A followup to my IRRP study: Provincial Taxation of High Incomes - The Effects on Progressivity and Tax Revenue

The revenue to be raised by a new tax bracket on high earners generated much debate during and after the federal election, given the Liberal plan to install a new federal tax bracket of 33% for those with taxable incomes over $200,000. With recent estimates coming from the C.D. Howe Institute’s Alexandre Laurin and from the Department of Finance suggesting revenues will be lower than the Liberal platform expected, it’s worth the time to review the evidence. But first we need to specify what we’re talking about.

When taxes go up, people might react in many ways. They could work less or invest less, for example. But, the recent focus of much of the economics debate on taxing high earners has focused on how high earners might respond through accounting maneuvers to avoid taxation. By shifting income off the personal tax form, a high-earner’s tax bill can be lowered. The degree of responsiveness is often characterized using the concept of an elasticity of taxable income (ETI).

Technically, the ETI measures the percent change in taxable income compared to the percentage change in the tax rate (actually the net-of-tax rate, to be precise). An elasticity of zero means there is no response, while an elasticity of 1.0 would imply a pretty large response. Technical derivations of the formulas can be found in my teaching notes here, or in the appendix of this IRPP book chapter I wrote with Michael Smart.

In applying the ETI concept to today’s policy questions, it is pivotal to understand that the ETI is not some fixed universal constant that can be applied to any situation. Instead, an ETI is a function of a given tax environment. Specifically, if a tax system is fairly ‘leaky’ so that it is easy for people to avoid taxes, then the ETI will be high. If the tax system affords few tax avoidance opportunities, the ETI will be low. This point was most effectively made by my former UBC colleague Wojciech Kopczuk in this academic article some years ago. What this means is that we can’t just apply any old estimate of an ETI to the current context.

What is a good pick for the value of the ETI in Canada? Alexandre Laurin provides a very useful review of recent estimates, with many centred between 0.6 and 0.7. But let me focus on my own estimate, from an about-to-be-published academic article co-authored with Michael Smart. We compare provinces through the period of 1988 to 2011 to see if the provinces who instituted higher rates see some shrinkage in the income reported in their jurisdiction. Using this inter-provincial variation, we have a central estimate of the ETI of 0.69.

The context here is very important–both the ‘1988 to 2011’ part and the ‘inter-provincial variation’ part. It’s pretty fair to argue that some things have changed since 2011. In particular, the residency of trusts has been tightened by court rulings, making it harder to shift income across borders using trusts. Moreover, our estimates are focused on provincial level tax changes. If part of the behavioural response we picked up comes from inter-provincial shifting, then we can’t just apply that ETI to the case of a federal tax change, since the federal government gets its money no matter if income is shifted across provincial borders or not.

In a draft paper, Michael Smart and I attempt to separate the ‘interprovincial shifting’ from other forms of response to higher tax rates. When we do that, our preferred elasticity shifts from 0.69 down to around 0.4. Is this large change in the federal vs. provincial ETI credible? We suspect that the expansion of the use of trusts over this period might explain a large degree of the inter-provincial elasticity. Through the 2000s, the use of Alberta Family Trusts expanded as wealthy Canadians in other provinces shifted assets to Alberta to take advantage of the 10% flat tax rate in Alberta.

So, if I were asked to predict what ETI to use to predict the impact of a new federal tax bracket for high earners in 2016, 0.4 would be my starting point.

But, it’s important to remember that ETI’s depend on the ease of tax avoidance. In their platform, the Liberals committed to several measures–ranging from making it harder to use small business corporations to avoid taxes to changing the taxation of stock options to greater efforts on administrative enforcement. I discussed these tax avoidance options in a piece for Maclean’s here.

My own view is that, if implemented successfully, these tax measures could substantially cut back on the scope for tax avoidance. So, from the starting point of 0.4, a forward-looking ETI for 2016 needs to be adjusted downward to the extent these tax avoidance measures are actually implemented. This was the basis for my conditional statement during the election campaign that the Liberal platform estimate of $2.8 billion of revenue from the new tax bracket was reasonable, if they implement serious anti-avoidance measures in the upcoming budget. Without those measures, a 0.4 ETI is most appropriate for a federal tax change, in my view. 

There are still many issues to consider. In my recent IRPP chapter with Michael Smart, we point out the difficulties of provincial governments using high-earner tax brackets to raise revenue. So, there are federal-provincial issues at play as well.

But more broadly, a progressive income tax is a vital part of our tax mix. Without the income tax, the rest of Canada’s tax system is likely regressive–meaning that high earners pay a lower proportion of their income in taxes than middle or low earners do. For that reason, a progressive income tax is necessary just to ensure that the overall tax system delivers the same proportionate rate across the income distribution. We need to continue the discussion on how best to design and improve our progressive income tax with that goal in mind.