The federal government is set to reduce corporate income tax (CIT) rates from 18% to 16.5% in January of 2011, and then again to 15% in 2012. The effects of these measures have been characterised in two ways:
- $6b per year in tax revenues. The Liberals use this number when they explain that they are choosing to use $6b/yr to finance social programs instead of corporate tax cuts, and (for reasons discussed below) the Conservatives don't seem to be disputing it.
- 233,000 jobs. The Conservatives are using this number to describe the costs of reversing the CIT cuts.
Although I wouldn't say that these numbers are meaningless, they should be put in proper context.
The problem is that these results are obtained by static analysis, where it is assumed that the tax change has no implications for behaviour: the only effect of a CIT change is to alter the numbers on the cheques sent to the Receiver-General. This is an odd assumption to make here, since the main objective of a CIT reduction is to induce a behavioural response, namely, increased investment and income.
One of the services provided by the US Congressional Budget Office is dynamic scoring, which incorporates behavioural responses into estimates for the effects of the budget balance. Our own Parliamentary Budget Office still doesn't have the resources required to do this properly, but it seems to me as though the true effect of the CIT cuts will be much less than $6b.
The only study I've been able to track down that touches on the issue is this 2004 paper from the Department of Finance. It's a calibration exercise, so we should interpret the results as properties of the model and not necessarily of the Canadian economy. Notwithstanding, the suggestion that the effects on the budget balance decline by more than a half as the effects of the CIT cut on investment and output make themselves felt seems plausible.
Here's a graph of federal CIT rates and the revenues they've generated:
Profits are a very volatile series and are also highly cyclical, so there's no reason to expect a tight link between short-term changes in CIT rates and movements in CIT revenues. But if you were making policy based on an assumption that CIT revenues were proportional to CIT tax rates, then you might want to see some sort of downward trend during a thirty-year period in which CIT rates were reduced by half.
(Note that the link between GST rates and GST revenues is remarkably stable: GST revenues were a roughly constant share of GDP before the GST rate cuts.)
It may be the case that the CIT cuts will cost $6b in the very short run, but the longer-term effects will be much less. It is a mistake to think that the effects of a 3 percentage point decrease in the corporate income tax will have the same effect on the federal budget balance as $6b/yr in new spending.
233,000: Characterising the effects of the CIT in terms of employment is one of those things that seems to make sense to everyone but economists. Nick might call this an exercise in applied orthogonality, namely, the insistence on using one dimension - employment - to evaluate policy:
I think we sometimes get fixated on a certain dimension, and analyse all policy questions in terms of how they look on that dimension, even when they matter on some other dimension that is orthogonal to the first. Involuntary unemployment is a problem. It's a dimension that matters. But reducing involuntary unemployment is not always the same as increasing employment. And some policies are good policies for reasons that have nothing to do with reducing involuntary unemployment, or even increasing employment. I can't think of any reason why the HST should have any effect in either direction on involuntary unemployment.
Short-run macro policy is about jobs. So is the minimum wage. So are unions.
The HST is not about jobs. Nor is free trade. Nor is environmental policy like carbon taxes. Most economic policies aren't about jobs.
To that list I would add corporate taxes.
The number itself comes from some back-of-the-envelope calculations in this study by Jack Mintz and Duanjie Chen:
Canada’s planned reductions in federal and provincial taxes on capital investment over the next several years will reduce this country’s METR to a level that should be competitive with these rapidly growing economies, enhancing Canada’s ability to attract capital investment in a volatile global economy. For instance, the three-point reduction in the federal corporate income tax will reduce the METR from 21.4% to 18.9% by 2013. Based on typical empirical estimates, a 10% reduction in the tax-inclusive cost of capital would increase the demand for capital stock by 7% in the long run. This would imply that the planned federal corporate rate cut would increase Canada’s capital stock by $49 billion and labour demand by 233,000 jobs in the long run.
The empirical estimates referred to are from this Department of Finance study.
Of all the lessons to take from the Mintz-Chen paper, the 233,000 number is almost certainly the least important, and is orthogonal - that word again - to the rest of the article. Indeed, it only makes sense if you assume that the increase in labour demand doesn't affect wages, and is completely described by an increase in employment. If the base is the 17.1m employed in May 2010, then we're talking about an increase of 1.4%.
But that isn't what will happen: the labour supply curve is much steeper than what this sort of analysis would suggest. And if workers take advantage of higher wages by working less, it may even be backward-bending. If employment isn't affected, then the long-run effect of a three percentage-point reduction in the CIT is to increase wages by 1.4%. Hardly earth-shattering, but you wouldn't leave it on the table.
If the CIT cuts go through and are later rescinded, we will not see a recession in which some 233,000 people lose their jobs. Instead, long-run incomes will be about 1.4% lower than they otherwise would have been.
I think rescinding the CIT cut would be a mistake, and that it will cost less than the advertised $6b figure. But it wouldn't be the worst mistake we've seen in the last five years.