Page 59 - IJPS-3-1

Yin N and Heiland F

vignette data and approach makes it possible to separately identify health effects
and reporting effects rather than a mixture. We illustrate this set of results with
counterfactual policy simulations in the Results section. Specifically we focus on the
disability policy effects on the reporting scale which in turn shifts the distribution of
self-reported disability in the country.
3 Results

3.1 Sample Description
The descriptive statistics by country for our analytical sample are provided in Table 3.
The table shows large differences in years of education, with low means in the southern
European countries. There are also obvious differences in the age composition, with,
for example, relatively few 66–70 year olds in Sweden. Most chronic conditions are
much more prevalent in the U.S. than in European countries. Still, the distribution of
self-reported severity of work disability in the U.S. is quite similar to what is observed
in European countries overall (columns “U.S.” vs. “Europe”). However, there are
notable differences in the raw distributions across the seven European populations.
For example, while all distributions are fairly right-skewed, Sweden are particularly
concentrated at “none” and the Netherlands is very concentrated at “none” and “mild”,
while the distribution is relatively more equal across the five categories in Belgium.
3.2 Predicting Reporting Scales: Results from Regression Analysis

Tables 4 provides the regression estimates of the respondents’ reporting scales using
their ratings on the disability vignettes (n = 6,652). The estimated effects from four
cut-point equations are listed from left to right for cut-point 1 (“not at all limited” to
“mildly limited”) to cut-point 4 (“severely limited” to “extremely limited”). As shown
in equation (2), the left hand side of each cut-point equation is the location of the cut-
point on the severity spectrum. Controls include a detailed set of individual-level and
country-level factors. The four cut-point equations are estimated jointly as the four cut-
points together determine an individual’s response scale.
In Table 4, a negative coefficient suggests that the respondents apply a lower
cut-point when determining the severity level of the work limitation, indicating a
more generous reporting style. The results are consistent with systematic reporting
heterogeneity at the individual and country level as some of the covariates are found to
be predictive of the location of the cut-points. We are particularly interested in, and we
show the results on, how the variation in disability policy generosity across countries
predicts people’s disability reporting scales.
Some of the policy dimensions are highly correlated, as evidenced by the magnitudes
of the correlation coefficients between them. The policies, such as minimum disability
level, sickness benefit level and duration, and unemployment benefit level and duration
(relative to disability benefits), are strongly and positively correlated, as the correlation
coefficients between any two of those dimensions are about 0.7 or more. Moreover,
these policy dimensions affect the vignettes’ ratings in the similar fashion, as indicated
by the correlation coefficients between each policy dimension and the vignettes
classifying. In addition, policy dimensions such as coverage and maximum benefit
level are also highly correlated. Therefore, in our regression estimation, we group the
correlated policy dimensions to address the potential collinearity.
As shown in Table 4, more extensive coverage and higher maximum benefit level
predict more inclusive rating styles at all four cut-points, that is, over the whole
work limitation severity spectrum. Permanence of the benefits is also associated
with applying lower thresholds, with the strongest effects at the middle of the work
limitation spectrum. Easier entry into disability programs with a mild work limitation,
reflected in lower minimum disability level, more generous sickness benefits, and
better disability benefits compared to unemployment benefits, predicts more inclusive
rating styles for work limitation severity.

International Journal of Population Studies 2017, Volume 3, Issue 1 53