I was a Benefit-Cost analyst
I did not intend to be. I intended in 1966 to be an electrical engineer but that diffused into a systems engineer and then mostly for transport systems. I did not use much circuit theory but feedback analysis became part of my worldview and helped my learning about complex systems, that I now just call the ecology.
Not me and not a spreadsheet, but what does number crunching tell us about the ecological future?
One thing led to another and in 1977 I was writing a benefit-cost manual for railroad projects. Four years later I was leading project to analyze Federal policy for railroad trackage rights and reciprocal switching. I went back to school for more systems engineering and a lot of math programming. In 1984 I was hired to be a benefit-cost analyst for the National Airspace System (NAS) Plan to modernize the automation of the air traffic control (ATC) system. And so on until retiring in 2011 from the ATC terminal-radar system program office. Expert in anything? Nope.
I led two lives and the other was getting embroiled in urban advocacy, highway battles and being Transport Chair for some groups. More on that in my Highway Revolt or Cookbook and HAC. My opus on all these subjects is Limits to Mobility (LTM) started in 1975. Benefit-cost (B-C) analysis is part of all this critique of what I call eminent-domain projects (EDP). The B-C part is mostly in LTM Part III. B-C analysis for projects by the State (our governance hierarchy) is to financial accounting as military music is to music. Why are we supposed to be impressed by this pseudo-dollar measure of projects by which the State changes the ecology?
The Pareto Optimum
What stuck with me from my first B-C manual writing was Vilfredo Pareto’s writings on welfare economics [e.g., Manual of Political Economy, 1906]. I then got into E.J. Mishan’s critique of B-C [The Costs of Economic Growth, Praeger, 1967; Technology and Growth, The Price we Pay, Praeger 1969]. Perhaps the target of these critiques is the Musgrave’s doctrine on Public Finance [Richard A. Musgrave, The Theory of Public Finance, McGraw Hill, 1959: Updated with his wife: Richard A. Musgrave and Peggy B. Musgrave, Public Finance in Theory and Practice, McGraw Hill, 1973 and later eds.].
I divide the issue into financial accounting and EDP. That maps to the theory of micro-economics and of welfare economics. Pareto made the distinction clear in two terms that have since lost their important distinction. For Pareto ophelimité is what motivates an individual to act and utilité is the analogous motivation for some public or State action. The difference has been lost in the English usage of “utility” that is both “useful” (to whom?) and the economics term for an imaginary measure of individual satisfaction.
The question of what motivates any behavior, at any scale including EDP, only leads to wide-open questions of learning (at several scales from genomic to individual) and accountability. When outcomes of an action are uncertain, that brings in risk and the scale of uncertainty usually brings in different outcomes for different people/things and so equity.
LTM arrives at the conclusion that the role of the State is to be accountable for risk and equity by preserving the forms (rules) of behavior. Anything else implies that the State knows better how the ecology (including our society) should operate. The State cannot define utilité anymore than we can measure ophelimité. In that case the attempt to get an objective measure, a cardinal number, for the outcome of any action is tempting. One of the surrenders to that temptation is B-C analysis. Another is voting that is not a number but a preference ordering that might (maybe) map to a number. That little subject was treated by a couple of great minds [Von Neumann and Morgenstern, The theory of Games and Economic Behavior, 1944, Sec. 3, The Notion of Utility, and Appendix, The Axiomatic Treatment of Utility].
We can get all wound up in various theories of welfare economic and social-preference ordering. Whether Kenneth Arrow [Social Choice and Individual Values, Yale University Press, 1951] put an end to the idea of an agreement about utilité is still widely discussed. There is interminable discussion of that in LTM Part III which is another good reason why it has not found a publisher.
I think Pareto cuts the knot. We act in some way to improve our situation “under the circumstances” that is the form of the ecology we are in and cannot change directly by our behavior. If you want to call that an increase in ophelimité, fine. The only data are in the action and a change in state (some measurable attributes) of the actor. Was the action a choice or compelled? That is another thing we cannot really measure. How do we know how to act in this way? We learned. Some of that is from long genomic trials and some from what we read in a manual or heard on the street. And some of it is to follow that crowd because, after all, there is more learning in social groups than individuals (you think?). Yeah, like lemmings.
Once we get this far in illogic we might as well agree with whatever the political hierarchy wants to do (utilité) because, after all, is that not in some way a product of social wisdom? Ha, ha. Look at who we are following now, and his clown car of minions.
We can easily go down any garden path to utilité. The one I focus on is the EDP mission: To demolish urban development to give more capacity to auto traffic where it does not fit. That is utilité absurd, but who notices? Who dissents? Plenty of us and yet it goes on. Change the climate, extinguish species? Yeah, same.
