I seriously want to know how it is government thinks it can control tarmac delays at airports.

The federal government will impose stiff penalties starting this spring on airlines that keep passengers waiting too long on the tarmac without feeding them or letting them off the plane � a remedy that will relieve many travelers but mean longer delays for a few.

...Under the rule, airlines that do not provide food and water after two hours or a chance to disembark after three hours will face penalties of $27,500 a passenger, the secretary of transportation announced on Monday.

In recent years, relatively few flights have been held on the ground for more than three hours � about 1,500 a year, or roughly one out of 6,200 flights � but that has been enough to affect more than 100,000 passengers a year and to create substantial public resentment.

�This is President Obama�s Passenger Bill of Rights,� said the secretary, Ray LaHood, using the term favored by proponents of like-minded legislation that is before Congress. The administration�s action does not require Congressional approval.

The right granted for disembarkment to passenger 1 is to make the time until takeoff longer for passenger 2 who decides to stay on the plane. Turning the plane around involves not only going back to the gate (if such gate is still empty; there's the possibility of fewer flights in order to have disembarkment points) and then, in the winter in northern parts of the US, a second de-icing of the plane. Air traffic controllers can only waive the rule, the DOT's press release says, if "for safety or security or if air traffic control advises the pilot in command that returning to the terminal would disrupt airport operations." I have no idea how big a carve-out that is. The law also applies to international flights.

The cost-benefit analysis of this is online; it says ExpressJet, the proposed new carrier to St. Cloud, would bear costs of $1.6 million per year for compliance. (It's the largest carrier of the smaller jet companies with planes in the 30-60 passenger class, the perpetrator of the Rochester overnight delay last August, and largest of the bunch.) Overall the cost of this program will be $100 million to the airlines

The government's delivery of these rights usurps a proper judicial function of torts. Most delays are the function of weather and air-traffic control, both beyond airline control. 7000 passengers were able to settle for $7.1 million over that Detroit fiasco that Northwest botched after snow clogged runways in 1999. That comes out to about $1000 per person. Will we have too much protection against unpredictable weather in the future to avoid penalties 27 times that much? Will it lead to cancellation of flights? The cost-benefit analysis guesses 2.5% of flights get canceled based on a single study that said if an airport gets closed for one hour for security reasons you get a 5.9% increase in cancellations. (See p. 104.) I find that figure for example a SWAG. There are many more.

Labels: cost-benefit analysis, economics

On the show Saturday I assigned my students some homework, to learn about QALY, a "quality-adjusted-life-year."

We might already have our answer:

In Great Britain, for example, the National Institute for Health and Clinical Excellence (NICE) uses "cost per QALY" to determine if patients should receive expensive treatment or drugs. It was with this formula that NICE calculated the precise amount six months of an average Brit's life is worth. As the Wall Street Journal reports, "NICE currently holds that, except in unusual cases, Britain cannot afford to spend more than about $22,000 to extend a life by six months." In other words, patients whose country has guaranteed them "free" health care are in some cases refused treatment because the incremental cost per additional QALY is too high.

Now, what are those treatments worth? From the British Office of Health Economics, a table from a 1993 article:

Labels: cost-benefit analysis, economics, health care

Washing your car or boat in the driveway or street is a residential ritual as American as backyard barbecues. But the state of Washington is telling its local governments they must prohibit home car washing unless residents divert the wash water away from storm drains, where they say it causes water pollution. 'I understand this is something people have done for a long time,' says Bill Moore, water quality specialist with the Washington state Department of Ecology, which is requiring the ban. 'It's not something we should be doing any longer.'

Source, via the Amateur Economist.

Labels: cost-benefit analysis, transportation

My friend and loyal reader jw sent along an article on the value of a life being used by EPA.

It's not just the American dollar that's losing value. A government agency has decided that an American life isn't worth what it used to be.

The "value of a statistical life" (VSL) is $6.9 million in today's dollars, the Environmental Protection Agency reckoned in May � a drop of nearly $1 million from just five years ago.

The Associated Press discovered the change after a review of cost-benefit analyses over more than a dozen years.

Now where do you get that $6.9 million from? There are two ways traditionally used in cost-benefit analysis. One is to take my estimated income over my lifetime, discount it to present value and come up with the sum. That number is likely to be relatively low, almost certainly less than $6.9 million. (If yours is not, congratulations -- you're a pretty rich fellow!)

