Back-of-Envelope Math on R&D

To flesh out the cost-effectiveness issue outlined below, consider the following figures.

McKenzie estimates that a 10% decrease in the cost of R&D due to a tax credit increases R&D by 2% in the short term and 7% in the long term, but that a 10% decrease in the effective tax rate on production increases R&D by 3% in the short term and 9% in the long term. At first glance, it seems that a general tax reduction would be more effective than a tax credit.

However, in 2003, Canada conducted $22.5 billion of R&D and Canadian governments collected $363.0 billion of taxes. Therefore, reducing the cost of all R&D by 10% would cost $2.3 billion per year, or $2.4 billion after the consequent long-term increase. Reducing overall taxes by 10% would cost $36.3 billion per year. Perhaps, by concentrating on reducing taxes that are most burdensome to business, this figure could be lowered to $12.1 billion.

Put another way, $1 billion of incentives would lower the cost of R&D by 4%, increasing R&D by an estimated 3% in the long term. However, $1 billion of business-tax cuts would reduce the effective tax rate by 0.8% at most, increasing R&D by an estimated 0.7% in the long term. Even if one accepts McKenzie’s estimates and assumes that business taxes can be cut cheaply, a dollar put into tax credits would have four times the effect on R&D as one put into tax cuts.

One comment

  • Any reason to think that more R&D spending will increase productivity?

    To my mind, that’s the only real reason for thinking that increasing R&D incentives would be a good idea. And the case has not yet been made: we’ve seen too many R&D boondogles whose only real purpose has been to provide an excuse for firms to play silly games in order to get free money from govts who wish to appear to be “doing something”.

    Clearly, *someone* has to do the R&D. But it’s by no means obvious that the gains to R&D will be captured by those who made the initial investment.

Leave a Reply

Your email address will not be published.