Pareto defined a criterion for any action with risk, that affects many (or much ecology) but with equity. The Pareto Optimum is a change in the ecology (the circumstance all are in) that leaves everyone no less well off (no decrease in ophelimité). Great. Now can any action meet that criterion and how do we measure it?
The Second Best
In welfare economics the “second best” is a form (rules, circumstance) of interaction that is, perhaps, the next best form to a pure economic form. For instance, highway congestion is a non-market of exploiting the urban access ecology. The option to make a (de)congestion price for trips emulates the marginal-cost pricing of market goods and is a second-best option for lack of the “first best” market form.
As discussed in the post on traffic assignment, the (de)congestion pricing has measurable utilité by reducing the total (social) travel cost. It is generally not a Pareto Optimum since the travel costs fall inequitably. As in the Manhattan case, Staten Island will squawk. How the revenues will be used, as on transit service, will further alter the utilité. But it is all a “public” (government regulation) decision according to a political calculus that is hardly a complete ecological accounting. There is risk and inequity that are more specific aspects of utilité and that should motivate the government machine. But don’t get me wrong: Regulating auto traffic where it does not fit is an attempt to preserve the urban ecology that indifference to auto traffic—and then EDP to put more where it does not fit—violated.
Defense is a function to mitigate risk to our form, our “way of life” and for all equally. But how can that be applied when it comes to acting on our form in the assumption that our way of life was defective? Demolishing or taking over cities is not defense. Consuming fossil resources is not defense. Utilité is perhaps the general Welfare but how to promote that is the question and inseparable from the complete ecology we live in.
We could compare urban pricing of auto traffic with commitment to “free” roads. That underlies the Federal highway EDP and in 1939 the explicit choice for the Interstate system. A counterfactual history of pricing traffic by its density imploes the comparison of quite different ecological forms and how is that to be done?
The “first best” economic form is a Pareto Optimum as far as that goes. The economic game as described by Debreu [Gerard Debreu, Theory of Value: An Axiomatic Analysis of Economic Equilibrium. Yale University Press, 1959] is about a distributed endowment of tradable goods and then voluntary transfers until the value of the game (its final, equilibrium state) is reached when no one has any ophelimité to gain by further trade. That raises questions of information and so risk. However it is assumed that everyone follows a path that is probably (with risk) an ophelimité gain at each move. It has to be a Pareto Optimum by individual choice. It is immeasurable because ophelimité is and all we have are the data of each move or that each stops changing. It is as inequitable as the initial endowment. Whether there is some better value is undefinable within this form but there could be a different, perhaps more equitable endowment. That is what the ecology and governance provides outside the economic game and includes EDP. But EDP just increases the risk and inequity of large changes in the ecology.
In practice we measure the economy by an aggregation of its measurable dollar tokens, as the GDP. What that implies has been examined by Daly [Herman Daly, Steady-State Economics, 1977]. GDP mostly signifies “growth” as a non-equilibrium extraction and inequitable distribution of endowment from the ecology that is not a voluntary agent in the economy. Drill Baby is all about that project: Sucking up more limited resources will probably increase the GDP with no reference to risk, equity or the state of the ecology. That is little thinking as are all projects for some partial evaluation that serves as utilité without regard for ecological risk or equity. So where is the clause in the Constitution about increasing GDP? What does “promote the general Welfare” mean?
Make it a Number
There is an analogy between trying to make ophelimité a number (you cannot) and making utilité a number (you cannot). Yet public decisions for EDP crave some persuasive proof that they are Good and you should re-elect Me. You should be satisfied with what your government is doing. But the collective “you” is as much a fallacy as personifying the State as an agent of ophelimité.
The Musgrave’s say a great deal about how to make Public Finance decisions, including for the allocation of the collected tax revenues. They know they are dealing with non-market goods and decisions so the relation to economics of the market is tenuous if not null. But there is accounting, of income and expense. The Federal State is just indifferent to whether that balances.
The decision process about income and expense (and debt) is political, in that medieval indulgence market that is Congress, except when Drill Baby also exceeds the Constitution and starts playing with tariffs. We are in the realm of State Planning that few of the players dare call socialism. But Public Finance is Welfare Economics (about utilité) if economics at all and Socialism. We just argue about who owns the means of ecological exploitation and who gains from it. The gainers argue that regulation of risk and equity is bad. The State blames ecological change on “what we choose”.
The relation of the State to the market-democracy is purely an embarrassment. It need not be. To preserve the economic and political form is not a Public Finance function but it is a formal risk management and equity function of the State as (I claim) the Constitution intended. The prior post argued that was “simply” a matter of complexity: The State is just not smart enough to be a super agent in the economy or any matter of accountable action under a social, and more generally, ecological form. The State can be concerned with preserving a form of law and justice, a more perfect Union. It should not incite polarization. Talk about counterfactual!
