Changes in disability reporting in the US and the Netherlands

In 2006, the WIA disability insurance (DI) programme was introduced in The Netherlands which reduced accessibility to disability benefits. In the first few years after the reform, DI rates have fallen dramatically and participation of 50+ workers has increased (García-Gómez et al, 2011). In this paper, we assess whether self-reporting of work disability among the Dutch has changed after the introduction of the new system, due to possible changes in norms. A possible mechanism is through the decline in the work disability prevalence in one’s reference group, which may have had an effect on perception of own work disability (Van Soest et al, 2011).

Using vignette questionnaires implemented in internet surveys (CentER panel in The Netherlands and American Life Panel – ALP) in 2003/2004, Kapteyn et al. (2007) showed lower thresholds for perceived work disability of the Dutch compared to Americans. Vignette correction of self-reported work disability reduced (by 60%) the large gap observed in self-reported work disability rates in the two countries. Given the changes in the DI programme in The Netherlands, one might expect the Dutch to have become “tougher”, i.e., less likely than before to consider that a certain level of functioning constitutes work disability. And, since there have not been similarly important changes in the disability insurance system in the US, this might have consequently lead to convergence between American and Dutch response scales.

We have combined the data of Kapteyn et al (2007) with data resulting from implementing the same questionnaire in the CentER panel and in the ALP in 2010/2011. We use these data to examine changes in reporting disability in these two countries. As in Kapteyn et al. (2007), we estimate models allowing for varying thresholds across countries, and in our case also across time. These are then used to predict counterfactual work disability in the US and in The NL in both periods using estimated response thresholds of the other country/period.

Applying old American scales decreases current disability rates in both countries. Applying new American scales on the other hand does not change the current disability rates. While thresholds for reporting disability used to be much lower in The NL than in the US, this does no longer appear to be the case, resulting in more comparable current unadjusted rates. Given the important reform of the Dutch DI programme, we could have expected to see this convergence between American and Dutch norms about what constitutes work disability. Our results however suggest this is mainly due to a considerable move of the Americans towards leniency and much less to the opposite evolution for the Dutch. This seems to be determined especially by an increase in the propensity of Americans to consider that mental health problems impair the ability to work. In particular, Americans and Dutch do no longer use different standards in this domain of work disability, which is what used to distinguish them the most.