david pritchard. bibliography.

Notes on Anthony Downs, Still Stuck in Traffic [3]

This was quite an interesting book, and came well-recommended. I suspect that most people read this book before the very similar "The Transportation/Land Use Connection" [9] by Moore and Thorsnes, but I read them in the opposite order. See my comments on that report for a general overview of the issues covered here. My comments here will mostly contrast Downs' approach with Moore and Thorsnes. I read the original edition of the book (Stuck in Traffic), and have only skimmed the new edition. I've mixed quotes from both editions here, with a reference to the page numbers in the other edition where possible.

The first thing that struck me about this book was the focus. Downs, an economist, aims this at the policymaker who wants to tackle traffic congestion. His first priority is not environmental, nor does he push (directly) for a change in modal split by including a "full-cost" accounting of the costs of the automobile. (Todd Litman, by contrast, seems much more concerned with the latter problem. More on this later.) Instead, all of his policy recommendations must be seen with respect to the goal of reducing traffic congestion. Naturally, many of his policy recommendations do have positive environmental impacts, and would decrease automobile usage. I found Levine and Garb's paper [7] useful for making these assumptions more explicit.

Second, his focus is on pragmatic policies for the American context. He is generally dismissive of transit-based solutions, mostly because the existing land-use in most American cities makes transit impossible in the near future. In this sense, his policy recommendations are not so useful in my context (Canada), where land uses are much denser and transit-friendly than in the USA. His pragmatic approach, however, is welcome.

Finally, unlike Moore and Thorsnes, Downs does pay attention to equity, which I definitely appreciate. At times, he seems a little cavalier in his approach, however—I don't buy the argument that pro-transit policies harm the poor. I suspect that he's sometimes a little selective with his statistics. On the other hand, I'm sure that transit projects do need to be evaluated on a case-by-case basis—major investments in subways can easily have regressive effects in some places, I'm sure. (For example, if the investment starves the bus system's funding, the poor are likely to be negatively impacted overall.)

Given these caveats, his analysis is well reasoned and fair. I don't agree with all of his recommendations, but that's because of differences between us; I am more interested in changing the modal split than I am in decongesting roadways, and I live in Canada not the United States. His decision to focus on congestion has advantages and drawbacks. On the one hand congestion is the main problem that most policymakers see, and by keeping a tight focus on this problem, he avoids the possibility of being labelled as an environmentalist. On the other hand, he quickly acknowledges that congestion is not really a solvable problem, at least while roadways remain free. Consequently, he rejects whole swaths of useful solutions (transit investment, major land use change, etc.) since they won't solve congestion, even if they might improve sustainability, equity, energy use or greenhouse gas emissions.

I did appreciate the updates in his new edition. Clearly, he has read some of his critics and adjusted his attitude over time.

The Texas Transportation Institute estimated that congestion "wasted" $67.5 billion dollars in seventy-five metropolitan areas during 2000 because of extra time lost and fuel consumed, or $505 per person, compared with what would have happened without congestion. Time lost in delays (at $12.85 an hour) accounted for about 68.5 percent of that estimated total cost; the rest was fuel costs.

In reality, these social cost estimates are based on a false premise: that peak-hour travel in these regions could have been accomplished without any congestion if only society had better policies. As explained in chapter 2, modern societies are organized in such a way that so many people need to travel during peak hours, morning and evening, that no feasible arrangements or policies could accommodate them all without significant delays. In short, a major amount of daily peak-hour traffic congestion is inescapable in every large metropolitan area in the world. Therefore, it is unrealistic to conclude that all the "excess travel time" experienced during peak hours versus nonpeak times when no congestion exists could ever be eliminated—and is therefore "wasted" because of ineffective policies. The hypothetical alternative of "congestion-free" travel during peak hours is an unattainable myth. So comparing that illusory alternative with what happens and declaring the time difference "wasted" is a misleading exercise.

