Citation Context

...udying the distribution of self-employment and pension income and not only that of wage income for understanding household income inequality trends. In fact, changes in the distributions of work and pension incomes account for most of the trend, but the changing distribution of pension income had an equalising effect across the whole period while work income distribution shows a widening gap between below- and above-the-median income quantiles. Changes in the self1Among several contributions addressing the causes of wage and income inequality, see Topel (1997); Gottschalk and Smeeding (1997); Acemoglu (2002). 1 employment distribution had a similar impact to employment on total household income, although self-employment accounts for less than one third of total work income, so its relative role is larger. In contrast to other countries, the increased probability of earning wives is found to have had an unequalising effect for households in the left tail of the income distribution. As for the upper tail, at first it contributed to reducing inequality but then, after 1991, to increasing it. However, little is explained by the changing distribution of family types. This methodology could also prove ...

Citation Context

...ncome inequality since 1991. [Figure 1 about here.] All quantile ratios in the bottom panel show a decreasing trend until the end of the 1980s and the beginning of the 1990s. They increased significantly between 1991 and 1993 and remained relatively stable afterwards (or decreased slightly in the case of the 95/5 quantile).12 During the whole period from 2004- 1991, the 95/5 quantile ratio increased by over 40%. 11The equivalence scale (7) assumes that all household incomes are gathered and equally distributed across the household and considers the size as an approximation of household needs. Buhmann et al. (1988) show that this scale provides a good approximation to most scales used in empirical studies of income distribution in developed countries, including some scales based on other characteristics in addition to size, such as the age of household members. 12Some readers might be sceptical about such a dramatic increase of equivalent income inequality between 1991 and 1993, arguing that aside from the crisis of September 1992, 8 Figure 2 presents the nonparametric density estimation of equivalent household income in the years 1977, 1991, 2004.13 Between 1977 and 1991 some households with equivalent...

Citation Context

... income in Italy between 1977 and 2004. Changes in the distribution of work and pension incomes explain most of the trend. The higher average likelihood of wage-earning wives had an unequalising effect on households on the left tail of the income distribution. Little is explained by the changing distribution of family types. JEL: D31, D63, C51 Keywords: Inequality trends, equivalent income, counterfactual analysis, Italy. 1 Introduction According to recent comparative studies on OECD countries, the highest level of equivalent income inequality is found in the US, followed by the UK and Italy (Atkinson et al., 1995; Brandolini and Smeeding, 2008). However, while ∗Acknowledgments: I am grateful to Frank Cowell, Roberto Artoni, Andrea Brandolini, Daniele Checchi, Massimo Florio, Martin Fournier, Stephen Jenkins and Marco Manacorda for helpful discussions on earlier drafts, to two anonymous referees for their detailed comments and to seminar participants at the London School of Economics, the University of Milan, the University of Modena, ISER, and the 2004 European Economic Association Meeting in Madrid for their feedback. I would also like to thank the Bank of Italy for providing the data. Usual disclaim...

Citation Context

...lent income. A household is defined as the group of people sharing the same dwelling, independently of their kinship.10 The equivalent income of individual i belonging to household h is equal to yi =∑ j∈h zj/S, where ∑ j∈h zj is the sum of all individual incomes accruing to household h (h = 1, 2, ...H), and S is the equivalence scale. A special case of the equivalence scale is the parametric class proposed by Buhmann et al. 7For further details on the data set, see Banca d’Italia (2006), and for a thorough discussion of the quality of the SHIW data and a review of other Italian data sets, see Brandolini (1999). 8The definition of total income excluding capital, building and real estate income is sometimes used also by researchers at the Bank of Italy, see for instance D’Alessio and Signorini (2000). 9In other words, all values grater than the 99th quantile were replaced with the 99th quantile without removing them. Some authors prefer to truncate the data set beyond a certain threshold. However if we were to perform truncation instead of censoring, we would lose all information (such as labour force participation, household composition and type of earnings) about top income households, which is rel...

