I had a read of the paper on welfare benefits and work commitment. It is a conclusion I would like to believe, but I have some doubts about the methodology.

They have certainly tried to be reasonably robust, not just doing simple correlations but controlling for all sorts of individual level effects in a multi-level regression, and they've looked at different groups and interaction effects and so forth. But two main problems remain IMO:

First, the dependent variable, the response to the statement "I would enjoy having a paid job even if I did not need the money" is purely a self-reported statement of opinion. Responses may well be affected by what people feel they "ought" to say, but more importantly it says nothing about what people would (or do) actually do in a situation where they are not in work but are receiving generous benefits. They might in principle think they'd enjoy having a paid job, but be less motivated to do anything about it.

Second, their measure of "Generosity" - they've taken the total spending per capita, in PPP terms, on a range of benefits and social support (unemployment, housing, sickness & disability, families & children), and divided by the proportion of people not employed.

The PPP factor is supposed to adjust for the difference between richer and poorer countries (the smae number of Euros has less purchasing power in Norway or Switzerland than in Hungary or the Czech Republic), but this still does not account for the fact that incomes per capita are still higher in Norway and Switzerland etc., even adjusting for PPP. So this does not adequately reflect how benefit rates relate to the income of employed people. The apparent "generosity" of Norway may simply reflect that Norway is richer and that both incomes and benefit levels are higher (even adjusted for purchasing power).

Moreover, averaging over all these benefits and then dividing by the number of non-employed people is extremely crude. Non-employed covers many categories, and plenty of employed people also receive benefits. For example, income top-up or childcare payments for people in work. This means that this measure says very little about the relative economic cost people pay for not being in work. Indeed, childcare benefits increase the economic benefits of work relative to not-work.

Thus, if the question one is trying to answer is "does paying people who are out of work generous benefits reduce their incentive to seek work?", then this measure of Generosity is not very well suited to the task.

Still, as one says, an interesting contribution to the debate, but far from settling it!