Callan, Keane, Savage, Walsh and Timoney have a new paper on the incentives to work in Ireland. This is an important topic and I am glad to see continued research on this.
For the first time, Callan et al. include the cost of working in their analysis. We showed earlier that this is an important consideration. Our paper was primarily intended to address a blind spot in labour economics (which disregards expenditure patterns); our calculations were illustrative only. Unfortunately, what was meant to be an academic contribution became a public issue.
Nonetheless, it is useful to compare numbers. Callan et al. use the latest available data, and their earnings and benefits data is the best available. Earlier McGuinness and O’Connell showed that our wage equations, which omitted age as that was absent from our data, were severely biased. Callan’s estimates of the costs of childcare are from the same source as their income data. More importantly, they have information on the age composition of the household, whereas we had to work with average costs for a typical family. In these regards, the new analysis is superior to our work.
The main difference, however, is due to transport costs. There, Callan et al. run into the same problem that we did: The primary data are incomplete and one needs to rely on secondary data. This introduces all sorts of biases.
I am not convinced that Callan et al. got this right. They include only the cost of commuting, disregarding extra school runs, social calls associated with work etc. They ignore that commuters would own a different car than people who stay at home. In our paper, we compared travel costs in work and out of work, correcting for income, regardless of whether travel was from home to work and back.
Callan et al. use three methods to compute the typical cost of travel to work.
Method 1 is distance times cost per kilometer. They find 17 euro per week. But that is for the cheapest car. The range is 17-41 euro per week.
Method 2 is public transport: 14-25 euro per week.
Method 3 is based on the National Travel Survey. The starting point is an unreferenced 78 euro per week in travel costs. They then attribute 1/3 of this to commuting, and find 25 euro per week. The language is vague, but 1/3 strikes me as the national average rather than the average of those in work.
I suspect that Callan et al. underestimate the travel cost due to work. According to their Table A1.1, travel costs in the range of 15-25 euro per week, increase the fraction of people who would be better off on the dole from 4% to 5% (a 25% increase). Extrapolating, this would be 6% for a cost range of 50-100 euro per week (a 50% increase).
Childcare raises the fraction of people who would be better off on the dole from 4% to 6% (for all) and to 12% (for those with children).
Combining childcare and transport costs, 6% x 1.5 = 9% (for all) and 18% (for those with children).
McGuinness and O’Connoll found 9% (no children) and 19% (1 child under 5).
I therefore think that, although the new estimates by Callan et al. are more carefully done than our initial estimates, the new numbers are too optimistic.
Note that all of these calculations ignore undeclared income.