The economics of tax incidence: paying the tax is not the same as bearing the burden

One of the more important things that distinguishes economists from non-economists is a familiarity with the notion of tax incidence. The statutory incidence of a tax (who sends the cheque to the Receiver-General?) is usually very different from its economic incidence (who is out of pocket?).

The basic intuition is simple enough. We all understand that if the government chooses to impose a tax on gasoline retailers of $0.50 per litre, customers can expect to see a similar increase in gas prices. Even if the statutory incidence falls on the sellers, the economic incidence is borne by the consumers.

The question of who ultimately bears the burden of the tax is almost entirely separate from the question of statutory incidence. (There's even a pejorative term - the 'flypaper theory' - for the claim that taxes stick to those who are first touched by it.) So what does determine the economic incidence of a tax?

The key determinant turns out to be the relative elasticities of demand and supply for the the good that is being taxed. If supply (demand) is inelastic, the quantity supplied (demanded) varies only slightly as the market price changes. The extreme case is an elasticity of zero, where the quantity supplied (demanded) is fixed, and the price is determined by the other side of the market. If supply (demand) is elastic, the quantity supplied (demanded) is very sensitive to changes in the market price. The extreme case is of infinite elasticity, where the quantity supplied (demanded) is indeterminate at a given price, and the quantity is determined by the other side of the market.

Let's look at the case where supply is relatively more elastic than demand. In terms of the standard graph of supply and demand, this means that the supply curve is relatively flat, while the demand curve is relatively steep. Suppose now that the government is considering a tax of τ:

• If the tax is levied on sellers, the supply curve shifts up by an amount equal to the tax, so that the after-tax supply curve remains the same.
• If the tax is levied on purchasers, the demand curve shifts down by an amount equal to the tax, so that the after-tax demand curve remains at the pre-tax level.

Here is a graph of the two cases:

In both graphs, the vertical axis is the selling price (the choice is arbitrary, and doesn't affect the conclusion). If the tax is paid on the supply side, the supply curve shifts from S₀ to S₁, the price increases from P₀ to P₁, and the quantity falls from Q₀ to Q₁. At Q₁, the difference between the new selling price and the amount received by the supplier after taxes is τ, the amount of the tax.

The yellow region corresponds to the amount of taxes paid by the consumer (in the form of a higher selling price), and the green rectangle is the tax paid by the sellers (in the form of a lower after-tax price). Given the relative slopes of the supply and demand curves, the change in the selling price is almost as large as the tax itself, and almost all of the tax is ultimately borne by the consumer, even though the statutory incidence is on the supply side.

Suppose now that the tax is levied on those who purchase the good. In this case, the supply curve is unaffected, but since consumers must pay the tax on top of the selling price, the demand curve shifts down by an amount equal to the tax. The demand curve shifts from D₀ to D₁, the price decreases from P₀ to P₁, and the quantity falls from Q₀ to Q₁. At Q₁, the difference between the new price and the amount paid by the consumer is τ. Once again, the green rectangle is the tax paid by the sellers, in the form of a lower selling price. The consumer benefits from this lower price but must still pay the tax, so the net payment is the yellow rectangle.

It should be clear by now that in terms of tax incidence, both graphs tell the same story. Since supply is relatively more elastic than demand, the burden of the tax falls relatively more on consumers, and relatively less on sellers. In the extreme cases of a perfectly inelastic (i.e., vertical) demand curve or a perfectly elastic (horizontal) supply curve, the entire burden is borne by the demand side regardless of where the tax is applied.

Now let's look at the case where demand is relatively more elastic than supply. The story is the same as above, but changing the slopes of the curves alters the relative sizes of the yellow and green rectangles:

Since demand is relatively more elastic than supply, the burden of the tax falls relatively more on sellers, and relatively less on consumers. In the extreme cases of a perfectly inelastic (vertical) supply curve or a perfectly elastic (horizontal) demand curve, the entire burden is borne by the supply side regardless of where the tax is applied.

