ESRI explains why dole report removed


The Economic and Social Research Institute (ESRI) last night published a statement specifying why a paper which said a high proportion of Irish people with children would be better off on the dole than in employment was withdrawn from its website.

“The Costs of Working in Ireland” paper was published on the ESRI website on May 22nd. It was subsequently removed because of what the body described as “concern that the public could be misled by its content”.

Last night the ESRI published a note explaining its concerns about the methodology used in drawing up the paper. It says the note has also been sent to the report’s authors. In the note, it says the equation used to estimate income in the paper, co-authored by Prof Richard Tol, “is poorly specified and applied to an inappropriate sample”. Its conventional way of estimating income when applied to a sample of full-time employees leads to “a very substantial reduction in the key findings relating to the proportion of individuals who would have a higher income in unemployment when the estimated costs of working are taken into account”.

The ESRI says the note issued last night shows how the approach adopted by the original paper generates “widely differing and therefore unreliable estimates of income from work”.

The note also says the institute is “concerned about inconsistencies” in the units of analysis used in the paper with expenditure data “measured at household level while income is measured at the individual level”. It questions assumptions made by the authors that all individuals who make the transition to work face the same set of costs regardless of age, education or marital status.

Finally, it says the model used to estimate wages and therefore the in-work and out-of-work results were not the most appropriate: “It is standard to use a Heckman model to adjust for the effects of selection bias when estimating predicted wages.”

Earlier this week Prof Tol, who left the ESRI some six months ago and who now works at Sussex University, said that two separate sets of data from the Central Statistics Office had been considered in the preparation of the paper. Asked whether the withdrawal of his paper was an example of his not being allowed by the institute to publish “true information”, he said: “It could easily be constructed as such. I don’t know why this decision was made or what pressure led to this decision.”