Wednesday, 9 June 2010

Tax incidence

Matt Nolan rightly notes that tax incidence in general equilibrium after prices have adjusted is far more complicated than just "this guy gets $25 more per week after a tax cut while this guy only gets $0.50" linear estimates from the initial change in tax rates.

That's true, of course, but the problem is bigger.

Matt starts by assuming that folks kinda get the scissors of supply and demand and that relative price elasticity determines tax incidence: statutory incidence is irrelevant. For simplicity, imagine a tax charged on the seller of a good that is perfectly elastic in supply and perfectly inelastic in demand. It currently sells for $5. The government says they'll charge the seller a $1 per unit tax. Voters love it because the sellers are big business, and everybody hates big business. Had it been a tax levied at point of sale on consumers, they'd have wailed and gnashed teeth because that's a tax on the poor customers who buy the goods. Of course, in this example, the price paid by consumers jumps to $6 regardless of whether the statutory incidence falls on consumers or producers. It's relative elasticity that decides who bears the incidence of the tax, not statutory incidence. And the inelastic side of the market bears the burden.

But voters are utterly incompetent at understanding this.

And here's the nicest bit of evidence on it from one of my very favorite experimental economics pieces of all time. Rupert Sausgruber and Jean-Robert Tyran's "Testing the Mill Hypothesis of Fiscal Illusion". I'd pointed to it in this prior post, but buried amongst other things.

Here's the experiment. Put a bunch of folks in a computer lab. Tell them they're buyers of good X. Each of them values good X at different amounts: the experimenter tells each buyer that he'll buy X from them at different prices for different quantities. That generates individual and aggregate demand schedules. Put in a computerized seller with a set supply curve. Let folks run a few rounds of trading with the computerized seller.

Then, let the buyers take a vote on an inefficient redistribution mechanism. If the resolution passes, a tax will be imposed on the buyers, per unit sold. The collected money will be divided evenly among buyers, but with some of it taken off the top as waste. Surprise surprise, nobody much wants that mechanism. Then, reframe it as a tax on the seller. Let them vote again. And, the buyers love it. Never mind that the incidence of the tax is identical across the two treatments. Indeed, in their design, the incidence falls entirely on consumers.

And that's why politicians love to use statutory incidence to frame things for voters. ACC: that's paid for by employer contributions as well as employee. Employees, you don't have to worry about the part your bosses pay. Corporate taxes: who likes companies other than hard-hearted economists? Producer taxes. When Canada shifted from the Manufacturer's Sales Tax to the GST, lots of idiots complained because where companies previously paid the tax, NOW the tax was going to be paid by consumers; Canada wound up with a pretty dirty GST, but still better than the prior manufacturer's sales tax.

In the Tyran and Sausgrubber experiment, voters eventually figured out that they bore the incidence and eventually voted against the inefficient redistribution mechanism. But that's in an experimental setting where the only things changing in the world are whether the scheme is in place or not. Under those conditions, it's pretty easy for the traders to see that the policy makes them worse off. But the real world is noisy, and there are always a hundred other things that voters can blame for the effects of their preferred policy. Indeed, they may wind up arguing that the business was evil for putting prices up when the tax was implemented and that regulations are needed to ensure that the tax works as it was "meant to".

Always ALWAYS beware of taxes that purport to fall on disfavoured groups. The economic incidence may not be what you're hoping it'll be. And that's even before all of the GE complications come in.

I teach this in my third year public choice class at Canterbury. Third year students, almost all of them econ majors. And at least half of them don't understand tax incidence before I show it to them prior to showing the Sausgruber and Tyran result. If folks almost finished an econ major don't get it, there's absolutely no hope in expecting voters to get it. And, of course, politicians well know that only statutory incidence matters to most voters, and so we get the policies we get.


  1. Hi Eric,

    Very interesting. I'd just like to say that my post didn't presume that people understood tax incidence - it just placed it as the first wrinkle in the common talk about tax cuts before the issue of GE raised its ugly head.

    I do find it hard to believe that people don't see incidence though - it is just such a fundamental point :/

  2. @Matt: you'd read my posts from 2009 on just what voters fail to understand about economics, right?

    That first wrinkle is effectively a 50' wall voters must hurdle before they even start thinking about the crocodile-infested waters of GE on their way to the ballot box to cast an informed vote. Or, they can just take the shortcut path of voting for what feels good. Hmmm.

  3. I'm a little reluctant to post this given the above comments re public lack of understanding, but as I am a member of the great unwashed I hope you'll forgive me if I as a stupid question, so here goes...