Furthermore, although the Texas Transportation Institute's estimates of congestion costs appear almost staggering when aggregated over an entire year, they seem much smaller when they are viewed on a daily basis. If there are 240 working days in a year, and each worker makes two commuting trips per day, and all congestion costs computed by the Texas Transportation Institute are allocated to the resulting 480 trips, then the estimated cost of congestion per person is $1.05 per trip, of which $0.72 is in time and $0.34 in cash. An annual average loss of thirty-six hours in delay over the same 480 trips is only 4.5 extra minutes per commuting trip each day. These costs seem much more bearable than the aggregated figures that are usually quoted in analysis of traffic congestion.

[[3] pp. 2-3]

As noted earlier, TTI's whole analysis of the "costs" of congestion is based on the false premise that there is some alternative state of the world in which everyone who wants to travel during peak hours could do so freely without delays. As British traffic expert P.B. Goodwin pointed out, "I cannot endorse statements of the form `congestion costs the economy 15 billion pounds per year, updated from time to time by inflation,' implying an annual dividend of 1,000 pounds waiting to be distributed to each family. This is a convenient, consensual fiction. It is calculated by comparing the time spent in traffic now, with the reduced time that would apply if the same volume of traffic was all traveling at free-flow speed, and then giving these notional time savings the same cash value that we currently apply to the odd minutes saved by transport improvements. This is a pure, internally inconsistent notion that can never exist in the real world. (If all traffic traveled at free-flow speed, we can be quite certain that there would be more of it, at least part of the time saved would be spent on further travel, and further changes would be triggered whose value is an explored quantity.)" [6]

[[3] p. 24]

The principal complaint against market-based strategies is that they put undue stress on low-income households and hence are economically regressive and inequitable. Such households are less able to pay the prices imposed than are higher-income households. Some arguments have been put forth to counter this charge, but they have not been entirely convincing.

The Bay Area Economic Forum of San Francisco has stated that many low-income workers commute "against the tide" since they live near downtowns and work farther out. Thus they would not have to pay congestion fees traveling to work. But this argument is not consistent with the commuting times of different income groups. In 1983 the average commuting time of households with incomes less than $10,000 was only 18.8 minutes—the shortest of any income group. Yet the average commuting time for workers living in central cities and working in the suburbs was 26.4 minutes. It appears that most low-income workers do not live in central cities and work in the suburbs.

Another point made is that the many low-income workers who already commute by bus would not have to pay congestion prices either but would benefit from the results. The question is, how many do use the bus? In 1983, 77.6 percent of all workers with household incomes below $10,000 commuted by private vehicle, whereas only 6.9 percent used public transportation. Even the suggestion that the money raised by road pricing could be used to improve transportation facilities used by low-income workers is questionable. A high percentage of adults commuting by public transit come from households with incomes of $20,000 or more—the figure in 1983 was 56.1 percent. Unless the funds from peak-hour tolls can be used to compensate low-income drivers directly, road pricing may have regressive effects.

[[2, p. 26]; also [3, pp. 80-81]]

I find the above passage very questionable. The most objectionable argument is in the third paragraph. How can 77.6% of workers with incomes below $10,000 afford to drive? Off the top of my head, the typical cost of operating an automobile is about US$5,000 in current dollars; for a below-average car and 1983 dollars, it's still probably $3,500. How can a household afford to spend 35% of its income on mobility? Carpooling is one answer, but it's still far from a good situation for the poor. The argument also hinges on current statistics, which are a consequence of the mobility-centric American transportation policy of the last several decades. In all likelihood, this statistic is merely a reflection of the absence of choice for the working poor. I wouldn't be surprised if many of that 77 percent would choose a different mode if it offered reasonable service.

To my intuition, the argument that public transit is mostly used by the middle class reflects the fact that good public transit is only available in larger, older, wealthier cities. How much do New York, Chicago and Boston alone distort this statistic?

Finally, the argument in the second paragraph seems to be mixing national statistics with urban statistics—a questionable argument, to my mind, and a very indirect measure of the commuting patterns of the poor.

Triple Convergence

Most vehicle drivers search for the quickest route, one that is shorter or less encumbered by obstacles (such as traffic signals or cross streets) than most other routes. These direct routes are usually limited-access roads (freeways, expressways, or beltways) that are faster than local streets if they are not congested. Since most drivers know this, they converge on the "best" routes from many points of origin.

