Thursday, March 11, 2010

Pigouvian taxes and subsidies

There is a theory in economics that an effective way to reduce the incidence of a bad behavior is to place a tax on it. The reasoning goes that this tax increases the cost of the behavior, thus by the law of supply and demand, the quantity demanded decreases. The use of taxes for the purpose of balancing out externality costs is referred to as a Pigouvian tax.

The implementation of these strategies in the real world tests both the entire regulatory framework and the economic theories. This naturally complicates things - are failures the result of the theory being wrong or merely due to flaws in implementation? Many public policies that seem to have great promise may actually be nothing more than the fiscal equivalent of building a levee on this mile of the river: flooding is prevented here but made worse downstream.

I have been approached on multiple occasions by bicycle advocates who have suggested that better bike policies and funding for bicycles can best be brought about through a campaign against car usage. The reasoning goes that by not funding more roads and by increasing taxes on gasoline, the "cost" of driving increases. By the Pigouvian reasoning, this will lead - nay, force - people to pursue alternative transportation options.

There is certainly some marginal effect, but it will not follow a uniform curve. This marginal effect will be a function of the attractiveness of alternatives, and these alternatives can be quite varied. The temptation is to be lazy, and to assume that some intersection of continuous curves defines the demand functions, meaning that each incremental cost increase of driving would have a similar, incremental decrease in driving. This, of course, is not the whole story, because the reality of the situation depends on the discrete decisions that are made. A small increase may not be enough to actually push a different transportation option into the top spot - people are actually quite uniform in their transportation situations and the biking/public transit system tends to have a more or less uniform cost. Furthermore, an additional "hump" that represents the force of habit may block people from changing habits for savings that are insignificant.

Therefore, there is a threshold below which only an insignificant portion of drivers would bother changing their plans. Above this threshold, a large number of drivers would seek to change their plans. And so, the question now arises, "What is the value of incremental driving disincentives both below and above the key threshold?"

Below the threshold, taxes on gasoline and congestion promoting policies amount to little more than regressive taxes. That is to say, the cost is passed through the drivers. Rather than changing their driving habits, it merely leads them to have less wealth available to purchase other goods in the economy. Owing to the dependence of many goods on gasoline transport, these taxes may also be expressed in the cost of goods.

The story is essentially the same above the threshold as well: those who are still driving with high taxes on driving are unlikely to change their behavior simply because driving starts to have a slightly higher cost.

The real effect occurs at the threshold, where the variable is actually sensitive. This is the only area where a case can actually be made that the pigouvian incentive concept is a good idea, because here the tax actually does bring about a reduction in the unwanted behavior. Here, the supply of the chosen alternative goods must be sufficiently flexible to absorb the relatively rapid change in habits that the cost increases bring about. But are public transit and bike networks actually capable of making these sudden changes? Not really. Here again, even at the threshold, the consumers will find they are unable to switch because the existing bike infrastructure is insufficient and the existing mass transit system is already operating at capacity.

A similar argument can be made for subsidies of public transit and bicycle infrastructure: they will be underutilized and fail cost/benefit tests without a complimentary policy of increasing the cost associated with existing behavior.

So, it is sensible to propose a steep tax, and to cut back severely on existing vehicle infrastructure, such that the effect is sufficient to bring incentives across the threshold, only if a comprehensive expansion of public transit and bicycle infrastructure is also implemented in parity.

In reality, the issue of cost pass-through is present in every tax and subsidy scheme. It can have a variety of unintended consequences. Generally, cost driven changes in consumer behavior come in the form of regressive policies. Care should be taken by well intentioned reformers to take the two steps outlined here: first, to find any thresholds in the discrete consumption bundles of actual goods; and second, to find and assess the role of potential bottlenecks in the desired new behavior pattern.

Additionally, there are two more potential problems in the basic implementation of such systems.

First, the shift in behaviors as a result of increasing costs may be unpredictable. Could increases in driving cost drive a portion of people to simply never leave their houses, instead becoming Television and Internet addicts? Could toxic waste disposal fees, instead of leading to reductions in toxic manufacturing processes, lead to the marketing of building material mixtures that contain these toxins (so that they do not need to be disposed)? The regulatory framework, as well as the nature of the goods in question, are the determinants of such results. Thus, taxes may need to be imposed on a large number of goods, and subsidies or other supply expansions may need to be effected on a similar scale.

Secondly, the political cost may be great. A politician in anything but a safe district would have to be incompetent to back a gasoline tax. Even safe politicians may want to vote against a gas tax simply to preserve the stream of automobile lobby dollars. At some point, advocates for social and environmental justice will need to assess the political realities. Though it is a difficult task, compromise will be necessary in all environmental agendas. For real political change, a great deal of consensus, log rolling, and popular awareness are essential.

In conclusion, there are four prerequisites for the use of a pigouvian tax/subsidy scheme:
1. Discrete understanding of the supply/demand for the goods to be taxed and subsidized in terms of the consumption bundles of the consumer demographics.
2. Ability of policies to simultaneously cross both the demand threshold for the good that is to be suppressed and any necessary subsidy to supply of the substitute good.
3. Adequate regulatory structure to ensure that new consumer behavior will actually be as planned. This may require a comprehensive set of many taxes and many subsidies across a large number of goods.
4. A strong political position for the elected officials responsible for the policies that bring about 1-3, as well as an electorate that is highly sympathetic with the cause that such policies represent.

No comments: