of the questionnaire in the surveys (he/she corresponds to the traditional concept of head of household).

7.

In preliminary analysis we considered other variables to describe family life course more accurately (children 0–5 years old in the household, children 0–14 years old in the household, individuals 65 or more years old in the household, individuals 75 or more years old in the household) but their effects were mostly captured by other variables already in the models (age of the reference person, size of households) due to collinearity problems.

8.

Southern regions include Sicily and Sardinia. We tested a geographical categorization in five areas, but north-western, north-eastern, and central Italy showed similar coefficients. Likewise southern Italy and islands (Sardinia and Sicily) showed similar coefficients in the models.

9.

While in some other countries it is usual to aggregate primary and lower secondary education, here we preferred to aggregate upper secondary and tertiary levels, given the low percentage of graduates that characterises Italy (OECD2012).

10.

We considered only the households in which at least one member was employed. We attributed to the household the social class of the individual with the higher employment status (following the usual hierarchy: bourgeoisie, white collars, petty bourgeoisie and manual workers).

11.

This dummy variable varies only at the second level.

12.

We used MLWIN software.

13.

If the number of units in the first level is large and contexts are homogenous in size, and the dependent variable is not too asymmetric, bias in estimates are not relevant (Goldstein and Rasbash 1996).

14.

Similar results were reached by Albertini (2013: 34) on the basis of different methods (index decomposition and quantile regression analysis) and focusing on economic inequality: ‘Descriptive statistics, therefore, signal that in the last three decades the relation between individuals’ social class and economic situation has not weakened’.

References

Albertini

,

M.

(

2013

) ‘

The relation between social class and economic inequality: A strengthening or weakening nexus? Evidence from the last three decades of inequality in Italy

’,

Research in Social Stratification and Mobility

33

:

27

–

39

.

https://doi.org/10.1111/j.1468-4446.2007.00155.x

Atkinson

,

W.

(

2007

) ‘

Beck, individualization and the death of social class: a critique

’,

The British Journal of Sociology

58

(

3

):

349

–

66

.

PubMed

Beck

,

U.

(

1986

)

Risikogesellschaft. Auf dem Weg in eine andere Moderne [Risk Society. Towards a New Modernity]

,

Frankfurt am Main

:

Suhrkamp Verlag

.

Beck

,

U.

and

Beck-Gernsheim

,

E.

(

2002

)

Individualization. Institutionalized Individualism and Its Social and Political Consequences

,

London

:

Sage

.

https://doi.org/10.1093/esr/19.1.25

Bernardi

,

F.

(

2003

) ‘

Educational performance at entry into the Italian labour market

’,

European Sociological Review

19

(

1

):

25

–

40

.

https://doi.org/10.2383/24195

Bernardi

,

F.

(

2007

) ‘

Le quattro sociologie e la stratificazione sociale [The four sociologies and social stratification]

’,

Sociologica

1

: 1–13

.

Biolcati Rinaldi

,

F.

and

Podestà

,

F.

(

2008

) ‘Two countries in one: the working poor in Italy’,

in H.-J. Andreß, and H. Lohmann (Eds), Working Poor in Europe: Employment, poverty and globalization.

Edward Elgar, Cheltenham

, pp.

203

–

26

.

Biolcati-Rinaldi

,

F.

and

Giampaglia

,

G.

(

2011

) ‘

Dinamiche della povertà, persistenze e corsi di vita [Poverty dynamics, persistence and life courses]

’,

Quaderni di Sociologia

LV

(

56

):

151

–

79

.

https://doi.org/10.1093/esr/18.3.371

Boudon

,

R.

(

2002

) ‘

Sociology that really matters: European academy of sociology, first annual lecture, 26 October 2001, Swedish cultural center

’,

European Sociological Review

18

(

3

):

371

–

8

.

Brandolini

,

A.

(

2009

) ‘L'evoluzione recente della distribuzione del reddito in Italia [The recent evolution of income distribution in Italy]’, in

A.

Brandolini

,

C.

Saraceno

and

A.

Schizzerotto

(eds),

Dimensioni della disuguaglianza in Italia: povertà, salute, abitazione [Dimensions of inequality in Italy: poverty, health, housing]

,

Bologna

:

il Mulino

, pp.

39

–

67

.

https://doi.org/10.1177/0038038597031003006

Breen

,

R.