During peak travel hours on weekdays, so many drivers converge on these best routes that they become congested, particularly in large metropolitan areas. Traffic on them eventually slows to the point where they have no advantage over the alternative routes. That is, a rough equilibrium is reached, which means that many drivers can get to their destinations just as fast on other roads. At times, the direct road may become even slower than alternative streets, and some drivers eager to save time will switch to them. Soon rough equality of travel times on both types of routes is restored at the margin. The opposite happens if travel becomes slower on alternative routes than on the expressway.

Several observations can be made about this equilibrium situation: it tends to recur, because most drivers develop habitual travel patterns; during equilibrium each limited-access road is carrying more vehicles per hour than each normal city street or arterial route because it has more lanes, more direct routing, and fewer obstacles; many drivers time their journeys to miss these periods because they do not like to waste time in heavy traffic; and at the peak of equilibrium, traffic on most expressways is crawling along at a pace far below the optimal speed for those roads.

Now suppose that the limited-access route undergoes a vast improvement—for example, its four lanes are expanded to eight. Once its carrying capacity is thus increased, the drivers using it move much faster than those using alternative routes. But this disequilibrium does not last long because word soon gets around that conditions on the expressway are superior.

In response, three types of convergence occur on the improved expressway: many drivers who formerly used alternative routes during peak hours switch to the improved expressway (spatial convergence); many drivers who formerly traveled just before or after the peak hours start traveling during those hours (time convergence); and some commuters who used to take public transportation during peak hours now switch to driving, since it has become faster (modal convergence).

This triple convergence causes more and more drivers to use the improved expressway during peak hours. Therefore its traffic volumes keep rising until vehicles are once again moving at a crawl during the peak period. This outcome is almost inescapable if peak-hour traffic was already slow before the highway was improved. If traffic is going faster than a crawl on this direct route at the peak hours, its users will still get to their destinations faster than users of city streets, which are less direct and more encumbered by signals and cross streets. Total travel times on these two types of paths will only become equalized if the limited-access roads are so overloaded that vehicles on them are moving at slower speeds than those on normal streets. Triple convergence creates just such an effect during peak hours.


These effects of triple convergence are short-run impacts because they involve persons who were already traveling each day during, or shortly before or after, peak hours. But there can also be long-run impacts of increasing the capacity of a major roadway. For example, widening an expressway may encourage more intensive propery development in the primary deestination it serves—often a region's central business district. More commuters will arrive at that destination during each hour while encountering the same degree of traffic congestion as before. Hence the road improvement may stimulate more real estate development instead of less congestion, or some combination of reduced congestion and intensified development. This impact clearly takes a considerable time to occur. Another more important long-run impact is that improving a roadway may cause more residents and businesses to locate along it in order to enjoy its upgraded access. These newcomers will then also use the expanded roadway, thereby adding to total traffic on it. This added traffic will offset some of the benefits that the original users of the road hoped to gain from expanding it in the first place. This type of long-run growth in demand caused by improving a roadway is called induced demand, since it was called forth by the roadway improvement. But regardless of whether upgrading a road evokes induced demand in the long run, any upgrading is certain to evoke convergent increases in demand in the short run, as just explained. Such short-run increases in demand caused by convergence are sometimes referred to as induced traffic.

Thus, because of triple convergence, expanding roadway capacity does not fully eliminate peak-hour traffic congestion, or even reduce the intensity of traffic jams during the most crowded periods—although those periods will be shorter. In fact, it is almost impossible to eradicate peak-hour traffic congestion on limited-access roads once it has appeared within a nonshrinking community. In theory, such congestion could be eliminated only if the capacity of those roads were increased to the extent that they could carry every single commuter simultaneously at the peak minute at, say, 35 miles per hour or faster. In nearly all metropolitan areas, that is impossible. Therefore, expansions of road capacity—no matter how large, within the limits of feasibility—cannot fully eliminate periods of crawling along on expressways at frustratingly low speeds.


The Converse of Triple Convergence

The triple convergence principle also operates in reverse. Any factors that increase peak-hour congestion on limited-access roads tend to cause more auto-driving commuters to shift away from those roads in peak periods to the same roads in nonpeak periods, alternative routes during peak periods, and public transit during peak periods. Such triple divergence has important policy implications.