    One question Eric, when Canada shifted to GST were retail prices increased by roughly the % of GST being imposed, as they were here? If this is the case then surely the consumer is worse off. The manufacturer is better off, the tax burden is shifted to the consumer but the manufacturer doesn't reduce the cost of goods to the retailer, the retailer is no worse off by imposing an extra % on the cost of the goods to cover the cost of GST. Assuming the tax loading has stayed static, but just shifted to the consumer, then there should be no need to raise prices if the manufacturer is content to run at the same pre-gst margin and reduces the wholesale cost to the retailer. So essentially the consumer loses out because of manufacturer opportunism.

    Or am I missing some fundamental economic principle here? <- quite likely!! :)

    Having said all this, I tend to favour GST as a means of revenue take, it seems much simpler and more transparent than less direct tax methods.

  4. Canada replaced a 12.5% Manufacturer's sales tax on a base that included exports with a GST of 7% on a base that excluded exports. Consumer prices dropped about as much as you'd expect from the difference between the two rates.

    Simple example. Suppose that a product can be produced at a cost of $1 in a perfectly competitive market where supply is perfectly elastic. A 12.5% manufacturer's sales tax is levied and paid by manufacturers. The price of the product, paid by consumers, is $1.125, $1 of which goes to the supplier. Now we switch to a GST of 12.5%. The producer sells the product for $1 and receives $1, but the consumer pays an additional $0.125 at the till as tax to the government. Suppose one manufacturer figures he won't reduce the cost of goods to the retailer. The retailer shifts to somebody else who'll supply the product instead. Competition works. Now, take the opposite case: suppose the supplier has monopoly power. If that's the case, and if he could just fail to reduce the price to the wholesaler, that means that he was a very poor monopolist before the tax change because he could have increased prices charged to the wholesaler prior to the change. If he has monopoly power, he'd be charging as much as he could before and after the tax change. So prices still have to drop.

    Try the wikipedia page on tax incidence for starts, but if there are remaining questions, do ask.

  5. @Lats,

    You said that retail prices increased by the amount of the GST when it was introduced here. In fact, that is not the case. the CPI in the December quarter of 1986 (the first quarter with the GST) increased by 8.9%. This compares to an average of 2.6% over the previous four quarters. As a first pass, then, the 10% GST only pushed up prices by 6.3%, which is pretty close to the 5.5% net effect expected from taking the 20% MST off. Now add in the fact that people delayed their purchases of manufactured goods until after October 1, 1986, to take advantage of the tax reduction, forcing retailers to offer more agressive sales in the September quarter and one can assume that the net effect was consistent with assuming that 100% of the incidence of both taxes fell on consumers.

  6. Thanks Eric and Seamus, I spent a few minutes googling for retail price increases around 1986, but didn't immediately stumble across the data, so I had to rely on my admittedly poor recollections from the time, and we know how questionable our memories can be. In my defense, I did say roughly the rate of GST, and given that it was 24 years ago, I'd say a 6.3% increase can be called roughly 10% :)
    Did people really delay purchases until after October 1 though? I recall a rush of retail sales prior to the introduction of GST for fear of a possible price increase. According to my brief googling retailers were caught off guard and ordered in excess stock, and were forced to discount stock after October in order to get rid of excess. If that is true then discounting post October was only indirectly a result of GST.
    Also, I'm sure one of you will be able to answer this for me, did the introduction of GST allow the taxation of a host of transactions between supplier and consumer that were previously hidden from the tax take under MST? I'm thinking of things like services and utilities that aren't manufactured goods in the physical sense.
    No doubt this stuff is all high school economics to you lot, but I went down the science path instead, so while I can grasp the concepts easily enough, my knowledge of particulars is at best hazy.

  7. > I teach this in my third year public choice class at Canterbury. Third year students, almost all of them econ majors. And at least half of them don't understand tax incidence before I show it to them prior to showing the Sausgruber and Tyran result.

    That's shocking. I've never had any formal training in economics, but after reading this stuff on your blog I think I've got a reasonable understanding of the major issues. Only thing I'm having trouble grasping is the stuff on intra-family externalities; I'll leave a comment on one of those threads when I've thought it through a bit more. How do students get through to third year without understanding this?

    In other news, can you give any reccomendations for online introductory economics resources? I've been looking at Wikibooks and Wikiversity, but leaves a lot to be desired.

  8. Tax incidence comes up in the lower level classes, but it takes frequent repetition of some ideas to make them stick.

    Intrafamily effects are at best underexplored in the theoretical literature. I'm pretty convinced that a Coasean argument ought to apply, but I've had a hard time finding explicit literature on it.

    Online intro econ resources...hmm. I'll have to think about it. But a pretty decent way of getting a handle on the main things going on is to follow Marginal Revolution and EconLog from the pro-market side, Scott Sumner on Macroeconomics, Brad DeLong for the somewhat market skeptic side (ignoring the rabid, usually correct but tedious, anti-Republican posts).