It was also argued in the prior post that the false response by the State to complexity is the fragmentation of formal policy into projects. This has the odd effect of bringing the State closer and closer to a personal decision to, say, build a backyard patio. It is just that a highway uses more paving. Is it not what you would do to increase your ophelimité? No. The fallacy is the crossing of scale categories from the contradictory social collective to the pseudo-individual.
This fallacy is compounded by trying to make utilité into a number. What number? GDP would do. But the economic process that produces GDP is complex. If the State confidently knew how to increase GDP, it would do so. The most naive approach is to reduce taxes. But that is not historically how we maximized growth [Robert J. Gordon, The Rise and Fall of American Growth, 2016]. It increases inequity and does nothing to mitigate risk. And that is just incompetence about GDP and let alone the complete ecological “value” that is its stable form.
What really happens when we alter the TOD to the AOS? Is the GDP greater in 2025 than it was in 1920? So much changed over time that such a comparison cannot be made, even for that GDP partial-accounting number. There is no super-formal invariant to make the comparison with. As far as the complete ecological state goes, that is no number at all. The ecology is many things interacting, many incommensurate measures and many invisible (and often conflicting) ophelimités.
The complexity forces any analysis into that same smallness that projects become. There have been attempts to do macro analyses of the highway program (see LTM Part III) and they basically share the problem of “compared to what”. The problem is the inverse of the prediction problem discussed previously. They lack a statistical form that meets the criterion (ergodicity that is an invariant form) for any valid conclusion over a large space-time sample. We rather get conclusions like: “Oh look, the economic activity increased at the freeway intersection” ignoring how the old CBD is doing. And the auto drivers in this access change are better off are they not? Yes, until they congest themselves in that part of the ecology that is not even second-best regulated.
Little, Uncertain, Inequitable non-dollar Numbers
B-C analysis can be no better than the smallness of the project it applies to. It is a matter of scale but also a matter of making several incommensurate outcomes of a decision into a dollar number. Other metrics have been proposed—such as entropy or energy—but that does not solve the problem that the state of any ecology has many dimensions and future outcomes are uncertain. The larger the scale of analysis attempted the more the complete ecology produces the outcomes and that is “complex”.
If scale is the space-time extent of outcomes—and so related to the power of the action—then the problems of quantifying utilité are about how many kinds of objects are affected by a project over what time, and then how those objects “decide” to interact in a changing form. The spatial dimension spans inequity among the objects affected and defies a Pareto Optimum. The time dimension is both an inequity between effects now and later, and an uncertainty of what the outcome will be given all the changes that enter over time (e.g., 1920 to 2025).
Now if you are ready for two impossible things before breakfast, let us predict the outcomes over space-time and map all the different measures of the outcomes into one measure, like dollars. Dollars seem “natural” only because they suggest the tokens of the economy. They are not “value” even in the economic sense but they suggest a model: The model of an investor who expects a dollar return on a dollar investment (RoI). The State is an accountable economic actor. No. Even highway engineers, if they work with a notional benefit in traffic flow via level of service (LOS) analysis, are not experiencing that benefit that is collective to some segment of traffic. And what will that traffic subsequently do? Congest itself for lack of an economic form. (The subtleties of induced traffic were discussed in the post on Assignment or LTM Part III).
Investment is associated with risk, and that is just what happens with a variety of future outcomes. Equity however is inherent. The economic sense of equity is having a stake in some enterprise and that belongs to a real agent accountable for the RoI. Equity as a distribution of outcomes (benefits and costs if so quantified) is taken care of by the identity of the equity-holder with the recipient of the RoI. There is no “distribution” because the investors will have one definite outcome. Only the risk over many investors is statistical.
Incidentally, a previous post discussed the financial handicap of railroads because of their accountability to fixed plant versus “free” highways. The question of railroad projects (including some State EDP) about what service where is overshadowed by the fact that railroads were financially accountable to equity. And that is how they failed against the State’s financial unaccountability. Project B-C of any sort does not redress that inequity.
Yet even the uniquely accountable investor is spread over time. That is where discounting comes in. It is supposed that a dollar later is worth less than a dollar now. That may be true in a growing, non-equilibrium economy. But in the aggregate utilité there is inequity over persons (or things) and the future persons (or things) will not agree that their outcomes are worth less than those of the people who exploited the ecology earlier. Equity becomes the issue with scale along with risk.
Much methodology is devoted to how you collapse many dimensions of measure into dollars. Alright, how much would you pay to have the TOD and trolleys back? I don’t know because I have no idea what they were worth back then and they are not something on a market shelf now.