Triple divergence is really an inadvertent form of demand management inherent in all intense congestion. Whenever peak-hour congestion on a roadway gets worse, the resulting decline in the desirability of using the roadway during those hours motivates some drivers to shift to other routes, other times, or other modes. This reduces the demand for that road during peak hours, thereby partially offsetting the worsened congestion until a new equilibrium is reached. Hence congestion is partially self-correcting through such triple divergence. This illustrates the basic nature of congestion as a balancing mechanism between supply and demand.

[[3, pp. 82-86]; also [2, pp. 27-30]]

Transit commuting is concentrated in central cities, especially a few large ones. In 2000, 10.5 percent of all central city workers commuted by public transit, compared with 2.9 percent of all suburban workers and 0.6 percent of all workers living outside metropolitan areas. However, the overall central percentage is distorted by the immense influence of New York City. About 53 percent of its workers use public transit, and they compose 27.8 percent of all U.S. transit commuters and 42.3 percent of all central city transit commuters. In all central cities excluding New York City, transit commuters make up only 6.6 percent of all workers. When New York City is excluded frmo total national data for 2000, the share of all workers over 16 using public transit declines from 4.7 percent to 3.5 percent.

Significant fractions of workers commute by public transit in several other large cities too. Table 9-2 shows the twenty-three cities with the greatest number of transit commuters in 2000. The sixth column lists the percentages of all city workers commuting by public transit. In six cities, including New York City, that fraction exceeded 20 percent; in nine others, it was between 10 and 20 percent. These twenty-three cities contained 50.6 percent of all U.S. transit commuters in 2000, though they encompassed only 10.3 percent of the total number of U.S. workers.

[[3] pp. 122-123]

The actual numbers in the table are dramatic: New York had 1,684,850 commuters in 2000, the number two city (Chicago) had 310,924, and it drops off quickly from there, down to 18,632 in number twenty-three (San Antonio). The total for the 23 cities was 3,069,103.

There's an excellent table showing the percent of household expenditures used for housing and transportation in several metro areas in [3, p. 131].

In theory, there is one exception to the conclusion that expanding transit capacity cannot reduce existing intensive peak-hour traffic congestion. It was pointed out by Martin J.H. Mogridge. However, his analysis only applies to regions in which the vast majority of peak-hour commuting is done on rapid transit systems with separate rights of way. Central London is an example, since in 2001 around 85 percent of all morning peak-period commuters into that area used public transit (including 77 percent on separate rights of way) and only 11 percent used private cars [8]. When peak-hour travel equilibrium has been reached between the subway system and the major commuting roads, then the travel time required for any given trip is roughly equal on both modes.

That must be the case because travelers can shift modes between transit and roads. If movement on the roads generally takes less time, then people will move from the transit system onto the roads. The added road traffic will slow the average speed on the roads—that is, increase travel times for specific journeys. So roads cannot maintain a travel time superiority over transit very long. Even expanding road capacity does not speed up movement on roads permanently. Rather, doing so simply draws more passengers from transit until the roads become so loaded they slow to the same travel time as transit. Since the percentage of travelers using transit is so great, there are always enough transit passengers willing to shift onto roads to swamp any improvement in speed from building more or wider roads. That is what differentiates this situation from conditions prevailing in every American city except perhaps New York City.

However, if travel time (including getting to the station, waiting, and getting from the station at the other end to the final destination) is less prolonged on transit than on roads, people will move from roads onto the transit system. But unlike roads, fixed-rail transit systems with separate rights of way do not become slower because they have more passengers—they only become more crowded. Transit cars get more and more jammed with standing-room-only crowds, but the trains move just as fast. At some point, the transit system becomes packed to maximum capacity during peak hours. An example is the Tokyo subway system, which becomes so crowded that official "pushers" are used to pack more people into each car. Then people stop moving from roads onto transit because there is no more room. (This occurs after the maximum number of trains have been put into service with the shortest possible headways.) At any passenger loading on transit systems below this maximum capacity, there must be a rough equilibrium with roads and transit or else more people would move off roads onto transit.