(

1997

) ‘

Risk, recommodification and stratification

’,

Sociology

31

(

3

):

473

–

89

.

https://doi.org/10.1111/j.1475-4991.1988.tb00564.x

Buhmann

,

B.

,

Rainwater

,

L.

,

Schmaus

,

G.

and

Smeeding

,

T. M.

(

1988

) ‘

Equivalence scales, well-being, inequality, and poverty: Sensitivity estimates across ten countries using the Luxembourg Income Study (LIS) database

’,

Review of Income and Wealth

34

(

2

):

115

–

42

.

Carbonaro

,

G.

(

1985

)

‘Nota sulle Scale di Equivalenza,’ La Povertà in Italia [Note on the equivalence scales]

,

Presidenza Del Consiglio Dei Ministri, Istituto Poligrafico Dello Stato

,

Roma

.

. http://www.nber.org/papers/w3964

Cutler

,

D. M.

and

Katz

,

L. F.

(

1992

) ‘

Rising inequality? Changes in the distribution of income and consumption in the 1980s, NBER working paper no. 3964

’,

American Economic Review

82

(

2

):

546

–

51

DiPrete

,

T. A.

and

Grusky

,

D. B.

(

1990

) ‘

Structure and trend in the process of stratification for American men and women

’,

American Journal of Sociology

96

(

1

):

107

–

43

.

https://doi.org/10.1177/095892879600600102

Ferrera

,

M.

(

1996

) ‘

The “Southern Model” of welfare in social Europe

’,

Journal of European Social Policy

6

(

1

):

17

–

37

.

Firebaugh

,

G.

(

2008

)

Seven Rules for Social Research

,

Princeton, NJ

:

Princeton University Press

.

Firebaugh

,

G.

(

1997

)

Analyzing Repeated Surveys

,

Beverly Hills, CA

:

Sage

.

, http://www.istat.it/strumenti/metodi/lineeguida.pdf

Fortini

,

M.

(

2000

)

Linee guida metodologiche per rilevazioni statistiche [Methodological guidelines for statistical surveys]

,

Istituto Nazionale di Statistica

Freguja

,

C.

and

Pannuzi

,

N.

(

2007

) ‘La povertà in Italia: che cosa sappiamo dalle varie fonti? [Poverty in Italy: What do we know from various sources?]’, in

A.

Brandolini

and

C.

Saraceno

(eds),

Povertà e benessere. Una geografia delle disuguaglianze in Italia [Poverty and well-being. A geography of inequalities in Italy].

Bologna

:

il Mulino

, pp.

23

–

59

.

Goldstein

,

H.

and

Rasbash

,

J.

(

1996

) ‘

Improved approximations for multilevel models with binary responses

,

Journal of the Royal Statistical Society. Series A (Statistics in Society)

159

(

3

):

505

–

13

.

https://doi.org/10.1080/713601612

Goldthorpe

,

J.H.

(

2002

) ‘

Globalisation and social class

’,

West European Politics

25

(

3

):

1

–

28

.

Hox

,

J. J.

(

2002

)

Multilevel Analysis: Techniques and Applications

,

Mahwah, NJ

:

Lawrence Erlbaum

.

Layte

,

R.

and

Whelan

,

C. T.

(

2002

) ‘

Cumulative disadvantage or individualisation? A comparative analysis poverty risk and incidence

’,

European Societies

4

(

2

):

209

–

33

.

https://doi.org/10.1177/000169930104400201

Layte

,

R.

,

Whelan

,

C. T.

,

Maitre

,

B.

and

Nolan

,

B.

(

2001

) ‘

Explaining levels of deprivation in the European Union

’,

Acta Sociologica

44

(

2

):

105

–

21

.

Leisering

,

L.

(

2003

) ‘

The two uses of dynamic poverty research – Determination and contingent models of individual poverty careers

’,

University of Bielefeld, Institute for World Society Studies

, Social Policy – Working Paper No. 6.

Leisering

,

L.

and

Liebfried

,

S.

(

1999

)

Time and Poverty in Western Welfare State. United Germany in Perspective

,

Cambridge

:

Cambridge University Press

.

Leisering

,

L.

and

Walker

,

R.

(eds.) (

1998

)

The Dynamics of Modern Society. Poverty, Policy and Welfare

,

Bristol

:

The Policy Press

.