The other way to find that out is to look at different jobs with different levels of safety or risk. Figure out the probability of a job fatality for the different occupations, and then compute how much additional pay the average worker receives in the riskier job. For example, one of the job opportunities I might have is to work as an adviser to, say, the central bank of Afghanistan. How much more would you have to pay me to go there versus, for sake of comparison, the central bank in Mongolia? Afghanistan is riskier, and if we can figure out how much riskier it is and compare it to the wage differential required to hire economists into both positions we have some measure of the value of a life. Indeed, this is what EPA is using:

...economists calculate the value based on what people are willing to pay to avoid certain risks, and on how much extra employers pay their workers to take on additional risks. Most of the data is drawn from payroll statistics; some comes from opinion surveys. According to the EPA, people shouldn't think of the number as a price tag on a life.

The EPA made the changes in two steps. First, in 2004, the agency cut the estimated value of a life by 8 percent. Then, in a rule governing train and boat air pollution this May, the agency took away the normal adjustment for one year's inflation. Between the two changes, the value of a life fell 11 percent, based on today's dollar.

EPA officials say the adjustment was not significant and was based on better economic studies. The reduction reflects consumer preferences, said Al McGartland, director of EPA's office of policy, economics and innovation.

"It's our best estimate of what consumers are willing to pay to reduce similar risks to their own lives," McGartland said.

But the EPA's cut "doesn't make sense," said Vanderbilt University economist Kip Viscusi. The EPA partly based its reduction on his work. "As people become more affluent, the value of statistical lives go up as well. It has to." Viscusi also said no study has shown that Americans are less willing to pay to reduce risks.

At the same time that the EPA was trimming the value of life, the Department of Transportation twice raised its life value figure. But its number is still lower than the EPA's.

There's no good reason for them to use different numbers, so that there has been convergence between DoT and EPA should be considered a good thing, caterwauling by liberal blogs notwithstanding. Here's a white paper from 2004 that EPA has posted that describes their study of VSL. Part of the problem is that people sort themselves into jobs in part depending on the attitudes towards risk. The report takes, for example, the observation that night clerks at convenience stores tend to be older and male. Older individuals tend to make more because of experience; males have been noted to earn more than females. Should all of the difference between the pay the night clerk and the morning clerk receive be attributed to the greater chance of armed robbery at night?

Interestingly, that paper includes a study done by Viscusi, which puts VSL at $6.9 million in one estimate. You would need to adjust that for inflation from 2003.

The implications of this for the drilling debate should be obvious. We have on one side a desire to reduce the price of gasoline and other energy products, which clearly increases the welfare of our citizens. Against that we have to weight the cost to the environment, which might include the loss of wildlife. What is the demand for wildlife? If you look at it in market terms, the location of ANWR matters as it provides less value to tourism than a similar-sized area on the California coast. We can compute the value of ANWR or any other place by looking at how much people will pay to travel there, and how many people do so. Take for example this 2001 piece that says 2500 people travel there and pay $4000 each to go. That's $10 million a year; discounted at 5% in perpetuity says the value of tourism in ANWR is $200 million. Now that number is almost assuredly too low -- there must be other tours. But as well, tourism would not fall to zero if rigs were put there. If we assume the tourists are there to see caribou, and if caribou will remain after the rigs are put in, some fraction of the 2500 will still go. A survey might tell you how many.

The benefits are likely to be large; sure, it might take years to get production (though Larry Kudlow (h/t: Dave) points out that in California, one estimate says you get drilling within one year), but that is an easy correction to the spreadsheet on which you do the cost benefit analysis.

Democrats have ignored tradeoffs on energy for years, and a few Republicans have joined them at least until now. A format exists to make sound calcuations about the tradeoffs. It is good to see liberals considering the meaning of VSL. Now to get consistent application of the method...

Labels: cost-benefit analysis, economics, Energy

Governor Pawlenty's bond proposal came out this morning, and the St. Cloud Times headlines what is not in the bond:

It does not include requested state dollars to expand the St. Cloud Civic Center or to remodel and expand the National Hockey Center at St. Cloud State...

...two major projects local leaders have had on their minds for quite some time. How long until we hear Sen. "No-no-no", who has in the past pumped for more for the hockey center, has another fit about Governor Pawlenty's thrifty ways?

SCSU is proposed for money for Brown Hall, an old classroom building that has laboratories as old as me. (Rep. Steve Gottwalt told me about touring the facility and the faces of the legislators as they walked through; I think they keep one lab particularly antiquated and nasty-looking for the tour.) In the note from the faculty union reporting this to us, the lobbyist noted,

Concern over neglected transportation infrastructure is causing transportation funding to squeeze other segments of the capital investment bill. This means we have a stake in the outcome of the gas tax debate�if roads and bridges are funded out of a gas tax increase it would mean more bonding authority would be left over for projects such as higher education projects.

I wanted to emphasize this point: The gas tax for funding transportation is not used instead of bonds (or in the slang used by DFL legislators, "the credit card") but in addition to bonds. We will pay just as much interest either way -- we'll just have less private investment increasing productivity with which we pay off those bonds with our future taxes. The gas tax makes government bigger; the discussion of the means of funding is just smoke. And it is highly unlikely the Legislature adjourns and the House goes to the voters without a bunch of money spent on bridges; the demand is still there.

There's plenty not to like about the governor's bond request -- I might start with $70 million more down the Central Corridor rat-hole -- but funding of bridges through bonding is certainly something to like. The "benefit principle" of public finance says that the people receiving the most benefit from a public good should be the ones that pay. That's one reason why we ask people to pay fees to enter state parks, for example. The alternative principle is "ability to pay." But in either case bonding can be the preferred option. Bridges last 30-40 years or more, and there is no reason why we should expect the current taxpayers to fund the driving of individuals using the bridge a generation from now; bonding assures that those driving the bridges are paying for them. And if investments in infrastructure like bridges are to increase state productivity, then the income of future generations will be higher than those in the present. If it is enough higher, the tax burden will be lighter on future than current generations, again arguing that some of the cost be shifted forward. Under any reasonable assumptions about the value of future Minnesotans' utility or satisfaction versus the value of present Minnesotans' utility, it is good public policy to use bonding. (Of course this may be controversial, as we've been arguing for some time over the rate of discount of future generations in the global warming debate. You'd have to work the math a bit to convince me it matters here.)

Now part of the problem, in my view, stems from what appears to be a formula that translates the size of the state budget to the size of the bonding proposal (which would put the amount planned here at about 3% of the biennial budget.) I had Rep. Larry Haws while I guest-hosted on the KNSI Morning Show a couple of weeks ago in which he made some reference to this; about the only sense I could make of this -- assuming there is some formula they use -- is that it keeps the rating agencies happy so that the interest cost of debt stays predictable. But one could easily imagine that cost-benefit analysis could be applied to the bridges to argue for an amount of bonding above this formula, as long as the return on investment was sufficient to cover the opportunity costs. Much like Nixon-to-China, it may be up to the DFL to come up with a way to use cost-benefit judiciously to make the case for a larger bond.

UPDATE: Lileks wonders about light rail. See moreover this from the LA Times (h/t: Peter Gordon):

Paradoxically, the MTA's rail projects, which required fare increases and reduced bus services, have cost the transit system riders. Using MTA data, our analysis indicates that they produced a drop in train and bus ridership of more than 3 billion boardings from 1986 to 2007.

Although we've now gotten back to 1985 levels in terms of public-transit use, the county population has grown by more than 2 million since then. That means, on a trip-per-capita basis, the transit system is still not performing -- by 20% -- as well as it did 22 years ago.

Labels: cost-benefit analysis, economics, legislature, Minnesota, Pawlenty

I first bought a microwave when Mrs. and I were married. I remember driving to several stores, speaking to salespeople, weighing characteristics carefully, then buying it. It was a substantial purchase for us (I'm quite sure it cost more than $200, and this was around 1988.) It traveled with us for about eight years.

About the time it failed, we bought our house. In the cabinetry of the kitchen is a space for a microwave, and one was there similar to ours. We wanted that one because it fit the space. The house was about ten years old, and it turns out so was the microwave. The seller was incredulous that we wanted the microwave but relented.

A year later, the microwave stopped working. We went to one store (the one with two words beginning with 'B') and talked with two salespeople. Fretted over it for about an hour, paid about $150 for a decent model, and took it home.

I then never thought about it again.

Last night I came home and both Mrs. and Littlest had sad looks on their face. The microwave did not work. I examined it and sure enough it did not work. "Do we get it fixed?" Simple cost-benefit analysis question, I replied. Ate a relatively cold late dinner and then went to the store and bought same. Opened laptop, looked at the same store's website. Saw nice one for $100. Took Littlest, bought same, came home and plugged it in.

Conclusions: Microwaves are an example of one-hoss shay depreciation. When my long-life lightbulb croaks, I am out some money, and then I have light. I get the same service from the microwave each time until it stops working, then I have zero service. On a rather difficult day, the $100 is a loss, but the decision is pretty simple: You no longer own a microwave; they cost $100; want one?

