The workhorse model for analysing whether social assistance benefit receipt creates dependence effects is overly simplistic and unreliable for empirical work, and patterns of time dependence are much more complex than previously assumed. These are the main findings of research by Christian Brinch, Manudeep Bhuller and Sebastian Königs, published in the September 2017 issue of the Economic Journal.

Tracking Norwegian households'' ''social assistance careers'' month by month over a ten-year period, Brinch and colleagues demonstrate that empirical estimates of time dependence effects strongly vary with the time scale chosen for analyses. They identify much richer path dependence in their monthly data than what has typically been assumed in standard models run on annual data.

The study finds that social assistance receipt seems to produce ''scarring'' effects among working-age persons in Norway: those who live in households that receive social assistance in a given month are on average over four times more likely to rely on benefits 12 months later than otherwise comparable households who received no benefits.

Already short benefit episodes moreover appear to have a lasting effect: model estimates suggest that at the beginning of a benefit spell, every additional month of social assistance receipt raises the chances of remaining on benefits by over five percentage points (that is, there is ''duration dependence'').

Any previous episode on benefits moreover permanently doubles a person''s likelihood of returning to social assistance again later (''occurrence dependence'').

While the study does not shed light on the exact drivers of benefit dependency, a number of reasons have been discussed in earlier research:

• A standard explanation is that income support might make it less attractive for low-income households to raise their earnings by taking up work or increasing hours thereby promoting future benefit receipt.

• Income support receipt may, however, also make finding work more difficult if there is a stigma attached to transfer payments that may make employers more hesitant to hire a benefit recipient.

• Another possible explanation is that past recipients may simply be more familiar with the procedures of filing a claim, such that they may be more likely to again rely on income support in the future.

''Time Aggregation and State Dependence in Welfare Receipt'' by Christian Brinch, Manudeep Bhuller and Sebastian Königs is published in the September 2017 issue of the Economic Journal. Christian Brinch is at BI Business School, Oslo. Manudeep Bhuller is at the University of Oslo and Statistics Norway. Sebastian Königs is at the OECD.