New taxes are usually inefficient or unpopular - and that's a good thing.

My last post contained a diagram suggesting that no tax is both popular and efficient. On twitter, Matt Cowgill suggested the diagram was wrong. Taxes on resource rents are both popular and efficient. Michael Kushnir suggested carbon taxes as another possibility. If Cowgill and Kushnir are right, then the set of efficient taxes interesects the set of popular taxes:

   

No tax is completely non-distortionary. Bev Dahlby, for example, estimated that the cost of generating an extra dollar of government revenue, once efficiency impacts are taken into account, is $1.11 per dollar of federal GST revenue, but $40.83 [sic] per dollar raised by the BC corporate income tax. In terms of the diagram above, Dahlby's analysis suggests that sales taxes are in the green, efficient circle, but corporate income taxes are not.

Few taxes generate enthusiastic popular support, but some are more popular than others. Those are the ones that fill the red circle.

The area labelled "both" in the diagram above includes two types of taxes. The first is the tax that produces concrete, tangible benefits, but does not distort people's behaviour in undesirable ways. For example, carbon taxes are popular - polls (or, at least some polls) suggest that the majority of Canadians support them - because they produce benefits, in the form of a cleaner environment. To the extent that they distort people's behaviour, they do so in desirable ways.  Taxes introduced during wartime have also had widespread support (their efficiency may be more debatable). Their popularity was achieved, as this Disney propaganda film shows, by framing them as part of a patriotic war effort: "taxes will keep democracy on the march."

A second type of efficient and popular tax is one that the majority of people do not have to pay, and the minority cannot avoid. The incidence of cigarette taxes falls on the minority of Canadians who smoke, grow tobacco, or own shares in tobacco manufacturing companies. The tax is popular with the majority (for US poll numbers, see here, here, or here) because they do not have to pay it. Unless cigarette taxes are very high, they are also relatively efficient: because nicotine is so addictive, smokers often would rather pay the tax than change their behaviour. (To the extent that cigarette taxes induce behavioural change, this may be a good thing).

Taxes on resource rents are also, I would argue, a tax paid by a minority within the population - the owners of the corporations that extract those resources (Andrew Leach has a primer on Alberta oil tax royalties here). Why? As a general rule, the economic burden of a tax falls on whoever cannot change their behaviour to avoid it. A tax on call centres can be shifted to call centre employees, because call centres operations can make employees a take-it-or-leave-it-offer: "accept these wages or we'll move to a lower-tax jurisdiction." Resource extraction companies, however, have to stay where the resources are if they wish to keep on extracting them. They have few alternatives to just paying up.

Any tax that falls on the owners of corporations - despite the inevitable rhetoric about pension plans and seniors - is a tax paid by the minority. The ownership of wealth is highly concentrated - Jim Davies has estimated that, globally,  the top 10 percent of the world's population owns 71 percent of its wealth. Taxes on wealth, with the notable exception of property taxes, are popular because they take from the minority and give to the majority. In those cases where the source of wealth is immobile, they are efficient too.

In sum, there is a good case for saying that some taxes are both relatively efficient and fairly popular. 

The thing is, most governments can't raise enough revenue to cover their expenditures using only popular and efficient taxes. (They can't raise enough revenue to cover their expenditures, period. The federal government, and most provincial governments, are running deficits.)  

A democratically elected government that seeks to express the will of the people, and also to run a reasonably efficient economy, should fully exploit any taxes that are both popular and efficient before looking to any other source of revenue. If new taxes are either inefficient or unpopular, that's a good thing, because it probably means that the taxes that are both efficient and popular are already part of the government's tax base, as shown in the diagram above.

The only time that new taxes can be both efficient and popular is when the government is not fully exploiting that small set of taxes, as shown in the diagram below:

The question is: why would a government find itself in this situation? Why would it ever raise revenue from relatively inefficient and unpopular taxes when it could be raising revenue from efficient or popular taxes?

One possible explanation is that changing the tax system is costly, so taxes cannot and should not fluctuate in response to short-term changes in the economic and social environment. In the case of carbon taxes, for example, public opinion has changed over time. From a political point of view, it make make sense to ignore poll results suggesting that carbon taxes have wide-spread support if that support is perceived to be weak and easily eroded.