Amazing what a bag of microwave popcorn will get you thinking about. I had the same popcorn Sunday and Monday night. Was the cost of Monday's popcorn $101 ($1 bag + $100 oven)?

Question: Is a bridge a one-hoss shay? We don't create sinking funds for replacements of lightbulbs or microwaves because you typically have enough money to replace them in a ready reserve. Poop happens. But we do for larger expenses like bridges. It has to be tempting to raid that fund. But when a bridge falls it's the same choice. You don't own a bridge; they cost $250 million; want one?

I think about this in terms of considering what the value of public capital is. What does it mean to depreciate a bridge or a public building? Do they generate less value as inputs to production as they age?

Labels: cost-benefit analysis, economics

I'm afraid that in the aftermath of the bridge collapse, cost-benefit analysis is getting a dirty word. Let us begin by pointing to this comment on a New York Times blog discussion:

The entire 1000 foot long section was tied together structurally to save money. It had no tolerance for partial failure. If one section failed, the entire section would go down. This more modern bridge was ugly as well as a poor design. This bridge was designed by modern engineers who have no sense of beauty and think they can calculate every decision on the basis of cost/benefit. They practice a destructive type of design called value engineering - taking out the expensive stuff if it�s redundant or optional.

Or take Joel Hirschhorn at the Daily Kos:

I can absolutely guarantee that there were countless discussions over the years by engineers, bureaucrats and politicians that combined risk assessment and cost-benefit types of thinking. It comes down to this: To bite the bullet and conclude that to provide maximum protection of public safety the bridge should be replaced would inevitably face debates about the incredibly high costs and the enormous difficulties of obtaining the funding, and also how obtaining such funding would imperil other government projects.

So the decision moves in the direction of higher levels of inspection to postpone the inevitable high cost/low risk scenario. On the other hand, all people connected to these kinds of discussions know one big reality: If the crap hits the fan and there really is a catastrophic bridge failure, then the money WILL be readily available to replace it! This is the way the system works. Of course, to replace the bridge and deal with all the many horrific impacts of the bridge failure AFTER the fact will cost much, much more money than if a planned replacement strategy had been adopted!. But that is exactly what the Minneapolis story ultimately is all about. It is what virtually all of our national infrastructure thinking is about. The New Orleans disaster was totally preventable, as every technically sound and objective analysis showed.

Which assumes, to start, that every preventable disaster should be prevented. Should it? Can we prevent them all? Again, at what cost? Compared to what? How do you know?

So what does cost-benefit analysis (CBA) ask the policymaker to do? Simply put, it totes up all the benefits of a policy change, such as bridge replacement or a different bridge inspection program, against the costs, all put into a common metric (most often, net present value.) Thayer Watkins of San Jose State provides a useful tutorial. Ed Gramlich, whose book is for me the standard of the profession, spoke a few years ago about where it can and cannot be used:

It can be quite helpful when relevant markets exist, when market or nonmarket techniques for valuing inputs and outputs have been developed, when scientific uncertainties are limited, and when time periods are not inordinately long. Should any of these conditions not be present, one still has ways of proceeding; but the methodology becomes more speculative, and the uncertainty band grows. One can still factor uncertainty into the analysis by providing probability distributions of net benefits from some policy change, but the resulting probability distributions are likely to be very wide. This uncertainty makes it difficult for analysts to give programmatic recommendations with any degree of confidence, and difficult to compete with special interests that have very clear ideas of their valuations of programmatic outcomes.

As a result, benefit-cost analysis is more or less useful in providing a framework for policy decisions, depending on the underlying properties of these decisions. If markets and valuation methods are clear and time periods are relatively limited, as with river and harbor projects, subsidy measures, tax or tariff changes, benefit-cost analysis is usually able to frame public decisions pretty clearly. But for programs involving research, climate change, or other types of extreme scientific uncertainty, decisions are framed much less clearly. Even in these areas, benefit-cost analysis has value, as I will try to illustrate, but one must be very careful to ensure that the results of the analysis are not misused.

Here's a rather simple example related to roads from John Quiggin, who is arguing whether to use cameras on roads to limit speeding on a highway:

People who formerly travelled at above the speed limit will go slower and take more time. On the other hand, since this will reduce both average speed and speed variance, there will be less accidents.

For the costs, I�ve assumed 15 000 cars per year, 20 per cent of whom speed consistently, maintaining an average of 130km/h (vs a limit of 110). I�ve given them a value of time saved of $20/hour (higher than is standard), and I estimate an annual cost of $10 million per year from enforcing the limit.