Here are two cases where this distinction has important policy implications:

1. Corporate taxes. I've gone through this point several times (most recently here) and I've compiled a reading list on the topic over here. If you assume that we are in a small open economy where capital flows freely, the supply of capital is relatively elastic. This corresponds to the first set of graphs, and so the result is that very little of the burden of corporate taxes falls on capitalists. Those who claim that owners of capital bear the entire burden are implicitly assuming - as did Harberger (1962) - that the supply of capital is perfectly inelastic. For large countries such as the US, the supply of capital is less than perfectly elastic, so the applicability of small open economy results can be problematic. But the empirical evidence for small open economies is pretty clear: the burden of corporate taxes falls mainly on workers.
2. Payroll taxes. These include employer contributions to EI and C/QPP as well as Worker's Compensation premiums. But as a HRCD survey notes, long-run labour demand is more elastic than labour supply, so the ultimate effect of payroll taxes is to reduce wages: "labour's share of the payroll tax burden in the long run is in the range of 87 to 100 percent."

If you're concerned about distributional issues, then this is an important concept. It's a mistake to go from noting that the statutory incidence is on a certain group of people to concluding that these are also the same people who actually pay the tax. Unfortunately, it's also a common mistake.

Comments

[Stephan Wehner]

It seems to me that for the "supply is relatively more elastic than demand" case, where the seller can pass the tax on to the consumer, the seller could have simply raised the price instead of waiting for the tax to be installed.

This should even work when there is (so-called) competition.

So I think your analysis for this case may not be complete: the price before the tax already incorporates the possibility of the tax.

[Min]

When I was a young man in Japan, the gov't increased the tax on alcoholic beverages sold by the drink. Immediately the bars and restaurants increased the prices of those drinks by around three times the tax increase. ;)

[Peter T]

I think that, as usual, the theoretical analysis contains a number of hidden assumptions. The most obvious one is that the payer of the tax is able to pass on the tax to consumers where elasticity of supply permits. There can be all sorts of constraints on this, ranging from risk of social or regulatory retaliation to ability to offset in other areas to various forms of corporatism. Before I bought the analysis (which suggests that all taxes on the rich will end up on the poor), I would like to see the results of extended case studies over time, taking into account other changes.

[slantendicular]

Does it follow that if the incidence of corporate taxes falls on workers, then reducing corporate taxes will cause the wages of workers to rise? If so, then have the significant corporate tax reductions in Canada over the past decade shown up in the wages paid to workers? How have they shown up, empirically?

[Mike Moffatt]

"Before I bought the analysis (which suggests that all taxes on the rich will end up on the poor)"

Keep in mind this isn't what Prof. Gordon is showing - just the incidence of *some* taxes (and to be fair, the incidence isn't being placed on 'the poor', rather it's being placed on 'labour').

Anyhow, a terrific post. I've been meaning to do something like this for awhile.

[Just visiting from Macleans]

Which graphs apply to Canada's banks, and the oil sands industry?

Do big sector fish in a small economic pond make the empirical evidence for small open economies is pretty less entirely clear?

[beezer]

Not an economist, but the post is clear enough. That said, pricing power comes from a wide array of circumstances that, in my opinion, may not be that susceptible to supply and demand.

Right now we have a decided lack of overall demand. So it would follow that in many cases pricing power is compromised. Yet corporate profits are at historic highs matched by historic highs in cash on hand. Someone obviously still retains a reasonable amount of pricing power.

Corporations cut supply, and employment, to match the reduced demand. Yet the profit figures indicate pricing was maintained. My questions is, what other factors might be influencing the pricing power maintenance?

[Luis Enrique]

How does the analysis of tax incidence, when the tax in question is a land value tax? The influential Martin Wolf is agitating for land taxes.

Here's what one campaign website* has to say:

IMPOSSIBLE TO PASS ON IN HIGHER PRICES, LOWER WAGES OR HIGHER RENTS. Competition makes it impossible for a business producing goods on a valuable site to charge more per item than one producing similar goods on less valuable land - after all, producers and traders at different locations are paying different rents to landlords now, yet like goods generally sell for much the same price and employers pay their workers comparable wages. The tax cannot be passed on to a tenant who is already paying the full market rent.

what do you make of that?