If such an equilibrium exists initially, but then some improvement of the transit system decreases the average travel time for movement on that system, that will make transit more attractive than roads. People will move off roads onto transit until transit still has lower travel times than roads but is at maximum capacity or transit and roads are back in rough travel time equilibrium. The improvement in the transit system has to be of a kind that reduces travel times to have this effect. Just increasing the number of passengers that the transit system can carry per hour will not make transit more attractive than roads to current road users, unless transit was previously faster than roads but at maximum load capacity.

Mogridge concluded that, in the long run, the speed of movement on the transit system—that is, its travel times—is the key factor deciding the overall speed of all modes in the city. The other modes have to adjust to come into equilibrium with transit's travel times. But overall travel times on the London transit system had not fallen for roughly a century. Mogridge believed that is why London's road traffic also had not changed in speed for about a century, even though a lot of roads had been built or improved.

[[3, pp. 132-134]]

Many people drive to work alone because they are able to park free. The amount they save from free parking is often much greater than the gasoline costs of commuting. Moreover, free parking is an employer-provided tax-free benefit, whereas public transit travel allowances paid by employers are subject to income taxes. If free parking were prohibited, many workers would stop using cars for commuting altogether, and even more would stop driving alone. Five studies have shown that an average of 66 percent of workers drove to work alone when employers provided free parking, but only 39 percent did so after employers ended that benefit. Shifts of solo drivers to ride sharing were much smaller when employers paid commuting allowances to all workers but still provided free parking.

Another way to raise parking costs would be to charge high fees on all vehicles entering parking facilities during morning peak hours such as 6 to 10 a.m. those fees would hit commuting workers but not most shoppers or persons running errands. To be effective, parking surcharges would have to be large. Otherwise they would not tip the balance of net commuting benefits away from driving alone to either ride sharing or using transit. This is clear from the fact that many workers now drive alone into congested downtowns in spite of high parking charges there.

Raising parking costs would be most effective in reducing peak-hour solo commuting where public transit services were readily available. But even if no nearby public transit existed, higher parking costs would surely encourage more ride sharing. Such fees would motivate many solo drivers either to double up or to ride transit to reduce costs, or to travel at other times when parking fees were lower or nil. Hence such fees should reduce peak-hour work trips—but they have to be substantial enough and imposed on all worker parking spaces.

There are three significant differences between charging high peak-hour prices on roads and on parking. First, it would be easier to collect peak-hour parking fees. Parking lots are all in fixed locations readily inventoried and visited by tax collectors. Many commuters already pay parking fees, and those fees could simply be raised. Persons who now park in free spaces provided by employers could be charged through those employers. Moreover, high parking fees would not slow the movements of traffic. In contrast, charging peak-hour fees to cars moving on congested roads poses formidable technical problems. Collecting parking fees on the millions of parking spaces now provided free by employers would also create a sizable administrative task. This might generate a whole new bureaucracy that further interfered with the activities of many organizations, especially small firms.

Second, parking fees would be levied against only some of the vehicles using toll roads during peak hours, whereas road pricing would charge all such vehicles. Long-distance trucks or other vehicles making trips through a region during peak hours would not pay parking fees; so they would not be deterred from driving then. Nor would such fees hinder drivers running errands that did not require long parking. Also, if many commuters shifted to ride sharing or public transit to avoid high parking fees, then nonparking vehicle users who had been avoiding travel during peak hours might shift into these periods. That would offset some of the decline in congestion caused by high peak-hours parking fees.

Third, higher parking fees would penalize all peak-hour auto commuters, whereas road pricing would penalize only those who used those roads on which peak-hour tolls were charged. So parking charges would in theory be more effective at discouraging peak-hour commuting—especially if they were levied only against solo drivers.

The Bay Area Economic Forum has proposed linking the elimination of free employee parking to a travel allowance for all workers [4]. Workers could use it to pay for parking if they wished to continue driving alone, to pay some or all public transit fares, or to pay part and save part if they switched from driving alone to ride sharing. For example, if an employee paid each worker a travel allowance of $75 a month and charged $75 a month for formerly free parking, then workers who wanted to continue driving alone would be no worse off. But they could gain economically by using transit at a fare cost of less than $75 a month or by splitting the parking fee with others through ride sharing. Each employer would be no worse off if it continued to fill its formerly free parking spaces or rented them to nonemployees. However, employers would then be paying workers using the public transit system a travel allowance they do not pay now. So this scheme would be most attractive to employers whose workers now nearly all drive to work and enjoy free parking.

