The Conservatives' decision to cut two percentage points from the GST removed roughly $12-billion from the federal government's revenues, and closing this hole in the budget balance will require some combination of spending cuts and tax increases. While the Conservatives have made it clear that they will not increase taxes, the opposition parties have proposed that instead of cutting programs, some taxes should increase. But the estimated revenues that are associated with these measures are overstated by a factor of at least two, and probably much more.
Jean-Baptiste Colbert - finance minister for Louis XIV- is quoted as saying that "the art of taxation consists in so plucking the goose as to obtain the largest amount of feathers with the least possible amount of hissing". In the jargon of modern economics, this principle translates as "tax the inelastic factor".
It is standard practice in Canadian policy circles to assume that for the purposes of estimating the changes in revenues from a given tax measure, people will make exactly the same decisions as before, with the only difference being the difference in the amount of tax they pay. This assumption is made because it simplifies the analysis, not because anyone believes that it's a reasonable assumption to make. All tax changes produce a behaviour response of some kind.
For some tax proposals -- increasing carbon taxes, reducing corporate income taxes -- the primary goal is to induce a desired behavioural response; the effect on revenues is generally a secondary concern. But if the primary concern is revenue generation, increasing taxes that are easily-avoided won't do much to increase revenues. It turns out that two measures circulating among the opposition parties -- increasing personal income taxes for high earners and increasing corporate taxes -- fall into that second category.
There is a broad consensus in the empirical public finance literature that when personal income tax rates increase, people report lower levels of taxable income; see here for a recent survey done by the Department of Finance. Roughly speaking, an 'elasticity of taxable income' (ETI) of 0.4 means that a 10 per cent decrease in the net-of-tax rate (one minus the tax rate) reduces reported taxable income by about 4 per cent on average, and up to 7 per cent for those with high incomes.
One of the policy proposals offered by NDP leadership candidate Brian Topp is to create a new tax bracket such that those with taxable incomes above $250,000 will see their marginal tax rates increase by 6 percentage points. When provincial taxes are taken into account, this reduces the net-of-tax rate by about 12 per cent. If we assume -- as Topp does -- that high earners will not respond to this increase and if we apply it to the 2008 tax base, the higher tax rate generates an additional $3-billion in revenues. But if we use an ETI of 0.4 to estimate the effect on the tax base, the increase in revenues falls to $0.6-billion. For values of ETI greater than 0.5, the effect on revenues is negative. (These numbers are based on my back-of-the-envelope calculations that are summarized here.)
A similar story can be told for corporate taxes: this study of OECD countries finds that a 10 per cent increase in the corporate tax rate reduces the tax base by about 7 per cent. It is standard practice in Canadian politics -- government and opposition parties alike -- to assume that this effect is zero. The actual effect of a corporate tax increase on revenues will be a fraction of estimates based on an assumption of no behavioural responses.
One of the reasons why the GST is such an effective generator of revenues is that it generates very little in the way of a behavioural response. The GST tax base is relatively stable: people generally buy almost as much after an increase in the GST as before.
It may be politically attractive to talk about taxing corporations and high earners, and there may be advantages for doing so that have nothing to do with revenue generation. But the effects of these measures on the budget balance are being greatly overestimated.
Stephen Gordon's recent posts and Twitterfeed can be viewed here.