* the economic literacy of which I can't vouch for

[anon]

Luis Enrique, land taxes are capitalized in the value of land, so yes they are simply paid by existing landowners. But the main selling point of the land tax is that if implemented correctly it does not distort economic activity like other forms of taxation do, so a LVT which was implemented as a tax shift might even _increase_ asset values by removing the excess burden which currently falls on real estate.

Incidentally, this reveals a gap in Gordon's reasoning about corporate taxation. Even in a small open economy, some of the incidence of corporate taxes might be capitalized in the value of firm-specific assets (including but not limited to commercial real estate). The tax would effectively fall on existing stockholders, but the elasticity of capital supply would be irrelevant.

[Frances Woolley]

Stephen - I wonder if there's a different way of labelling your figures that clarifies the differences between them.

The ones on the left hand side are showing things from the demander's point of view: their demand curve hasn't changed, but suddenly firms are charging them more for any given amount they want to buy. Another way of saying this: in the left hand side graphs, the place where the supply and demand curve cross is the *before tax* (e.g. including HST) price.

The ones of the right hand side show things from the suppliers point of view: their before-tax cost structure hasn't changed, and the price they need to receive after tax to make it worthwhile supplying goods hasn't changed, but suddenly, because of the tax, people aren't willing to pay as high an after-tax price for their goods. So on the right hand side graphs, the intersection of supply and demand is the *after tax* price.

I find the idea of a selling price confusing - when I buy something, the receipt shows a before-tax price, an amount of tax, and an after-tax total. So what's the selling price? How many people actually know what the legal statutory incidence of HST or CPP premiums etc is anyways?

[Luis Enrique]

thanks anon - sorry to be dim, but what does "land taxes are capitalized in the value of land" mean?

(I'm using to thinking "capitalized" is to do with accounting treatment of capital expenditure - amortization in P&L - I can't figure how that relates to land taxes)

[Stephen Gordon]

Frances - I was trying to come up with a way of distinguishing between the case where the tax is on one side and on the other. The idea behind the graph is that if the purchaser pays, she does so on her way out of the store at some sort of tax kiosk. Probably not the best choice.

[anon]

Luis Enrique, the value of land is generally understood in economics as the net present value of the stream of all future rents (or the unobserved equivalent of "rents" if the real estate is not rented out). A land value tax is effectively a tax on future rents, so the sale price of land is lowered accordingly.

This is actually a significant problem with land value taxation: its incidence is felt immediately in lower asset values for landowners (leading to lower apparent wealth), but the benefits occur over time and are far less concentrated. This means that land taxes have to be implemented gradually and as a tax shift from distortionary forms of taxation, otherwise they are politically indistinguishable from a simple taking or confiscation of property.

[Luis Enrique]

thanks v much anon.

I'm not sure I understand why landlords would not increase rents upon the announcement of an LVT. If they tried to, the gradient between rents in high and low LVT areas would increase, so you'd predict some (but not all) tenants to move. So I wouldn't expect landlords to be able to pass on the whole LVT in higher rents, but why not some? Also, if tenant did move to low LVT areas, that'd increase value of land in these areas.

[anon]

Luis Enrique, I'm not sure what your point is, but it is assumed that landowners are charging as much rent as the market will bear, both in high- and low-LVT areas. This makes sense because (1) due to competition, no single landowner can charge more than what the market will bear. (2) land does not have a "cost of production" which equates to the rent being paid: its supply is entirely inelastic.

[Luis Enrique]

My point is to doubt the argument that because landowners are charging as much rent as the market will bear they cannot partially pass on LVT. Rather than think about single landowners, think about all landowners making a simultaneous price change. The announcement of an LVT sounds like a good coordinating device to me. Put it this way, if an LVT was announced, how much would you bet on zero impact on rents?

[anon]

"Rather than think about single landowners, think about all landowners making a simultaneous price change."

