I used to blog fairly regularly about the Basic Income (aka Citizen's Income, aka Guaranteed Annual Income), but the passage of a decent interval of time, Kevin Milligan's recent Economy Lab columns ( , ) and Erin Anderssen's long article in Saturday's Globe and Mail gives me an excuse to revisit the issue and perhaps repeat myself. I had almost finished writing this before I saw Kevin's most recent column, so there's some overlap that I can't be bothered to edit out at this point.
It doesn't matter if you don't know what the various acronyms stand for; the point is that for low incomes, the rate is almost 100%, and there are even income levels where earning more market income will reduce after-tax income. Many low-income earners will - quite sensibly - choose not to work if doing so will reduce their incomes. For a given level of taxation, the small general equilibrium model I discussed here predicts that participation rates and output are higher under a basic income program than under conventional social assistance.
Actually, the labour supply issue is one of the odder things about the BI debate. A common objection to BI proposals is that "people would get money for just staying at home and not working". My response is that this would be better than the system we already have, in which people are paid to stay home and are penalised if they accept work.
But efficiency arguments aren't the only thing to consider; the point of these programs is to reduce inequality and/or ensure that people can maintain a certain minimum standard of living. And it's not clear to me that the basic income model will always dominate conventional social assistance.
If the amount of revenue available for redistribution is 'small', the welfare gains from giving it all to the least well-off - that is, those whose marginal utility is highest - may be greater than the total utility gains generated by giving the entire population an equal share of a tiny pie. And in fact, this is exactly what happens in the small GE model. If taxes are low, the fact that conventional social assistance offer higher payments to those in most need outweighs the efficiency losses associated with lower participation rates. For tax rates higher than a certain critical threshold, basic income offers higher social welfare, and the unconstrained optimum under basic income is higher than under conventional social assistance.
Of course, as the generosity of the BI increases, so does the likelihood that those working for low wages will stop participating in the labour market. This isn't necessarily a bad thing: substituting away from consumption and towards leisure or home production can be welfare-improving. (See also Nick's post on instrumentalism.) But since it reduces the tax base, tax rates would have to increase in order to compensate. BI advocates have a developed the bad habit of dismissing this point, and they would do well to take it more seriously.
So basic income proponents should be prepared for making the case for more tax revenues. How much more is a question that has seen surprisingly little serious work. And by 'serious', I mean analyses that incorporate behavioural effects.
There's a proposal being floated around Quebec policy circles that would ensure that someone who works 16 hours per week at minimum wage would still have an income that would put her above Statistics Canada's poverty threshold for Quebec (about $12,000). These amounts are far from lavish, but the costs are surprisingly high. In this report, some of my Laval colleagues estimate that the lower bound is on the order of $2.2b. To put this in perspective, one percentage point of the Quebec TVQ (the provincial GST) generates about $1.2b in revenues. Financing even this modest proposal would require the equivalent of at least two extra percentage points to the TVQ.
The RSS tag the Globe put on Kevin's recent Economy Lab post says "If it's not implemented properly, a guaranteed annual income could become a very costly program". I think it's better to say that if it's not costly, then the BI isn't being isn't being implemented properly. There's really not much point in arguing for a BI if you're not prepared to argue for significantly higher taxes.
Update: Earlier today, Kevin Milligan posted a good line on Twitter that summarises the main point quite well, in 140 characters or fewer:
The GAI impossible trinity: you can only have 2 of 3:low clawback rate; big base transfer; same cost as existing system.