If you incentivize it, will they come?

On Monday, I argued investment incentive policy is best understood within a political framework that takes seriously the electoral incentives state and local officials face. Looking through the New York Time's interactive database of investment incentives, it is striking how widely states vary in the amount of incentives offered. What explains this variation? One possible explanation is individual agency. Perhaps Texas has a very large incentive program because of the political influence of G. Brint Ryan; this is the implicit argument forwarded in the New York Times Investigative Series on Investment Incentives. Comparative political economists would point to institutional variation; differences in governance structures and how susceptible local governments are to corruption may explain the extent to which states pursue incentive programs.* Partisanship might matter too, although it is difficult to make the case that voters see incentives as clearly benefiting benefit at the expense of workers. Indeed, as I mentioned on Monday, experimental evidence suggests voters see incentives through the prism of job creation. This makes a partisan-based mechanism less plausible.

While people and institutions may help explain a portion of variation in incentive programs, I’d argue structural conditions are most important. (This probably won’t surprise readers of the blog – contributors here tend toward thinking about the world in such terms.) As I mentioned in Monday’s post, states and localities are working to attract jobs in a context of open capital markets. Consequentially, absent transaction costs, capital is mobile while labor is relatively fixed and this makes capital strong. Capital gets locational incentives because its exit option is credible. Labor, however, is captive so governments can tax it more. The problem with this view (besides the fact that suggesting governments face little pressure to reduce taxes on middle-class workers will get you laughed – and voted - out of Washington these days) is that the global economy, while open, is not frictionless. Transaction costs, or in network terms, negative externalities are important.

There is a large literature in economics on agglomeration effects – basically the idea that centers of economic activity form due to positive externalities generated by the success of a few enterprises.** In the 1950s, Detroit was perhaps the best example of one of these centers. Successful, large manufacturing enterprises require deep supply chains, preferably located with geographic convenience to reduce transportation costs and to decrease production times. Competitors often locate nearby to be better able to recruit management, design, and other knowledge workers. In network terms, when a fit enterprise center emerges, preferential attachment reinforces that center.

Today, thriving centers of economic activity in the US include New York City, the Silicon Valley, and perhaps even NC’s own Research Triangle Park. It is then not surprising that, according to the New York Times report, California is reducing its incentive program, which is already comparatively small at $112 per capita. New York and North Carolina have relatively low per capita incentive programs, $210 and $69 respectively. Compare that to the three largest incentive programs on a per capita basis – Alaska at $991, West Virginia at $845, and Texas at $759. Economic geography matters. The states with the largest incentive programs are those that either never generated large centers of economic activity, or whose centers have become obsolete as our economy has shifted from manufacturing to services.

Economic centers form due to a confluence of factors, some of which governments have control over and some that they don’t. Investments in education and infrastructure can provide a skilled workforce and inexpensive access to energy, telecommunication, water, and transportation networks. And, it is true that offering locational incentives may reduce governments’ ability to invest things that will actual increase their locality’s fitness. But, this ignores the fact that incentive programs are fundamentally designed undermine powerful network effects that concentrate economic activity. That is why they are so inefficient; because they are swimming against the current. Fitness is not the whole story – preferential attachment entrenches economic centers. So, incentives ultimately are big risks – if you are lucky, you may attract enough high-quality enterprises that you can build a thriving center. But, network dynamics are working against you.

* Nate Jensen pointed me to this particular NBER working paper , which finds evidence that corruption increases incentive programs.

**See here (firewalled) for a review.

Shall We Continue in Sin, So That Grace May Abound? God Forbid.

I used to blog sometimes about how many things we call "public goods" really aren't. People label things they like "public goods" because it eliminates opposition of the public provision of these goods. So folks call all sorts of things "public goods" which are not public goods: education, health care, etc. Actual public goods, which are both non-excludable and non-rival in consumption, are pretty rare. I stopped harping on this because I thought I'd made my point and nobody else seemed to care.

But now I see Mike Munger twisting himself into knots over whether roads are public goods, so I'd like to revisit the topic. Munger's conclusion is that roads are public goods, kind of, sometimes. But not other times. He reaches this conclusion by comparing the marginal cost of use under different scenarios: if the addition of the marginal car has a zero (or near zero) impact on the cost of the using the road then it is a public good; otherwise it is not.

This is mistaken in the same way that it is mistaken to say that the "Tragedy of the Commons" is a story about externalities (or public goods). A public good is not defined by comparing the cost of additional units of consumption at various margins. For true public goods the marginal cost of additional unit of consumption is negligible at all margins. That is the definition of a public good: increasing consumption does not reduce the amount of consumption available to others. When we compare costs at varying margins all we're doing is talking about relative levels of scarcity. Public goods are not, cannot, be sensitive to scarcity.