Citation Context

...decomposition of inequality changes in the US (with a methodology analogous to the one adopted here) can leave unexplained more than 40% of the total change in some inequality indices (Daly and Valletta, 2006, Table 2). Single-equation regression-based decomposition of inequality indices applied to Chinese data found a residual between 40% and 90% of the total change depending on the inequality index used (Morduch and Sicular, 2002, Table 2), while structural equation equivalent income inequality decomposition of Taiwan equivalent income inequality leaves about 25% of mean change unexplained (Bourguignon et al., 2001, Table 4). 14 has important policy implications. The increase in inequality after 1991 possibly suggests more redistribution, considering that among the EU countries during the 1990s Italy was the one with the lowest redistribution from top to bottom (G, 2004) and the lowest level of unemployment benefit coverage (Boeri and Brandolini, 2004). However, the differing effects of changes in the distribution of dependent employment, self-employment and pension income suggest the need to avoid a uniform approach for all types of income. As the number of wives in the labour force increased more for ...

Citation Context

...lds with equivalent income below e5000 improved their relative position and after 1991 the trend partly reversed. [Figure 2 about here.] A number of factors certainly played a role in such a distinctive trend in Italian equivalent income inequality. The consumer price annual inflation rate has gradually decreased since peaking at 20.8% in 1980 and has never been higher than 5% since 1992. As low income households are typically more vulnerable to inflation due to their lower ability to hedge against it (Mulligan and Sala-i-Martin, 2000), their limited bargaining power in the political process (Albanesi, 2007) and the frequently regressive effects of bracket creeping (Immervoll, 2005), the decreasing rate of price inflation may have contributed to the downward trend of inequality up to the end of the 1980s, becoming a much less relevant factor afterwards. Following the September 1992 currency and financial crisis, Italy went through its deepest economic crisis since WWII, which led to a series of tough policies, aimed at controlling public finances and limiting the expansion of public debt. They included two thorough reforms of the pension system, a reform of collective bargaining, the tightening o...

Citation Context

...nt household income distribution, possibly due to the recession of the early 1990s and the introduction of a number of new taxes on small business income after 1992. In contrast to the previous period considered, changing female participation in the labour force had an unequalising effect 12 regardless of the inequality index used (accounting for an additional 6-15% of total change).19 [Table 5 about here.] The result of the inequality enhancing effects of participation in the labour force by wives is in contrast with what was found, for instance, by Daly and Valletta (2006) for the US and by Biewen (2001) for East Germany, but not with what was found by others (see for instance Shorrocks (1983); Karoly and Burtless (1995) or Cancian and Reed (1998)). As for Italy, a simple logit regression of the probability of participation in the labour force by wives conditional on their own personal characteristics and their partner’s income, age and years of education shows that, while in 1977 women coupled with welleducated partners were, ceteris paribus, less likely to work, the reverse was true in 1991 and 2004. As better educated people have a higher average market income, the increase in participatio...

Citation Context

...n 1977 and 2004. Changes in the distribution of work and pension incomes explain most of the trend. The higher average likelihood of wage-earning wives had an unequalising effect on households on the left tail of the income distribution. Little is explained by the changing distribution of family types. JEL: D31, D63, C51 Keywords: Inequality trends, equivalent income, counterfactual analysis, Italy. 1 Introduction According to recent comparative studies on OECD countries, the highest level of equivalent income inequality is found in the US, followed by the UK and Italy (Atkinson et al., 1995; Brandolini and Smeeding, 2008). However, while ∗Acknowledgments: I am grateful to Frank Cowell, Roberto Artoni, Andrea Brandolini, Daniele Checchi, Massimo Florio, Martin Fournier, Stephen Jenkins and Marco Manacorda for helpful discussions on earlier drafts, to two anonymous referees for their detailed comments and to seminar participants at the London School of Economics, the University of Milan, the University of Modena, ISER, and the 2004 European Economic Association Meeting in Madrid for their feedback. I would also like to thank the Bank of Italy for providing the data. Usual disclaimers apply. †University of Milan ...