A second explanation is that, in Canada, it is possible to form a majority government with support from a minority of voters. It doesn't matter if only 40 percent of Canadians (say) oppose a carbon tax - if those 40 percent are the ones that have put the present government in office, that government is not about to turn around and introduce a carbon tax. 

A third explanation is that governments do the right thing by protecting the rights of a minority against a majority. Historically, Canada imposed a head tax on immigrants from China. The tax had widespread popular support. It might even, at low levels, have been efficient, in the sense of not distorting people's behaviour. (Impoverishing people isn't inefficient.) But popularity and efficiency are not the only criteria for judging taxes: the head tax was a bad tax because it violated horizontal equity, by discriminating against one group of people.

A final possibility is that the premise I began this post with - that introducing a carbon tax would be both popular and efficient - is wrong. There are various estimates of the social cost of carbon. The number used by the US government is rising to $36 per tonne of CO₂ emissions. The BC carbon tax is based on a price of $30 per tonne of CO₂ emissions, which works out to a tax of 6.67 cents per litre of gasoline.  Right now the average Canadian pays federal and provincial gasoline taxes of 25 cents per litre - over and above provincial and federal sales taxes on gasoline. These gasoline excise taxes are much higher than a carbon tax based on widely accepted estimates of the cost of CO₂ emissions would be. I don't particularly like this conclusion, but I can't, in all honesty, say that, given present levels of taxation on gasoline, and current estimates of the cost of CO₂, that I can make a strong efficiency case for a carbon tax. 

But that's a good thing. When any new tax is either inefficient or unpopular, it means that governments doing their job: deriving the maximum possible revenue from popular and efficient tax sources before tapping into unpopular or inefficient sources of tax reveue.

Comments

[Dan F]

Certain kinds of taxes, like carbon taxes, might be a popular solution... unless you have an organized and sustained public relations campaign against them by one political party and their supporting industries (as we saw in 2008).

[Mike Moffatt]

Interesting piece!

RE: Carbon tax - transportation makes up less than 24% of CO2equiv emissions in Canada. What about the remaining 76%?

[Bean]

Well, we eat things, and most importantly, we breathe in oxygen and breathe out CO2 and water. How is that accounted for in discussions of CO2 emissions?

And, along those same lines, this article claiming to have a baby with a zero carbon footprint is patently impossible. There is no such thing - if the baby is eating, drinking, and breathing, it has a carbon footprint.

No doubt manufacturing is a biggie, though. It takes energy input to change raw materials into something else.

[Wonks Anonymous]

The elasticity of demand for tobacco is an empirical question. My understanding is that even "addictive" substances have been found to have rather elastic demand.

No mention of the Georgist "single tax" on the unimproved value of land? I guess the resource tax was pretty close.

[Andrew F]

Frances, as I recall, the carbon tax proposal from 2008 effectively excluded gasoline from the carbon tax on the grounds you raise. I believe the federal excise tax was to be replaced by the carbon tax.

[Fran R Woolley]

Wonks "The elasticity of demand for tobacco is an empirical question".

A government that is attempting to maximize tobacco tax revenues will increase taxes until it reaches the elastic portion of the smokers' demand curve. So, yes, it's entirely possible that, given existing levels of cigarette taxes, the demand for tobacco is elastic, i.e., greater than one.

Andrew F - interesting observation. The BC carbon tax, however, has increased the pump price of gasoline.

[Fran R Woolley]

Wonks "No mention of the Georgist "single tax" on the unimproved value of land?"

A tax on land values ends up being a tax on the value of living close to jobs and parks and other amenities - a tax on land values encourages urban sprawl. Now perhaps there's some way of separating out the value of land and the value of living close to the city centre - I don't know what that might be, though.

[K]

Frances,

There is a *huge* difference between expert estimates of the expected loss of catastrophic events and the market price of insurance of those same events, the difference being governed by the covariance of said event with the market portfolio. There is no question that the market charges very large risk premia for insurance of systemic risk events. Global warming obviously qualifies as a systemic risk.