You're assuming that landowners can collude to raise rents above the market price and keep land out of use. It may be true that the announcement of a LVT is a feasible coordination device, but collusion would break down in the long run (disregarding special cases such as LVT being higher than the available rent, which would result in the abandonment of property rights by landowners) Generally speaking, LVT tends to encourage effective land use compared to no tax.

[Bill]

A mixed bag, but worth a read ...

http://economix.blogs.nytimes.com/2010/07/23/who-ultimately-pays-the-corporate-income-tax/

[Mike Moffatt]

All that being said, I agree with Krugman in that Fisher's recent speech *is* depressing. It's one thing to make a mistake (IMO a justifiable one), but it's another thing not to learn from it.

[Marc]

I think Irwin Gillespie got this one right when he argued that owners pay corporate income tax up to some global (or in Canada's case, US) tax rate, and any excess is shifted to consumers or workers. So as an overall policy issue, with Canada's corporate tax rates lower than in the US, there is no reason to believe that CIT gets shifted to workers.

Even if you don't buy that, there are likely to be some differences across industries: some industries are regulated in terms of price or actual rate of return; most service industries are not really mobile in the sense that they need to have market presence in order to make any money; and others need access to resources -- in the case of tar sands there are very few places companies can invest money.

So really this is about a handful of industries, mostly in manufacturing, where the open economy metaphor is appropriate. But note that well before recent CIT cuts, Canada had a lower rate for manufacturing and processing, as compared to service industries.

My default would thus be to assume that CIT is paid by owners and is thus is a progressive tax.

[Stephen Gordon]

Reference?

The second and third paragraphs are wrong. The argument is based on the openness of capital markets.

[Marc]

Disagree. Ultimately, this issue depends on how one views how companies make investment decisions. That is, they invest based on the hope they will make lots of money, above and beyond their costs, but overall this is a highly uncertain process. If they make money, they pay tax on it. Along the way they try to minimize their labour costs, and charge as much as they can for the good or service being produced. End of story.

That's why you can only cite a handful of flawed empirical studies to back your case.

For the Gillespie reference, you'll be pleased to know that when I googled it, I came up with a WCI blog post!
http://worthwhile.typepad.com/worthwhile_canadian_initi/2006/08/canada_already_.html

[Stephen Gordon]

No, the issue is where the savings comes from to finance those investment decisions.

And in the reference, Gillespie and his co-author *assume* that the CIT is borne by owners of capital. It's not a result derived from a set of explicitly-stated assumptions or supported by data.

[Just visiting from Macleans]

It seems to me Marc is questioning the assumptions - not the theory. I agree with the points he raises.

[Marc]

Intuitively, I'd agree that payroll taxes are shifted to labour as companies look at their total wage bill inclusive of payroll tax. I'd also argue that carbon taxes or sales taxes on inputs are shifted to consumers as these are a cost of doing business. And business price as a mark-up over average costs.

It is just the CIT shifting that I have some quibbles with. I don't think Stephen has made a compelling counter-argument to the intuition Gillespie brought to this many decades ago, especially in the Canadian case.

Here's a contrary bit of data that disproves Stephen's whole point: over the past decade we have seen major corporate tax cuts, but at the same time a rising profit share (and decreasing wage share) of GDP. In Stephen's theory, CIT cuts should be passed on to workers. But no. Why? Corporations try to make as much money as they can.

Open capital markets? Capital is what capital does: invest in order to make a profit, on which they then pay CIT. Like I said, it is a simple story and economists do a dis-service when they create complicated stories like marginal effective tax rates and shifting of CIT onto labour.

By the way, I did a Canadian tax incidence study that follows Gillespie and updates tax incidence up to 2005:
http://www.policyalternatives.ca/publications/commentary/putting-fairness-back-canada%E2%80%99s-tax-system

[Stephen Gordon]

Here's a contrary bit of data that disproves Stephen's whole point: over the past decade we have seen major corporate tax cuts, but at the same time a rising profit share (and decreasing wage share) of GDP.

The theory doesn't say anything about income shares, even conditioning on everything else. Wages should increase, but so should investment and the capital stock. The net effect on income shares is indeterminate.