The impact of this policy on solo driving is accentuated by present income tax laws. They treat any employee travel allowance as taxable income, but do not allow travel costs—either parking fees or transit fares—to be deducted from personal income as business costs.

A milder form of reducing free parking for solo-driving commuters would be to end the income tax deductibility of all employer expenses connected with providing such parking. That would include capital and operating expenses for building and maintaining parking spaces used by solo commuters. Also, public transit subsidies for employees, or ride sharing and van pool expenses, could be made tax deductible. These changes would encourage business firms to reduce free parking for solo drivers and to increase their support of other means of employee commuting.

[[2, pp. 70-72]; also [3, pp. 192-195]]

[T]he biggest impacts of changing densities on average commuting distances are caused by moving from very low to medium densities, rather than from medium to very high densities. Thus, moving from 1,000 to 5,000 persons per square mile would cut average commuting distances much more than moving from 5,000 to 10,000, even though the absolute increase in density is much greater in the second case. The reason lies in the basic way circular metropolitan areas grow. If 1 million persons resided in a perfectly circular metropolitan area, its radius would be 17.84 miles at a density of 1,000 persons per square mile, 7.98 miles at a density of 5,000 persons, and 5.64 miles at a density of 10,000 persons. The overall radial difference between the lowest-density case [and] the highest density one is 12.20 miles. But 81 percent of that difference lies in going from a density of 1,000 to a density of 5,000; only 19 percent lies in going from 5,000 to 10,000. Thus, to conserve energy and shorten total travel, it is more important to avoid having new growth occur at very low densities than to have such growth occur at very high densities.

[[3, p. 207]]; also [2, p. 85]

The above is a very interesting bit of analysis. It's perhaps not that relevant a stat; the average for the Greater Vancouver Regional District is 22.2 persons per hectare (5,750 persons per square mile). At the very low end of the Vancouver scale, places like Maple Ridge/Pitt Meadows have 7.0 persons/ha (1,800 persons/square mile), so I suppose their stats apply there. Vancouver proper has a density of 50.9 persons/ha (13,188 persons/square mile), so we're off his radar. (All stats from GVRD, and I should note that they represent land area with water bodies excluded. Vancouver is definitely not a circular region!)

At any rate, Downs' analysis is sound, I'd say. He's basically just describing the "low-hanging fruit"; there are still very good reasons to improve density beyond the 5,000 mark. Energy use can drop very dramatically once densities are sufficient to allow walking and cycling to most destinations, and transit viability also depends on densities above 5,000. Downs is focusing on automobile-based travel alone.

"The most important policies necessary to control rapidly rising transit costs are not in the area of land use, but rather in the area of labor relations" by keeping labor costs down. This is true because 70-85 percent of the operating and maintenance costs of public transportation are labor costs.

[[3, pp. 210-211]]

I'm very wary of the quote above (and who's he quoting, anyway?). I haven't looked at extensive statistics regarding the costs in a transit operation, but this feels like another agenda creeping into the analysis. Labour costs are an issues, without doubt; in Vancouver, TransLink had a big fight with its unions when it wanted to introduce shuttle buses for low-density areas that can't provide the ridership to fill a regular bus. The viability of such shuttle buses is more tied to labour costs that traditional buses. However, the same issue can be solved by tackling land uses. Moreover, transit is already cost-competitive with automobiles. If the costs of automobile use were more transparent and less subsidized, the operating costs of transit systems would be less contentious an issue, since they would still be lower than automobile-based transport.

Most of the links just discussed involve the effects of varied residential densities on public transit usage. But causality sometimes flows in the opposite direction. When many commuters into a business district use public transit, developers may be motivated to increase the district's nonresidential density. For example, when a new public transit system is built serving a downtown, more people can commute there without causing any greater peak-hour roadway congestion—hence without raising average commuting times. Yet commuting times influence workers' choices of where to live and work. If more workers can reach downtown in the same commuting time as before, more will want to work there. The suggestion that this may justify developers' building more office space or other facilities there is borne out by huge expansion of office space in the downtowns of both San Francisco and Washington after the construction of mass transit systems serving them.