To understand where this logic ends consider that under Munger's definition -- public goods are good, and less-used roads are the most public goody of all roads -- we should build a bunch of roads (and bridges) to nowhere. Almost no one will use them, so the marginal cost of an additional vehicle will be the closest to zero that it can possibly be. Let's start building!

This is the sort of absurdity for which Saul of Tarsus admonished the early church in Romans 6: if God's grace is good, and grace is only extended to cover sins, then should we sin as much as possible in order to maximize grace? Of couse not. Similarly, we should not build roads which will not be used.

Roads are excludable: to use them you must possess a motor vehicle as well as an assortment of licenses and insurance contracts which permit you to operate that motor vehicle on that road. You and your vehicle must also physically be in the place where the road is. Roads are also rivalrous in consumption: the more people use them the fewer additional people can use them without congestion. Roads are therefore not public goods. Ever.

It does not necessarily follow that there should be no public provision of roads. Just because something is not a public good does not mean that there is no reason for public provision of it. There may be a case which can be made on consequentialist grounds that collective action (via taxation) to provide a non-public good is justifiable. I believe that many roads will pass this sort of cost-benefit test. But this case needs to be made on its own merits.

And if we make that case on its merits, we will likely come to the opposite conclusion of Munger: scarcely-used roads in rural areas are the ones which should be tolled/taxed. Why? Because the case for public funding of roads is not that they are public goods, but that they increase efficiency by reducing transaction/transportation costs. They function like a utility in an environment where a monopolistic market structure is likely to be more efficient than a competitive market structure so long as the monopolist is not a profit-maximizer (i.e., where the monopolist's producer surplus is redistributed to consumers, i.e. where the monopolist is a government -- subject to an electorate -- rather than a firm). Those efficiency gains will be highest when and where the roads are used the most, and lowest when and where the roads are used the least. Public subsidization should be highest where there is the most potential for efficiency gains. This occurs in the busiest areas.

If we see lots of congestion on some roads that is a signal that we should build more roads there. Not to make roads more like public goods (by reducing the cost of the marginal unit of consumption), but to try to reap the social gains from whatever economic activities are causing the congestion. If we cannot build more roads (because there is no empty land, say) then we should build some other transportation network, like bike paths or subways, to allow people to engage in productive activity more easily. The positive spillover effects from such investments are more likely to pass a cost-benefit test than in a rural area.

I'm not opposed to congestion pricing in general, but we need to recognize congestion pricing for what it is: a tax on productivity. People don't drive into Manhattan during rush hour because they enjoy it. They go through that nightmare to get to work, often in high-wage/high-productivity sectors of the economy. I'm not sure why we'd want to discourage that.

Why We (Sometimes) Shouldn't Internalize Externalities

At the end of my previous post, I suggested that we might not always want to force people to internalize externalities. This is anathema to most economists and indeed believers in the market as the most efficient allocator of resources. After all, if the market allocates efficiently, and externalities are market failures, then externalities should be corrected. In many cases this will be true, but not, I think in all.

First the argument in favor of internalizing externalities. In the case of positive externalities, if the people generating the benefits are not able to capture the value of those benefits, then they will tend to not produce those goods. Even if these are not public goods, wealth creation is good for its own sake as people will tend to enrich others' lives with their innovations as well as use their wealth to employ other people*. In other words, we want people who are capable of creating new things to create those things, ceteris paribus. If people are not compensated for doing so, they will tend not to bother, or so the argument goes. In the case of negative externalities, if people are not forced to internalize those costs they will tend to over-produce those goods. That's bad for society as well.

To see where this can go wrong, let's take an extreme example. Bill Gates has been the world's richest man, although he's currently #2 behind Carlos Slim. What's more, no one believes that Gates has secured his wealth illegitimately. However, the benefits to the economy of Microsoft's products have not all been captured by Gates. As intermediate goods, Microsoft products allow people to produce other products, and to do so more efficiently than they otherwise might. Though I can't find the study now, a few years back someone estimated that Microsoft had generated hundreds of billions of surplus value to the economy, mostly by reducing transaction costs, improving efficiency in a number of business environments, and generating network externalities. These innovations have literally benefitted hundreds of millions, if not billions, of people all over the world.

You can probably see where I'm going with this. The fact that Gates has not been able to internalize all of the benefits from his products has not stopped him from creating and distributing them. On the contrary: in order for Gates to be as wealthy as he is he has to make sure his products are broadly dispersed, which is why he gives some of them away for free. He has been rewarded very well, but no utilitarian calculus that I'm aware of would suggest that we should redistribute away from ordinary people who have benefitted from Microsoft to Gates. He has more than enough. So we should modify the common claim at the beginning: internalizing (positive) externalities only needs to be done when we're operating near the margin where innovators might choose not to innovate without more compensation. And if we think in those terms, it isn't immediately obvious that we're usually operating at that margin. It is well-known, for example, that popular live music events are generally "inefficient" in that they usually sell out quickly, which indicates that there is quite a lot of surplus that is being captured by consumers rather than producers (or other consumers, who might be willing to pay more for a ticket in an auction setting). Economists generally think this is puzzling, but from a utilitarian perspective it's arguably a very good thing.