The correct price to charge for carbon emissions is, at the very least, the market price, not the expert expected loss, since the market price is the fair price of compensation for taking the loss (that's why it's the market price).

But the market price is surely not enough. The fact is, we simply do not care as much about our descendants as we care about our selves. (If we did, there'd be very little equity risk premium since we could diversify over arbitrarily long time periods). The correct market price would be the price of an intertemporal trade reflecting the preferences of all generations.

Since we can't establish a market for intergenerational risk transfer, all we can do is try to estimate the theoretical market price of that risk. The best discussion that I'm familiar with is Martin Weitzman's "On Modeling and Interpreting the Economics of Catastrophic Climate Change." Weitzman makes the point that need to overweight the negative outcome, but also that when we consider model parameter uncertainty, then the warming distribution becomes extremely fat tailed (t rather than normal, for example). The price of risk will not be driven by the mean, but by the not-insignificant chance of a 5C+ temperature rise this century.

Anyways, it's obviously a complex and contentious subject, but personally, I'm inclined to believe that any estimate that comes out the US government will hopelessly underestimate the charge that unborn generations would impose for suffering the risk.

[Andrew F]

Frances, don't you pay the 'proximity tax' either in the form of a land value tax or in the opportunity cost of purchasing land? A land value tax would encourage higher density near places where people may want to live. You can control sprawl with zoning.

[K]

Frances,

"A tax on land values ends up being a tax on the value of living close to jobs and parks and other amenities - a tax on land values encourages urban sprawl."

This is not correct. How do you propose that imposing a tax on land is supposed to impact the free market rental value of that land? If the supply of the land is unchanged, what actual economic decision is impacted by the land value tax?

The answer is that the land tax only impacts the the profit of owning the land and therefore the market price of selling it, not the price of renting it. The ground rent, as explained by Ricardo, "is equal to the economic advantage obtained by using the site in its most productive use, relative to the advantage obtained by using marginal (i.e., the best rent-free) land for the same purpose, given the same inputs of labor and capital." (Wiki) Adam Smith also pointed out that taxes on land cause no distortion on industry.

Imagine that we tax land so much that the value of the land (to the landlord) drops to zero. Does that, in any way, change how much you would pay to rent that land? If you want to encourage sprawl, i.e. disuse of land, you should tax land improvements (i.e. structures). Shifting taxes from improvements, to unimproved land value leaves free market allocations in tact. Land taxes are a totally free lunch.

[Fran R Woolley]

K: " personally, I'm inclined to believe that any estimate that comes out the US government will hopelessly underestimate the charge that unborn generations would impose for suffering the risk."

I agree, but take a look at the estimates that are out there - you won't find much that's over about $50 per tonne. Arguing for a large carbon tax basically means arguing that most of the existing estimates of the price of carbon emissions are wrong.

Andrew F: "A land value tax would encourage higher density near places where people may want to live."

That's not obvious to me. Yes, I can see that developers, buying expensive downtown real estate, would try to extract the max value from it by building as densely as possible. But wouldn't people also move to the exurbs to get cheap land?

[K]

Frances,

"But wouldn't people also move to the exurbs to get cheap land?"

Imagine two equivalent plots of land. They both have an annual rental value of $5000. Interest rates are 5% perpetually, so the market value of each plot is $100K. Now the government imposes a $5000/year tax on the first plot, but not on the second. The market value of that plot therefore drops to $0. You have $100K and want to build a house. Do you want to 1) buy the first plot for $0, invest your money at $5000/year and pay $5000/year in land tax, or 2) buy the second?