Lucchini

,

M.

and

Sarti

,

S.

(

2005

) ‘

Il benessere e la deprivazione delle famiglie italiane [The well-being and deprivation of Italian households]

’,

Stato e mercato

(

74

):

231

–

66

.

https://doi.org/10.1146/annurev.soc.34.040507.134619

Mayer

,

K. U.

(

2009

) ‘

New directions in life course research

’,

Annual Review of Sociology

35

(

1

):

413

–

33

.

Mejer

,

L.

and

Siermann

,

C.

(

2000

)

Income Poverty in the European Union: Children, Gender and Poverty Gaps, Statistics in Focus

,

Luxembourg

:

Eurostat

.

Morlicchio

,

E.

(

2012

)

Sociologia della povertà [Sociology of poverty]

,

Bologna

:

il Mulino

.

Negri

,

N.

and

Saraceno

,

C.

(

1996

)

Le politiche contro la povertà in Italia [Policies against poverty in Italy]

,

Bologna

:

il Mulino

.

. http://dx.doi.org/10.1787/eag-2013-en

OECD

. (

2013

)

Education at a Glance 2013: OECD Indicators

,

OECD Publishing

Ranci

,

C. (ed.

) (

2010

)

Social Vulnerability in Europe. The New Configuration of Social Risk

,

Basingstoke

:

Palgrave Macmillan

.

Rasbash

,

J.

,

Steele

,

F.

,

Browne

,

W. J.

and

Goldstein

,

H.

(

2009

)

A User's Guide to MLwiN, Bristol: Centre for Multilevel Modelling, ISBN

:

978-0-903024-97-6

.

Rodriguez

,

G.

and

Goldman

,

N.

(

2001

) ‘

Improved estimation procedures for multilevel models with binary response: a case–study

’,

Journal of the Royal Statistical Society: Series A (Statistics in Society)

164

(

2

):

339

–

55

.

Saraceno

,

C.

(ed.) (

2002

)

Social Assistance Dynamics in Europe. National and Local Poverty Regimes

,

Bristol

:

The Policy Press

.

Saraceno

,

C.

(

2003

)

Mutamenti della famiglia e politiche sociali in Italia [Changes in family and social policies in Italy]

,

Bologna

:

il Mulino

.

Schizzerotto

,

A.

(ed.) (

2002

)

Vite ineguali. Disuguaglianze e corsi di vita nell'Italia contemporanea [Unequal lives. Inequalities and life courses in contemporary Italy].

Bologna

:

il Mulino

.

Snijders

,

T.

and

Bosker

,

R.

(

1999

)

Multilevel Analysis

,

London

:

Sage

.

Sørensen

,

A. B.

(

1998

) ‘Theoretical mechanisms and the empirical study of social processes’, in

P.

Hedström

and

P.

Swedberg

(eds),

Social Mechanisms. An Analytical Approach to Social Theory.

Cambridge

:

Cambridge University Press

.

Taylor-Gooby

,

P. (ed.

) (

2004

)

New Risks, New Welfare. The Transformation of the European Welfare State

,

New York

:

Oxford University Press

.

https://doi.org/10.1080/14616690903056005

Vandecasteele

,

L.

(

2010

) ‘

Poverty trajectories after risky life course events in different European welfare regimes

’,

European Societies

12

(

2

):

257

–

78

.

https://doi.org/10.1093/esr/jcq005

Vandecasteele

,

L.

(

2011

) ‘

Life course risks or cumulative disadvantage? The structuring effect of social stratification determinants and life course events on poverty transitions in Europe

’,

European Sociological Review

27

(

2

):

246

–

63

.

Walker

,

R.

(

1994

)

Poverty Dynamics: Issues and Examples, in collaboration with Karl Ashworth.

Avebury

:

Aldershot

.

Whelan

,

C. T.

and

Maître

,

B.

(

2008a

) ‘“

New” and “Old” social risks: Life cycle and social class perspectives on social exclusion in Ireland

’,

The Economic and Social Review

39

(

2

):

131

–

56

.

https://doi.org/10.1111/j.1468-4446.2008.00213.x

Whelan

,

C. T.

and

Maître

,

B.

(

2008b

) ‘

Social class variation in risk: A comparative analysis of the dynamics of economic vulnerability

’,

The British Journal of Sociology

59

(

4

):

637

–

59

.