It's budget season, and there's a federal deficit to worry about. The government wants to cut spending, and the opposition parties want to increase taxes on high earners as well as corporate income taxes (CIT) instead. (No-one wants to increase the GST.)
In a recent Economy Lab post, I went through the arithmetic of calculating the extra tax revenues you might get from increasing taxes on high earners (spoiler: not much). In this post, I'm going to look at how much extra revenue the federal government could expect to receive from a higher CIT rate.
If you assume that there's no behavioural response, then each percentage point added to the federal CIT will generate roughly $2b in new revenues. So you'd conclude that the January 1, 2012 reduction in the CIT rate from 16.5% to 15% would reduce revenues by about $3b, and increasing the federal rate from 16.5% back to (say) 24% would increase CIT revenues by some $15b - almost one per cent of GDP.
This is the the sort of answer 'static analysis' gives. In a world in which multinationals file 57,000-page tax returns, one can only marvel at the faith in human nature among those who would make policy based on the belief that the only behavioural change on the part of corporations to an increase in CIT rates will be to put larger numbers on the cheques they send to the Receiver-General.
Here's why no-one should think that the link between increasing corporate tax rates and increasing corporate tax revenues is simply a matter of higher tax rates => higher tax revenues. Data are from the OECD:
A linear regression through these data would give you a negative relationship between CIT rates and CIT revenues: an increase in CIT rates actually reduces CIT revenues as a share of GDP. Of course, this result is almost entirely driven by the outlier that is Norway. But even if you discard Norway, you'd have a hard time extracting a significantly positive relationship between CIT rates and CIT revenues. As far as the Canadian experience goes, the steady reduction in CIT rates hasn't had much effect on revenues.
The key concept in this literature is the tax base elasticity: the parameter describing how the CIT base responds to variations in the CIT rate. I've come across two recent and relevant studies that use Canadian data to estimate CIT base elasticities. One is by Jack Mintz and Michael Smart (Journal of Public Economics version here, working paper version here), and the other is by Bev Dahlby and Ergete Ferede (International Tax and Public Finance version here, older version here). Both find strong evidence that CIT bases are highly sensitive to changes in tax rates.
But I'm somewhat reluctant to use these estimates for the task at hand. What concerns me is that much of the data variation that drives these two sets of estimates is variations across provinces, and indeed the story seems to be that tax shifting across provinces is easier than tax shifting in and out of Canada. (Provincial governments who think that a higher CIT is an effective way of increasing revenues should prepare themselves for disappointment.)
I'm inclined to give more weight to the results from this study, which uses country-level data from the OECD. The elasticities they obtain - around -0.7 for the tax rate - are substantially smaller than the ones obtained using Canadian data, so if anythings, using their numbers is going to err on the side of being too optimistic are far as revenue generation goes.
These models all have dynamics in which the effect of changes in the tax rate on the tax base increases over time. I've ignored them, mainly because the exercise is to estimate the effects of a tax change on the budget balance for FY 2012-13. But they probably do go some way in explaining the lack of any visible link between tax rates and tax revenues in the scatter plot above.
Another thing to consider is that the relevant CIT rate is the combined federal-provincial rate. An increase in the federal rate will reduce the size of a common federal-provincial tax base. Even if provincial PIT rates don't change, they will be applied on a smaller tax base and revenues will fall. Provincial CIT revenues in 2011 were $22.034b, which works out to an average provincial CIT rate of 11.04%.
The policy experiment is to change the combined CIT rate, see what effect it has on the tax base, apply the (new) federal rate and the (old) provincial rates and compare the result to the base case. I've also done the exercise for the Mintz-Smart specification, using a net-of-tax elasticity of 2.5 - the low end of their range, and not far from what Dalhby-Ferede find.
Change in federal revenues Change in provincial revenues Static analysis Dynamic scoring:
Riedl and Rocha-Akis
Mintz and Smart
Static analysis Dynamic scoring:
Riedl and Rocha-Akis
Mintz and Smart
15.0% -$3b -$1.80b -$1.42b 0 +$0.88b +$1.15b 15.5% -$2b -$1.18b -$0.92b 0 +$0.58b +$0.76b 16.0% -$1b -$0.59b -$0.45b 0 +$0.28b +$0.38b 16.5% 0 0 0 0 0 0 17.0% +$1b +$0.57b +$0.41b 0 -$0.28b -$0.38b 17.5% +$2b +$1.13b +$0.80b 0 -$0.54b -$0.76b 18.0% +$3b +$1.68b +$1.16b 0 -$0.80b -$1.13b 18.5% +$4b +$2.22b +$1.49b 0 -$1.06b -$1.49b 19.0% +$5b +$2.75b +$1.80b 0 -$1.30b -$1.86b 19.5% +$6b +$3.27b +$2.08b 0 -$1.54b -$2.21b 20.0% +$7b +$3.78b +$2.33b 0 -$1.77b -$2.57b 20.5% +$8b +$4.28b +$2.56b 0 -$2.00b -$2.92b 21.0% +$9b +$4.77b +$2.77b 0 -$2.22b -$3.27b 21.5% +$10b +$5.25b +$2.95b 0 -$2.43b -$3.61b 22.0% +$11b +$5.72b +$3.11b 0 -$2.64b -$3.95b 22.5% +$12b +$6.19b +$3.24b 0 -$2.84b -$4.29b 23.0% +$13b +$6.64b +$3.35b 0 -$3.04b -$4.62b 23.5% +$14b +$7.09b +$3.44b 0 -$3.22b -$4.95b 24.0% +$15b +$7.54b +$3.51b 0 -$3.42b -$5.27b
People who are using static analysis to make their revenue projections are overestimating the revenue gains from increasing corporate taxes by a factor of at least two in the short run, and by even more in the longer term.