[K]

Here's Adam Smith in the Wealth of Nations:

"Ground-rents are a still more proper subject of taxation than the rent of houses. A tax upon ground-rents would not raise the rents of houses. It would fall altogether upon the owner of the ground-rent, who acts always as a monopolist, and exacts the greatest rent which can be got for the use of his ground. More or less can be got for it according as the competitors happen to be richer or poorer, or can afford to gratify their fancy for a particular spot of ground at a greater or smaller expense. In every country the greatest number of rich competitors is in the capital, and it is there accordingly that the highest ground-rents are always to be found. As the wealth of those competitors would in no respect be increased by a tax upon ground-rents, they would not probably be disposed to pay more for the use of the ground. Whether the tax was to be advanced by the inhabitant, or by the owner of the ground, would be of little importance. The more the inhabitant was obliged to pay for the tax, the less he would incline to pay for the ground; so that the final payment of the tax would fall altogether upon the owner of the ground-rent."

Obviously one feels like a bit of a crank quoting the classicals. But truly, I don't think there has been any progress on this bit of economic theory for a good 200 years. If anything, there was probably significant regress in the early 20th century, as the marginalists, especially J B Clark, fought valiantly to completely expunge finite resources as a production factor from economics, leaving just labour and capital (totally mutable "jelly"). This also coincided with the invention of the "long run" as the period after which all resource constraints magically disappear.

[Purple Library Guy]

The thing about the inefficiency of certain taxes is I think vastly overstated, for two basic reasons.
The first reason is, outfits like the C.D. Howe Institute are paid to lie about this kind of thing. So there's no reason to believe or pay attention to people, such as Bev Dahlby, who work for C.D. Howe. If one listens to that sort of people, the rich have no motivation to reduce corporate or wealth taxes because not a penny of any of that actually comes from them after the tax evasion, passing costs to consumers and so forth. Makes one kind of wonder why they spend so many millions lobbying for what they claim doesn't matter.

The second reason is that characterizations of the "efficiency" of certain tax measures have some validity in an abstract space, "all else being equal", but not necessarily in the real world. So for instance, you can say that "all else being equal", if one taxes corporations more they will just evade taxes more, or move operations elsewhere or whatnot. But any government attempting to tax corporations has no obvious reason to allow all else to be equal. If the corporations attempt to evade tax more, the government can respond by cracking down on such evasion. If the corporation wants to move operations elsewhere, the government has many potential levers, from protectionism, to taxing things which cannot readily be moved if the corp wants to do business with customers in the country, to outright expropriation (You're mothballing your plant as you move production elsewhere? You won't be using it? Fine. We'll buy that for a pittance, maybe hand it to the employees).

By the way, where are we on the efficiency of Tobin taxes?

[FR Woolley]

Purple Library Guy - I'm not totally convinced by Bev Dahlby's numbers either, but I think he's more or less right about which taxes have higher efficiency costs and which taxes have lower ones.

One thing that I'm getting at in this post is that there are trade-offs: things that are popular, like greater income equality and justice, usually have a price in terms of economic efficiency. That price may be worth paying, but it's wrong to deny that it exists.

"If the corporation wants to move operations elsewhere, the government has many potential levers, from protectionism, to taxing things which cannot readily be moved if the corp wants to do business with customers in the country, to outright expropriation (You're mothballing your plant as you move production elsewhere? You won't be using it? Fine. We'll buy that for a pittance, maybe hand it to the employees)."

Yes if the government wants to follow Castro's model, not if the government wants to be part of the World Trade Organization.

[K]

Frances,

"but take a look at the estimates that are out there"

But the estimates are almost all costs based on mean warming scenarios, without any particular weighting on the tails. If the mean was no warming, but the width of the distribution as a result of AGW was 5 degrees instead of 0.5 degrees, does that mean that the fair price is zero? Also, fair pricing should *massively* overweight the catastrophic scenarios (way more than the objective expert probabilities). The economic theory underlying most of the public estimates is totally bogus.

Also, discounting is a huge issue, and the lower estimates inevitably discount at the high, return of capital rate. If you take an intertemporal utility optimizing perspective, it quickly becomes clear that the correct discount rate is very low. Here's Weitzman again (from "How Should the Distant Future be Discounted When Discount Rates are Uncertain?"):

"The bottom-line message that we wish for readers to take away from this paper is the following. When future discount rates are uncertain but have a permanent component, then the “effective” discount rate must decline over time toward its lowest possible value. Empirically, this important feature can have significant ramifications for climate-change CBA — by weighting the distant future much more heavily than is done by standard exponential discounting at a constant rate."