As B-C is scaled beyond the little confines of a little project, or real investment, the uncertainty of the outcome measures, the uncertainty of their dollar-valuation and the inequity increases. There is inequitable risk and that is exactly the result of EDP. B-C does not mitigate that. It does give a false sense of objectivity especially when the risk of the predictions is unstated (as is often the case). But who understands probability and statistics? More is Better and projects reduce to that promise. But what for whom?
There are politicians who would prefer to shed their accountability for decisions by appealing to some algorithms that give a positive B-C number (with discounting we presume). At best that adds up to a set of incommensurate decisions, but biased toward those that have a real economic payoff to someone. And that ties back into the political-economic power loop of inequity, regardless of ecological outcome. Like now.
When is B-C Right?
The answer to that is also in scaling, toward the small rather than toward the EDP that is powerful enough to change ecological form.
I am not cataloguing here all the critique of B-C method, a lot of which is in LTM Part III. Mishan in particular has a nice example of how even two parallel and competing transport projects are formally coupled and cannot be analyzed in part. The management of the ecology cannot be in part and cannot be reduced to B-C numbers. Individually we do not attempt that management. The State cannot be allowed to do it incompetently.
I also take particular aim at the doctrine (promoted by the Musgraves but deep in welfare economics) that B-C as an “efficiency” measure must not be constrained by equity. Then failure to attain a Pareto Optimum is no bar to a utilité decision. There are three layers of fallacy to this doctrine:
The axiom that dismisses equity as a policy criterion.
That efficiency, inevitably as some project to consume the ecology, is a valid measure for the State, rather than leading to a tragedy of the commons (TOTC).
Ignoring that particular efficiencies (mobility and bandwidth) are also parameters of ecological instability (the primary topic of LTM and previous posts).
The State is accountable for equity, as a matter of justice (a different concept). If not, then we get what we are getting and who evaluates that? The rich or the poor and homeless? Who runs the State? Oh, I see. The doctrine is really about inequitable efficiency. We get from the gainers that “all boats float” (known as the potential-Pareto optimum if, optionally, there is State redistribution, unlike now).
As a onetime B-C analyst am I guilty of gross deception and indifference to inequity? Here is where I enter the ambiguous domain between a pure dollar investment (private finance) and a project that can be bounded so that its outcomes are reasonably certain, reasonably financial and reasonably equitable. But I add: Doing the analysis of outcomes also informs the design of the project.
I would say that my analysis examples meet the criteria. In my own defense. But let me recall the ATC modernization analysis. We (at the Mitre corporation, working for the FAA) did two phases of the analysis ca. 1985 and 1989. In the latter I was also liaison to the Congressional Budget Office (CBO) that was reviewing the analysis. We showed a substantially positive B-C and RoI, discounted and with risk analysis on costs. Amazingly, the cost prediction from the risk analysis (taking the mean) was quite close to the escalating project cost estimate of 1994. And then Congress cancelled the project. That was moaned about previously and see my memoir AERA 3 and the NAS Plan.
The NAS Plan decision by Congress is some proof of the divergence between politics, public finance and project B-C. It also shows that politicians keep moaning about things they are responsible for. But if you accept ATC as a formal regulation of air commerce the issue is not a formal change but a maintenance of that form. The failure of the State in that, from “decaying infrastructure” to indifference to ecological collapse, says something about the State in complexity more than any B-C critique.
For me, doing the B-C analysis was a way of relating the state of many parts of a system to system performance and outcomes, in this case mainly in flight safety and efficiency. Given such metrics we really have a cost-effectiveness analysis. If you know what you have to do, do it efficiently. But that is a narrow criterion, emulating our own efficient behavior, to get what we can in a Pareto Optimum. There is a valid issue of just how much air transport we should have, and that is a matter of further ecological regulation in our fly-drive AOS. We then leap from any possible B-C scope to the hierarchical levels of governance and their responsibilities which is what LTM is about.
I conclude there is a valid domain of project analysis. That must include the people responsible for the risk of any system operation. I made that argument in a previous post on the participatory ConOps in system engineering. Highway analysts use primarily a LOS metric and meeting some LOS criterion becomes a cost-effectiveness issue. I have plenty of complaints about the ecological indifference of such a scope of efficiency but that leads to a formal scale of urban governance that we lack. In particular we have no urban-regional governance to supplant the MPO as the transport-project programming entity (that is not by a B-C analysis anyway). The region just fragments into little, efficient, cost-effective segments and no one regulates the risk and equity of urban form. There is no accountability for what happens because it is blamed on “what we want”. But that is a TOTC, not a Pareto Optimum in an economic form. Or are you confusing the ecology for Amazon?
I conclude with where B-C is certainly right: I invest X dollars, as in my retirement accounts, and I expect a dollar RoI that keeps me alive (and in books). I chose the investments, I have the equity, and my prediction is good, so far. But I am accountable and I learned about risk management from my career, including the B-C analyses. How does the State learn to manage its endowment?