Consequently, one way to strengthen the market for office and other space within a business center is to build more off-road transit facilities to serve it. That is undoubtedly why downtown business interests so strongly support construction of new fixed-rail transit systems, especially if they can obtain federal subsidies to cover much of the costs. Portland, Oregon, is a striking example of this relationship. Its city officials have deliberately promoted new light rail systems in part to strengthen the region's downtown. This relationship also holds for residential densities, especially in outlying locations. Thus the better the public transit service to a residential neighborhood, the higher the density of housing that can be supported there, other things being equal. That is why high-density housing clusters have appeared around mass transit stops in Toronto and Arlington County, Virginia. Persons living there can commute by transit during peak hours without encountering highway congestion, and more people are encouraged to live near such transit stops than would live in those areas if the stops did not exist. However, for this relationship to bear fruit, local residents must permit previously low-density development near transit stops to be converted to higher-density development.

[[3, pp. 212-213]]

Do residents of low-density areas block choices of higher densities by others?

Nearly all suburban communities have zoning ordinances that control the densities at which new homes can be built or existing ones redeveloped. Typically, these ordinances severely restrict the amount of land on which relatively high-density housing can be developed. That includes both multifamily housing and single-family housing on small lots. Analysis of this practice by many urban economists and by the Advisory Commission on Regulatory Barriers to Affordable Housing has clearly shown that suburban zoning often prevents the creation of higher-density—and therefore relatively low-cost—housing. Many suburban governments pass zoning ordinances deliberately designed to prevent lower-cost housing within their communities. Their residents fear lower-cost housing located nearby would reduce the market values of their own homes. Also, they do not want to live near households of lower socioeconomic status. So they adopt laws that raise the cost of building new units, for example, by requiring relatively low-density housing.

Many residents of such exclusionary communities benefit from restrictions preventing the construction of lower-cost housing and entry of lower-income households because such rulings drive up the market prices of their own homes. In this way, they also attain the kind of local socioeconomic mixtures they prefer. But these policies impose costs on the low- and moderate-income households excluded from such communities. Since the beneficiary households generally have much higher incomes than the penalized households, this amounts to a regressive redistribution of welfare. In my opinion, it is therefore socially undesirable.

Insofar as low-density settlement is accompanied by the widespread use of such restrictive zoning ordinances, it reduces society's efficiency and welfare by inhibiting the choices of households that would prefer higher-density housing units. Such units could often be relatively inexpensive because units of moderately high density are less costly to build than those of very high density or very low density. Moreover, many low-wage workers employed in communities where housing is too expensive for them must commute from other communities, often driving long distances. Thus, restrictive zoning contributes to longer average commuting journeys than would otherwise occur.

[[3, pp. 221-222]; also [2, pp. 93-94]]

Some transportation economists theorize that people everywhere have a rough travel time "budget" of about 1.0 to 1.5 hours per day, whether they are poor pedestrians or wealthy car owners. Andreas Schafer and D. Victor surveyed residents of two African villages, thirty-six cities in Europe, Asia, and Latin America, and twenty developed nations to discover how much time those residents spent, on the average, in travel each day. The average results from these groups ranged from slightly under 1.0 hour to slightly over 1.5 hours. These are remarkably similar results considering the variety of income levels, travel modes, cultures, ethnic groups, and stages of economic development [10].

[[3, p. 274]]

Many Americans are concerned that rising inefficiencies caused by traffic congestion in the United States will make metropolitan areas economically less competititive with those in other nations. These worriers cite extensive use of public transit in many other world cities, inferring that mobility there is superior to mobility in our auto-dominated system.

Contrary to widespread opinion, travel by private automotive vehicles dominates overall ground passenger movement in most western European nations, though total use of public transit there is higher than in the United States. Table 16-4 shows the percentage of passenger miles traveled by specific modes in 1997 for the 7 European nations individually and 15 European nations combined.