How about negative externalities? Arguably this is a more difficult case, but if we keep using the same utilitarian logic I've presented it isn't hard to argue that poorer people should be given some license to push negative externalities onto richer people. People have used this sort of argument to suggest that developing countries should not have to adhere to the same climate change restrictions that richer countries should adhere to. The question is where the margin that tips over into inefficiency lies. I admit that these cases will be much more in the eye of the beholder than the case in which Gates' positive externalities are broadly shared, but there isn't any ex ante reason to discard the idea.

*You don't have to be a believer in strong-form trickle down economics to generally accept this point.

Negative Externalities and the Tragedy of the Commons

In a classic example of a tragedy of the commons, a field is used by several farmers to graze their cattle. None of the farmers have the right to restrict access to other farmers. The field can support a certain number of cattle, say ten, without being depleted. But any additional cattle added to the field beyond ten will deplete the field and everyone will suffer. In this case, every farmer has an incentive to add the eleventh cow, as the benefit from doing so will accrue only to that farmer, while the costs will be shared by all. Therefore, the owner of the eleventh cow has imposed a negative externality on the owners of the ten when he brings the cow to the field. The tragedy is that because individual incentives deviate from collective incentives, and there is no external enforcement, everyone is bound to suffer.

This example, or one like it, is used to illustrate many problems in the world. Pollution, overpopulation (as in the original Garrett Hardin article that I believe remains the most-cited in the history of the journal Science), climate change, financial regulation, etc. In many cases, we focus on the ability of institutions like government, property rights, or bargained contracts to overcome this tragedy and ensure a more efficient outcome.

The problem is that it makes no sense. At least, it makes no sense when translated into the language of externalities, which nowhere appear in Hardin's original article. To see why, remember this: there are no property rights in the commons. So while the eleventh cow is the one that puts the commons over the edge, this is only true because of its cumulative effect with cows one thru ten. Were any of of those ten absent (i.e. not just the eleventh), the presence of the eleventh would be no problem. Since no farmer has any more or less right to access the commons than any other, all the farmers are equally culpable when the commons becomes depleted. The owner of cow #11 is doing no more (or less) damage than the owner of cow #4. Since all are culpable and all share the cost, there is no externality. All the farmers are contributing equally to the problem, all are equally suffering the cost. The removal of any of the cows would solve the problem, and there is no established reason for the claim that the eleventh cow is the one that should be removed.

The logic of externalities requires an established property right that is being infringed upon. It also requires that someone is forced to pay the costs of an action without accruing the benefits from it. When we realize this, we also realize that the problem of externalities is a much smaller one than we first thought. It does extend to manufacturers that pollute a river that makes fishing impossible downstream, but only if the fisherman has a claim to the river and does no polluting on her own. It does not extend to a financial crisis that culminates in some people being unemployed or losing wealth via asset depreciation, unless (perhaps) the crisis was caused by fraud.

The question that remains: what about climate change? Is this an externalities problem? Every society that has the capability to pollute does so, albeit at differential rates. While the infamous Summers memo was obviously sarcastic, there is a kernel of truth embedded in it: pollution is a by-product of economic activity, all of which generates waste, and the lack of which is a greater threat to human well-being in the developing than climate change. The pollution levels of even low-polluting industrial countries like Japan would be more than enough to warm the planet if all countries polluted at that rate. This is analogous to the owner of the eleventh cow: whenever anyone comes into possession of this cow, they will add it to the commons and everyone suffers. We don't fault the owner of the eleventh cow for ownership, we fault the structure of the interaction. If the only difference between the culpable and non-culpable is opportunity, can we really feel comfortable ascribing blame?

Secondly, is it really clear that people in Bangladesh (say) have a rights claim to an climate environment that is not two or three degrees warmer than it is right now? How did they acquire this right? This is an important point when considering the politics. Some countries, such as Russia and Canada, will likely benefit from a warmer climate. Some countries will suffer. Some countries, like the United States, will probably not be greatly affected in aggregate, but within the country some groups will benefit while others will suffer. Similarly, taking large efforts to prevent or mitigate climate change will benefit some and harm others. How do we resolve this without some sense of who has what claim to what type of climate?

None of which is to deny that even if climate change does not represent an externalities problem, it nevertheless may constitute a human problem. And really I'm less interested in climate change per se as I am in the fact that the tragedy of the commons is not necessarily an externalities problem as we usually think.

I'm planning to follow this post with another post on why we might not want to force people to internalize externalities in all contexts.