"things that are popular, like greater income equality and justice, usually have a price in terms of economic efficiency"

Though the land tax, being 100% efficient and, as Adam Smith describes above, incident on "rich competitors... in the capital, [who] would not probably be disposed to pay more for the use of the ground," ought to be a win-win on the popularity-efficiency frontier.

[Stephen Gordon]

The first reason is, outfits like the C.D. Howe Institute are paid to lie about this kind of thing. So there's no reason to believe or pay attention to people, such as Bev Dahlby, who work for C.D. Howe.

Damn pointy-headed intellectuals. Why don't they tell me what I want to hear?

[Patrick]

I don't think the comment about CD Howe is entirely fair. Fraser, sure. They deserve a bit of scorn. But CD Howe has always seemed to me to be perfectly reasonable.

[K]

Frances,

What are the GHG cost estimates that you are referring to? Looking around I found this 2007 survey by the IPCC. Summary:

"More than 100 estimates of the social cost of carbon are available. They run from US$-10 to US$+350 per tonne of carbon. Peer-reviewed estimates have a mean value of US$43 per tonne of carbon with a standard deviation of US$83 per tonne. Uncertainties in climate sensitivity, response lags, discount rates, the treatment of equity, the valuation of economic and non-economic impacts and the treatment of possible catastrophic losses explain much of this variation including, for example, the US$310 per tonne of carbon estimate published by Stern (2007). "

My understanding is that estimates have been rising since 2007 and the lower estimates generally don't take into account the issues raised by Weitzman and others.

[Fran R Woolley]

K - yup, those numbers, with a mean value of $43, sound about right. To argue for a gasoline tax of more than 25 cents per litre, you'd need to be right up at the high end of those estimates.

Stern (2007) assumes a discount rate of zero, which is why that number is so high. His arguments against discounting are something I've been thinking of blogging about - they deserve to be more taken more seriously than they have been.

It's interesting to contrast these numbers with the price of emission permits under the EU cap and trade system. They have been trading down around a euro or two per tonne - I don't know the exact number, but the point is it's ridiculously low. I've been trying to work out whether or not it's possible for me to buy some. I figure I could completely offset my carbon footprint for a dollar a week, and live a happy and guilt-free life.

[Andrew F]

Also Canadian Centre for Policy Alternatives.

[Bob Smith]

Purple library guy,

The statement that corporate taxes get shifted to others (consumers and, in particular, workers) is a statement of the LONG-TERM impact of corporate taxes. In the short-run, the cost of those taxes can't be readily shifted (because wages are sticky in the short-run, or because short-run capital is fixed). In the long run, though, capital is not fixed (companies can decide not to continue to invest in a particular factory) and wages tend to be more flexible, so the long-run cost of corporate taxes can get shifted away from the owners of capital.

So the observation that corporations lobby against high corporate taxes isn't inconsistent with the observation that their shareholder don't bear the long-run cost of those taxes.

Moreover, it's worth recalling that the people making the decisions about who a corporation should lobby are typically employees (i.e., management) who are generally not significant shareholders. So the statement that the cost of corporate income tax will be borne, in the long-run, by employees is entirely consistent with the observation that management (i.e., the most highly compensated employees in the company) uses corporate resources to lobby against higher corporate income taxes.

[Alex Bollinger]

Good post, but I disagree that policy should be using that definition of efficiency. Instead of worrying that taxes will distort behavior (much of this behavior is undesirable in that it produces externalities, like second-hand smoke and global warming), perhaps the state should adopt a richer definition of efficiency. It's hard to see how the protection of inefficient markets, like fossil fuels, is itself efficient.

I also think that the optimal pigouvian tax on gas is grossly underestimated because of inherent limitations of the cost-benefit framework as currently practiced. Estimating the value of environmental consequences that have no equivalent in current markets (what's the value of a species that could be lost? Etc.) isn't something the framework is ready for. Polling that asks people to put a price on these things is going to produce low-ball answers, both because people worry about what others are going to pay (public goods game) and because people generally say they'll pay less than they actually end up paying. Since the incompleteness of markets is obvious here, there's no reason to assume that preferences are complete either.