Thus passenger cars account for 81.8 percent of all person miles traveled in 15 European nations and over 80 percent in 5 of 7 major nations there. True, all forms of public transit combined provide 15.6 percent of passenger ground travel in Europe versus less than 2 percent in the United States. One result is that commuting times in these nations are longer, on the average, than in the United States. Across 15 European nations, the average one-way commuting time in 1996 was 38 minutes versus 22 in the United States in 1995. Denmark, which had the highest percentage of total public transit travel, also had the longest average commuting time of 43 minutes; Italy had the shortest time of 23 minutes. [1, Benchmarking, Section III. Inputs and Outputs].

[[3, pp. 286-287]]

I really take issue with the above quote. His table denominates travel as a "share of annual passenger kilometers traveled, by mode, Europe." This choice of measurement biased in favour of long trips rather than short, and completely excludes all trips by foot or bicycle. Furthermore, he includes Denmark, Germany and the Netherlands in the table, all of which have high levels of bicycle trips. But by excluding bicycles, he makes it look like 79% of Danes travel by car, and 82% of Germans and Dutch. On that basis, he draws his conclusion that Europe is little different from the U.S.—clearly false. His source does include non-motorized modes, but has not combined statistics for all modes due to data limitations. His source also goes into much more detail about trip lengths, etc.

According to French transportation expert Christian Gerondeau, the average automobile commute in most of Europe is about 19 minutes each way. (In the Paris area, it is about 25 minutes.) In contrast, the average transit commute is 49 minutes each way. That is why about 70 percent of European commuters use private cars versus about 15 percent by public transit. He believes only people who do not have a choice of using cars are riding on transit. [5]

[[3, p. 289]]

I'd like to see the source of these statistics.

The number of registered automotive vehicles [in Beijing] rose from 600,000 in 1990 to 1.6 million in 2000, a gain of 100,000 per year. Although a lot of road building has occurred, it has not kept pace with this huge rise in vehicle ownership and use. Hence traffic jams are a daily occurrence. In 1998, Beijing banned bicycles on a 300-meter stretch of Xisidong Street from 7:00 a.m. to 8:00 p.m. because they were clogging this major street. However, there are still several million bicycles in use every day in Beijing.

[[3, p. 290]]

Wow, that's a nightmare discussion. "Clogging" the street, indeed. His sources for this paragraph are two very rational positivist media articles, talking all about "modern" this, "streamlined" that and "three-dimensional" expressways.

The appendices of the books are also excellent, and highly recommended.


Commission for Integrated Transport.
European best practice in delivering integrated transport.
Technical report, Commission for Integrated Transport, London, UK, November 2001. http://www.cfit.gov.uk/docs/2001/ebp/index.htm

Anthony Downs.
Stuck in Traffic: Coping with Peak-Hour Traffic Congestion.
Brookings Institution Press, Washington, D.C., USA, 1992.

Anthony Downs.
Still Stuck in Traffic: Coping with Peak-Hour Traffic Congestion.
Brookings Institution Press, Washington, D.C., USA, 2004.

Bay Area Economic Forum.
Market-based solutions to the transportation crisis: Executive summary.
Technical report, Bay Area Economic Forum, San Francisco, CA, USA, 1990.

Christian Gerondeau.
Moving peopl and goods in europe.
In Driving America Conference, Washington, D.C., USA, 1999.

Phil Goodwin.
Solving congestion.
Inaugural Lecture for the Professorship of Transport Policy, University College London, 1997. http://www.cts.ucl.ac.uk/tsu/pbginau.htm

Jonathan Levine and Yaakov Garb.
Evaluating the promise and hazards of congestion pricing proposals; an access centered approach.
Technical Report 2/11, Floersheimer Institute for Policy Studies, Jerusalem, Israel, 2000. http://vtpi.org/levgarb.pdf

M.J.H. Mogridge.
Travel in towns: jam yesterday, jam today and jam tomorrow?
Macmillan Press, London, UK, 1990.

Terry Moore and Paul Thorsnes.
The transportation/land use connection.
Technical Report 448/449, American Planning Association, Chicago, IL, USA, January 1994.

Andreas Schafer, D. Victor, and Robert W. Johnson Jr.
The future mobility of the world population.
Transportation Research A, 34:171-205, 2000.

David Pritchard 2007-12-10