We are unable to do this for the ratio measure of perceived pay inequality. Year dummy variables—one or two, depending on how many years of ISSP data are available for the particular country—are the independent variables of interest. The regressions include a variety of controls for individual characteristics: education years of schooling completed , income family income , subjective class position, employment status employed or not employed , union membership member or nonmember , age and sex.
If a year dummy variable is statistically and substantively significant in such a regression, this heightens our confidence that a genuine change has occurred in the perceived level of inequality or in support for redistribution. We do not show the full results of these regressions here they are available on request.
Instead, we simply note them in the text. The focus of the median-voter model, and of much of the broader interest among social scientists in redistribution, is on the degree of intended generosity of social programs. However, both of these measures fuse intended generosity with need.
Expenditures and actual redistribution will be higher if programs are structured more generously but also if more people are unemployed, elderly, poor and so on. As suggested earlier, this is a problem for testing the median-voter hypothesis: if market inequality and redistribution both increase over time, the latter could be a product of increases in the number of people making use of redistributive programs rather than of changes in the generosity of those programs.
A better strategy for assessing the intended generosity of programs is to use a measure of program details Esping-Andersen, ; Korpi and Palme, ; Allan and Scruggs, Such a measure directly taps the degree of intended generosity of redistributive programs, which is the theoretically relevant concept. Until very recently, no such measure was available with over-time data.
We use several here. Three are from a data set compiled by Lyle Scruggs on various program details for three key types of redistributive policies: public pensions, unemployment insurance and sickness insurance. The data include measures of eligibility criteria, replacement rates and benefit duration. We use Scruggs' calculations of decommodification scores, which are based on a revision of Esping-Andersen's scoring procedure Scruggs, ; Scruggs and Allan, , to examine over-time trends in redistributive policy generosity.
These data are available for all of the countries we analyze. Higher decommodification scores indicate greater program generosity. Unfortunately, these data are available for only five of our eight countries and this measure does not take into account eligibility criteria or benefit duration. There are other sources of comparative data on inequality, public opinion and redistribution. For inequality, these include data from the University of Texas Inequality Project n.
For public opinion the chief alternative is the World Values Survey n.
There are only eight countries for which we can effectively match up these two databases for multiple years. Two are Nordic European countries: Norway and Sweden. Two are Continental European countries: Germany and Italy. As we noted in the introduction, it frequently has been observed that there is a negative, rather than positive, cross-sectional association between inequality and redistributive policy generosity across countries. Our interest here is in trends over time.
Should we focus on the over-time variation within each country? Or should we concentrate on the variation across countries in change over time? In our view, the latter would require asking too much of the available data on public opinion. We are not especially confident that they accurately tap the magnitude of shifts in public opinion.
We are even less confident that they are suitable for assessing differences across countries in the magnitude of such shifts. Normative views on this issue tend to vary considerably across countries Kelley and Evans, ; Marshall et al. Hence there is reason to be skeptical about whether or not similar changes in mean responses in, say, Sweden and the United States would indicate true similarity in the degree of change in the perceived level of inequality in the two countries.
We therefore confine our analyses to within-country trends over time. We treat the eight countries as, in effect, a set of case studies. For each country we begin by examining trends in inequality in the s and s. We turn next to trends in awareness of inequality and support for redistribution.
We then examine trends in the generosity of redistributive programs. Note: There are no social assistance generosity data for Norway. Vertical axes of some charts are truncated.
For variable definitions and data sources, see the appendix. The two inequality charts suggest different conclusions about recent trends in inequality in Norway. The trend for earnings inequality among full-time employed individuals suggests no noteworthy change during the s and s. Data are available for only 8 years. However, the data points are spaced sufficiently well across the two decades so that we can reasonably infer that the pattern has been one of a constant level of earnings inequality. In contrast, the time series for pretax—pretransfer household income inequality indicates an increase in inequality since the late s.
The rate of increase tapered off in the second half of the s, but the level of market income inequality at the end of the s was higher than at the beginning of the decade. Were these trends in inequality perceived accurately by Norwegians? Unfortunately, ISSP data are available only for and Consistent with the trend in earnings inequality, the data suggest no change in the perceived degree of pay inequality between and The line is flat, despite the increase in market income inequality during this period.
It too suggests no change. Regressions controlling for compositional shifts in the population indicate that there actually was an increase in the perceived level of market income inequality and in support for redistribution between and , but the increase was very small in magnitude.
How, if at all, did redistributive policy change? The decommodification indexes for pensions and sickness insurance indicate no noteworthy shift during either decade. There were some changes in pension policy, but they amounted to short-term shifts that were reversed shortly afterward. For instance, there was a decline in generosity in that was reversed in The same thing happened in and and in and The generosity of unemployment insurance was increased in and , but then remained constant through the remainder of the s and the s.
Unfortunately, data on social assistance benefits are not available. Summary : How consistent with the median-voter hypothesis were developments in Norway in the s and s? On the one hand, earnings inequality remained largely constant through the two decades, and so too did the perceived level of pay inequality and income inequality, the degree of support for redistribution, and the degree of redistributive policy generosity.
On the other hand, market income inequality among households increased significantly between the late s and the late s without a corresponding increase in the perceived level of income inequality, in support for redistribution, or in the generosity of redistributive policy.
Thus, some developments in Norway are consistent with the median-voter hypothesis while others are not. The first chart in the first row indicates that, unlike Norway, Sweden experienced an increase in earnings inequality among employed individuals. This occurred mainly in the s.
The data in the second chart in the first row indicate that market household income inequality also increased.
However, this increase occurred in the s and early s, before the rise in individual earnings inequality. In the mid-to-late s market inequality among households declined slightly. Note: Vertical axes of some charts are truncated. Like for Norway, the public opinion data for Sweden cover only the s, as Sweden was included in the ISSP social inequality modules in and but not in The first chart in that row shows that on average the perceived ratio of the pay level of a corporate chair to that of skilled and unskilled workers widened, and the second chart suggests an increase in the perceived level of market income inequality among Swedes.
Support for redistribution also appears to have increased during the s, though only during the first half of the decade.
The chart in the third row shows that the mean level of agreement that government should reduce income differences rose between and and then remained constant between and Regressions controlling for compositional shifts in the population suggest that these apparent increases in the perceived level of market income inequality and in support for redistribution were real, though because of data limitations it is possible to control only for age and sex.
Did the Swedish government respond by making redistributive programs more generous? Pension generosity was increased in the early s, but that preceded the rise in inequality.
In the late s and early s, after a decade of rising income inequality, the generosity of Swedish pensions was reduced.