Tax policy colloquium, week 12: Emily Satterthwaite on VAT exemptions for small businesses

This past Tuesday at the colloquium, Emily Satterthwaite presented Electing Into a Value-Added Tax: Survey Evidence From Ontario Micro-Entrepreneurs. This interesting empirical work has both a quantitative and a qualitative dimension, derived from surveying small suppliers in various lines of business at Toronto-area farmer's markets. (Now there is some empirical research that I'd actually like to do - going to farmer's markets, which I do intensively anyway from spring through fall, at least in years when there actually is a spring.)  The subjects were people who are not required to register as businesses under Canada's VAT, because their annual gross receipts are less than $30,000 in Canadian dollars ($23,000 US). But they are allowed to register voluntarily if they wish.

The research sheds light on the design question of how high or low small-business VAT exemptions should generally be. In addition, micro-entrepreneurs' behavior (and expressed attitudes or knowledge) around elective VAT participation may also be more generally illuminating, both about VATs and, more generally, this sector of the economy.

1. My priors on high vs. low mandatory VAT registration thresholds
In practice, VAT small-business mandatory registration thresholds vary quite significantly. I gather that there is no small business exemption in Sweden - if you have $1 of relevant sales, you are supposed to file. In Canada, as noted above, the threshold currently stands at about $23,000 in US dollars, and is trending down annually, since the nominal amount hasn't been changed in more than 20 years and isn't indexed to inflation. In the UK, by contrast, the threshold for mandatory VAT participation exceeds $100,000 in US dollars.

While I haven't previously thought much about whether VAT mandatory registration thresholds should be low or high, I come equipped with attitudes (which I am of course quite willing to reexamine) suggesting that one would want to aim towards the low end.

Now admittedly, in favor of a relative high registration level are the points that:

(a) The social value of accurately measuring and collecting each dollar of correctly determined tax revenue is generally much less than a dollar. A payment of tax is a transfer, so the dollar is just moving from one dollar to another. Getting it right is obviously worth something - presumably, in efficiency and/or distributional terms - or else we'd just have a lump sum tax of some kind, but the marginal value of correctness is presumably just some fraction of the full dollar. This of course is standard Kaplow et al.

(b) Small businesses are likely to have higher marginal compliance costs per dollar of revenue collected than big ones; also the marginal administrative costs of auditing them may be relatively high. So one might have to climb up the scale a bit before it's worth it.

But there are also a bunch of reasons or arguments for wanting to aim low. For example:

(a) VAT exemption amounts generally function as a notch or a cliff - unlike, say, income tax exemption amounts. E.g., if you're one dollar under the VAT registration ceiling, you don't have to collect any VAT from your customers. But once you hit the ceiling you have to collect it all, from the first dollar onwards. One of the students in the class found this great article about problems that this has been causing in the UK. Setting the threshold high tends to result in a bigger notch, and under the Sweden approach there would be no notch. The notch literature suggests that they're generally bad as a design matter, unless the notch occurs at a low point in a multimodal distribution. Not clear how or why one would find such a thing in small business size, however.

(b) VAT exemptions can in effect create a tax preference for small business, inefficiently steering consumer demand towards them and inducing them to stay under the threshold. If you want a comprehensive and relatively neutral tax base, significant exemption thresholds will be at least a matter of regret.

(c) Consider again the point that small businesses are associated with higher marginal compliance and administrative costs. I noted above the possible conclusion that this may support exempting them from the tax. But suppose we look at it the other way around. Small businesses generate negative externalities if they're exempted. If it's better to have a comprehensive system with not just tax payments but information reporting that extends as broadly as possible, then one may think of the small businesses as imposing disproportionate costs on the system, rather than the system as imposing disproportionate costs on them. Or one may adopt a Coasean joint causation perspective, a la the railway and the hay fields.

Again, my hunch from all this tended to come down on the side of setting thresholds low rather than high. But on the other hand there's a paper by Michael Keen and Jack Mintz, modeling the broader social welfare effects (but in light, I suspect, of the authors' considerable empirical knowledge), that suggests it may often be optimal to set the threshold relatively high. I tend to have a very high regard for those two individuals' work, so that does move the needle for me a bit.

2. When do or should small suppliers voluntarily register to participate in the VAT?
Again, Canada allows small suppliers voluntarily to register for VAT participation, and the paper's main contribution is exploring when and (in their own stated terms) why they choose to do this or not.

But for starters, what one should think of voluntary registration? We tend to think of choice as good, especially where the state benefits from more people participating (so there presumably is no downside if they voluntarily opt in), unless one is especially concerned about the cost of having to choose. With respect to tax elections in particular, however, it's often the case that (i) electivity is good if people are using it mainly to lower their compliance and planning costs, but (ii) it's likely to be bad if they're using it to lower their tax liabilities, since the value of the $$ to the government is an externality from their standpoint and it's unclear why this filter would relate closely to whom we want to bear higher vs. lower taxes.

But anyway, when should we expect people to opt into the Canadian VAT? Financially, it tends to have both an upside and a downside. The upside is that one need not charge the VAT on sales directly to consumers. The downside is that VAT-registered businesses that sell directly to you will still charge the VAT, but you won't get it refunded. This is especially disadvantageous if you then sell to another VAT-registered business, in which case, the ultimate downstream VAT collected ends up being higher than if all were registered, as there is unreimbursed cascading for the liability charged on your mid-stream purchase.

The paper's empirical findings are roughly consistent with this. It finds no significant effect on the upstream side (i.e., whether a given farmer's market micro-entrepreneur purchased inputs from VAT-registered businesses), but it does find significant effects on the downstream side (i.e., whether one sells directly to consumers, discouraging registration; or to other VAT-registered businesses, potentially encouraging it).

There is also some indication that informal considerations may matter. E.g., registering or not might involve either signaling or communicating type, although the alternative theories that might apply here are numerous. These might also feed back into influencing the normative analysis.