As already noted, the cost per life lost is above $20 million, between $5 million and $10 million so we only need to prevent one or two fatalities per year to get benefits>costs (there have been about 8 deaths per year in the last five years).

So too with a cost-benefit analysis of bridge inspection. You cost out the increased travel time and increased probability of accidents in a narrowed road environment (or perhaps using detours, whichever is better) versus the benefit from the information gained by additional inspections. Those benefits, of course, are subjective; it certainly was higher on Friday when they inspected the De Soto bridge here in St. Cloud than it would have been last Tuesday. Still, do you see anything wrong with that decisionmaking rule? Charlie Quimby does not:

We all make judgments about how long to keep pouring money into our cars. If you're like me, you take good care of your car but drive it for as long as you can. At some point, it clearly becomes structurally deficient � burning some oil, quarter panels rusty, on its third set of rotors, muffler corroded, maybe a crack in the windshield. But it's driveable, safe and still far less costly to operate than a new car with a monthly payment.

I once drove a rotary engine Mazda until it caught fire for the third time. It was a snowy winter, after all, and the blazes on the engine block could be quickly extinguished, with no apparent effect on other functions of the car.

I'm no economist or bridge inspector, but after the third fire, I could see it was getting to be time for replacement.

We should not blame MN/DOT for its analysis, assuming no grievous misjudgments were made about bridge condition. But we should revisit the question of where we want to be as state on scale of new car to flaming Mazda.

Which is fair criticism. Any CBA will have a set of sensitivity analyses that demonstrate the conditions under which a project -- perhaps more frequent bridge inspections, or bridge replacement -- will either be supported or rejected. You and I can look at that sensitivity and come to different decisions. CBA doesn't give you just one answer; it gives you a framework for rational analysis. And to decide ex post that those who looked at the CBA and decided the more expensive projects for shoring up the bridge were not just wrong but malign is a hindsight fallacy.

But the criticism of Hirschhorn is still in need of answering: Did the MnDOT inspectors lean on the scales to favor visual inspection in order to avoid having to go to the boss and say the bridge was unsafe?

To that end it's instructive to note the authorship of the 2001 and 2005 reports on the bridge's safety were not from MnDOT. These independent assessments go exactly to Gramlich's recommendation that government officials doing CBA need to use independent agencies to verify their work. The University of Minnesota, which did the 2001 assessment, did not have incentives to fiddle the numbers.

What I suspect is that the complaints are not about CBA itself but about the budget process, which reinforces CBA just as CBA reinforces it. Gramlich writes:

The federal budget process is one of the most criticized governmental processes around, with observers routinely pointing to projects that should have been, but were not, screened out. But the federal spending and budget process seem to help benefit-cost analyses of projects. The budget imposes discipline, if imperfectly. When projects threaten to become too costly, as the SSC [superconducting super collider --kb] did and dredging the Delaware River may be, OMB [the Office of Management and Budget] and other budget cutters can be expected to join the fray and often support the results of the benefit-cost analyses.

Indeed, the relationship can be symbiotic. OMB needs benefit-cost analysis to weed out projects (otherwise they would be acting in a peremptory manner). Benefit-cost analysts often welcome the sheer political heft of OMB. Those opposing the dredging of the Delaware River will find stopping the project easier with the help of OMB than without it--just as stopping the SSC was easier once costs started escalating rapidly in an era of increasingly tight budgets.

Yes, "increasingly tight budgets." In any cost-benefit analysis one must impose a cost of funds used for a project. What are the alternative possibilities for the funds to be used? For example, the Sunday paper here carried a story of fifteen bridges in the area that are "structurally deficient". Some bridges are seventy years old. Yet under current circumstances and with current traffic flows, there are other projects more in need of help.

The detractors of CBA seem to live in a world where tradeoffs do not exist. To respond to the first comment, 'beauty' is a good that competes for resources with other goods, like education or meals on wheels. So too with safety. As "vinod" notes on his blog, Thomas Sowell's "tragic vision" means that libertarians and conservatives understand tradeoffs are real, while the anointed dogs believe that "problems must be solved in totality with 'no one left behind' and 'no one left at risk'." Even in a world with a few black and gray swans, you can't fix everything.

I have a few loose ends to tie up, to wit:

• how do you factor the black swans into CBA?
• "isn't it really heartless of you to price the value of a life in these calculations, King?"

In case other comments are needed, I'll hold that discussion off until tomorrow.

Labels: cost-benefit analysis, economics, Minnesota