Economic analysis

We’ve been evaluating pipeline projects all wrong

We should ask whether the social returns on investment are sufficiently large to outweigh the costs of resources used
FILE - In this Wednesday, Feb. 1, 2012 file photo, miles of pipe ready to become part of the Keystone Pipeline are stacked in a field near Ripley, Okla. President Barack Obama says that the proposed Keystone XL pipeline project from Canada to Texas should only be approved if it doesn’t worsen carbon pollution. Obama says allowing the oil pipeline to be built requires a finding that doing so is in the nation’s interest. He says that means determining that the pipeline does not contribute and "significantly exacerbate" emissions. (AP Photo/Sue Ogrocki, File)
Etienne de Malglaive/REA/Redux
Etienne de Malglaive/REA/Redux

Last week, I wrote a piece critiquing the use of economic impact analysis as a means of evaluating the social benefits of new infrastructure or other major projects. My main critiques of this method centered on two areas: first, that the method does not differentiate with respect to what the spending produces in the end and, second, that the method does not generally account for the fact that at least some of the spending comes at the expense of other activities which would otherwise have occurred. In other words, the method does not account for the value of what is built or for the opportunity cost of the resources used in producing it.

While it’s easy to criticize, some readers rightly asked what the alternative would be—what should we do instead? There’s a relatively simple answer to this: benefit-cost analysis. If we want to ask what the net benefits of a project are, the logical approach would be to assess the benefits and costs accruing to relevant stakeholders, and to then determine the net benefits (or net costs) of the investment. In effect, we should be doing the same calculation from a societal perspective as firms would undertake for their investments—we should ask whether the social returns on investment are sufficiently large to outweigh the costs of resources used.

There are three main challenges with respect to this method, each of which are present within our current discussions over pipelines. First, determining standing, or whose benefits and costs should factor into the analysis. Second, converting non-market benefits and costs into a comparable metric often presents a challenge. Finally, determining how we should weigh benefits and costs over time both in the near and long term is very controversial.

The question of standing with respect to benefit-cost analysis is controversial in many ways since you are effectively asking who counts and who doesn’t. Perhaps the best illustration is with respect to greenhouse gas emissions and pipelines. New pipelines, and the associated resource development, will produce greenhouse gas emissions, some of which will likely be incremental to what would have otherwise occurred. To determine how best to include the costs of these incremental emissions in your benefit-cost analysis, you’d need to assess the likely impacts associated with emissions (i.e. the climate sensitivity) and the damages associated with induced changes in climate. While these are very challenging to estimate in-and-of themselves, even supposing you could do so accurately, you’d then need to determine the share of those damages which you consider to impact the relevant population. In other words, do we assess the damages to Canadians associated with incremental GHG emissions, or do we assess global damages? The two figures will vary by multiple orders of magnitude.

The same issues apply on the benefits side. Improved pipeline infrastructure will allow Canadian oil resources to be more efficiently shipped to higher value markets, increasing their value. But, some of the rights to these resources are owned by non-Canadians. Should we consider additional profits accruing to non-Canadian shareholders as part of the social benefits of a pipeline project, or should we consider only the share of incremental profits as well as the increased taxes and royalties which will be captured by Canadian shareholders and governments? If you spend much time around the pipeline file, you’ll notice that many who say we should count the global costs of climate change don’t feel quite the same way about the appropriate reach of standing with respect to counting increased profits from resource extraction and vice versa.

Once you’ve determined standing, then the next step of the analysis is the quantification of costs and benefits to those with standing. For benefits such as the expected increases in resource values or financial costs like those from potential cleanup costs for oil spills, this is relatively straightforward, since they are already measured in dollar terms. Where this gets difficult, and where a great deal of economics research has been focused for decades, is when we try to measure environmental and other non-market costs in comparable ways. Of course, nothing in economics says we would need to do this in dollar terms—we could do the whole analysis in acres of old growth forest equivalents or polar bear equivalents where the increased revenues in financial terms would be converted to natural capital equivalents, but that’s not the direction in which most of the research has gone. Economists tend to use either stated (i.e. surveys) or revealed (i.e. looking at behavior) preferences of individuals to estimate values for intangible benefits and costs in dollar terms.

Expressing the value of non-market amenities and costs in dollar terms is not popular, but we will end up doing so implicitly in most project evaluations, so let’s make it explicit. Consider the example of an oil spill—either the non-market costs (over-and-above clean-up costs) are zero (which seems indefensible), infinite (also indefensible in my view), or somewhere in between. Reasonable people can argue about how we should measure those costs, and it’s important to recognize that this is a discussion which goes far beyond traditional economics, and into ethics, sociology, and other disciplines. The fact that it’s difficult or controversial to assign a number between zero and infinity doesn’t mean that we should use zero or infinity.  If you want to calculate net benefits, you need costs and benefits to be in the same units or the math doesn’t work.

Some of these same issues also arise when it comes to valuing a “job”—a salary paid to someone is not a gift, but a payment for services, and a worker employed on this project may or may not have been employed elsewhere in its absence. So, how would a cost-benefit analysis value a job? Supposing that the salaries are paid by Canadians to Canadians, you’d be close to zero-sum as long as the labour market is at or close to full employment. It would be reasonable to assume, in that case, that the workers employed building a pipeline would be otherwise employed building something else in the absence of the pipeline project, and so the impact on their well-being is limited. Similarly, from the employer’s perspective, they pay the wages, but also receive the value of the labour, and in a tight labour market those two should be close to equal at the margin.

The analysis gets more complicated in a slack labour market in which the opportunity costs of labour will generally be lower, and there may be social benefits over-and-above the product of labour from having higher employment. These should be factored-in to a complete analysis.

The final issue is one where, again, we have to look beyond economics and be content that there likely is no right answer: how do we value costs and benefits over time and across generations? This is certainly relevant to the issue of valuing environmental damages from climate change, where the U.S. government for example publishes a range of potential social costs of carbon emissions for different assumptions on how we value future costs and benefits. It’s easy to say that we should not discount future benefits or costs relative to current ones, but that would lead to some bizarre conclusions. Reasonable people should be able to agree that some time-value of money is acceptable (if you don’t, can I borrow some money and pay you back in 50 years, without interest?), but we should welcome disagreement, discussion, and research on exactly what the time value should be or at least the range we should use for evaluation of projects.

In my piece last week, I said that we were doing project evaluation wrong. How do we do it right? We should deploy a technique analogous to what the project proponents are doing—we should look at the resources which we, as a society, would supply to that project were it granted a permit and what our returns on that investment would be, adjusting for risks. It’s entirely reasonable to suggest that a project could have private returns which far exceed its social returns (which may even be negative). Preventing projects with negative social returns from being built is why we have the regulatory process in the first place, so let’s demand that the process assess these returns carefully and using the best techniques we have available. Taking spending, cycling it through multiplier a model and yielding an economic impact is not that technique, and we can and must do better. Let’s stop using it.