[Fran R Woolley]

Bob: "So the statement that the cost of corporate income tax will be borne, in the long-run, by employees is entirely consistent with the observation that management (i.e., the most highly compensated employees in the company) uses corporate resources to lobby against higher corporate income taxes."

Lovely observation! Thank you.

[Declan]

I'm puzzled by the implication that the only externalized cost of driving is climate change. I'd guess that a true accounting (even one subject to the limits Alex is correct to note above) would show the current gasoline taxes as being sorely inadequate to even cover the non-climate related externalities (noise, smog, congestion, social mobility, etc. etc.). Also puzzled by the implication that a carbon tax would only affect gasoline purchases.

[Bob's your uncle]

Lovely observation! Thank you.

Sheesh. Do I really need to explain why higher paid management might want this?

[primedprimate]

Frances, I would love a post on discounting. This topic has recently been on my mind a great deal and I've just finished reading some interesting papers by Partha Dasgupta, John Quiggin, and Caplin and Leahy, etc. I also recently read a very useful 'summary-type' Asian Development Bank paper on current practices and relevant literature: http://www.adb.org/sites/default/files/pub/2007/WP094.pdf

Also, the most recent issue of Science has this brief piece by a constellation of luminaries (Arrow et. al.): http://www.sciencemag.org/content/341/6144/349.short

I enjoy reading WCI immensely and I am very keen to read your (and the commentariat's) views on discounting.

[Bob]

BYU,

I think I just did.

[primedprimate]

I guess my comment was lost in the spam filter. Oh well, it was merely a response to one of Frances' comments and only tangentially related to the main thread.

[Bob Smith]

Declan,

Certainly the traditional rationale for the gasoline tax was to pay for construction/maintenance of the road network (which, I note, most estimates suggests it fails to do). Which would provide a fairly compelling rationale for both (a) a higher gasoline tax (or other forms of road pricing) and (b) a carbon tax.

Mind you, good luck getting that passed.

[Fran R Woolley]

primedprimate - send me a direct email if ever one of your messages is lost in the spam filter. Once Typepad decides you're spam, that's it - all of your comments on any typepad blog will go to the spam filter for ever and ever. Timely rescue may or may not prevent this from happening, but it's worth trying.

[erik]

Frances

You should definitely do a post on discounting. The arguments against pure time discounting, when something has an effect over many generations, are completely watertight (and pretty strong within a generation as well, given that you are an utilitarian). I cant imagine that there is one single philosopher who would disagree.

There are, on the other hand, many second best arguments as to why we should use higher (or lower) discount rates on public projects (as a shorthand, instead of explicitly including a full model of the second best economy).

[Majromax]

The question is: why would a government find itself in this situation? Why would it ever raise revenue from relatively inefficient and unpopular taxes when it could be raising revenue from efficient or popular taxes?

There's another possiblity, and that's that the qualities of efficiency and popularity have a nonlinear dependence on the tax rate -- essentially Laffer curve thinking.

For example, a 5% excise tax on cigarettes is an efficient/popular tax, since it does not encourage evasion, but a 10,000% excise tax would have nearly universal evasion, making it an inefficient tax even if it remains equally popular.

From a strict revenue-optimization standpoint, the government should consider expanding the tax base (to less efficient or less popular taxes) at the point when the marginal loss to new tax evasion equals the marginal income gain. Since avoided taxes are avoided in their entirety rather than in the margins, this will be at some finite rate.

In reality, governments may wish to expand the tax base far sooner than that. The strict revenue-optimizing rate will necessarily have a high rate of evasion, and it's probably to the government's benefit to have a low rate of tax evasion for other, social purposes. As other articles (here, I think) point out, tax evasion has a strong network-effect component whereby someone is more likely to evade taxes if their peers are as well. Additionally, in some cases tax evasion scales: it's not that much harder to smuggle 20,000 cartons of cigarettes as it is to smuggle 10,000. There might even be hystereisis effects.