Pur Autre Vie

I'm not wrong, I'm just an asshole

Thursday, February 14, 2019

Amazon and Local Taxes

I don't have strong feelings about Amazon's decision to pull out of its plans to build a large office complex in Long Island City. It probably would have been good for the city, but the city does not really need it and perhaps it will be built somewhere that needs it more. [Edited to add: My cursory review of the Amazon proposal seems to indicate that the city would have been a substantial net winner from a financial perspective, which probably explains why NYCHA tenant leaders supported it. Of course as I observe below, there would have been losers as well as winners, but unless I'm missing something the claims about NYC "saving" $3 billion as a result of the project's cancellation seem to be highly misleading, comparable to the claims about how Britain could "save" hundreds of millions of pounds a week by withdrawing from the EU.]

I would like to make an observation about the politics of location decisions, though. It is easy to get angry about "big corporate giveaways," but they are often rational from the perspective of the community offering them, and getting rid of them would probably require complicated and intrusive laws.

One source of confusion is lack of clarity as to the net and gross benefits being conferred upon the corporation and the community in which it proposes to locate. Imagine a corporation that plans to build a facility that will generate $10,000,000 per years in taxes, while costing the town $1 million in additional government services. Imagine that the corporation is indifferent among three different locations for its facility. One town offers a tax rebate of $5 million per year, another offers $7 million, and the third refuses to offer anything on the grounds that corporate giveaways are bad. The corporation accepts the highest offer and pays a net $3 million in taxes per year, while costing the town $1 million in services. The town comes out ahead by $2 million. Nevertheless the headline may be something like, "Mayor Approves $8 Million Corporate Giveaway" (counting both the tax rebate and the town's expenditure on services), which may draw the ire of voters due to confusion between the gross amount of the rebate and the net amount of taxes paid by the corporation. Still, though, you can't say that the mayor made a bad decision for the city, and you wouldn't advise a civic-minded mayor to turn down such an offer (unless, I suppose, other corporations were making better offers and the city couldn't accommodate all of them).

There are three or four objections that I think you could mount against this sort of rebate. The first is simply that local officials may be bad at calculating the net effect of a business opening a facility in the municipality. If so, in a competitive bidding situation there may be a "winner's curse," where the bidder that makes the biggest upside error in calculating the benefits "wins" the auction and ends up overpaying.

The second (related) objection is that local officials may be corruptible, which exacerbates their tendency to overbid.

The third objection is the most subtle. If you allow corporations to get towns to bid for their presence, then you erode the ability of towns to generate tax revenue from businesses. In other words, you might say that even the most cost-justified tax rebate, with abundant net benefits to the city, is still regrettable because the corporation shouldn't be entitled to any tax concessions at all.

The fourth objection, which I won't address further, is that I've been speaking of the city's interests as if they are unified. In fact different people will be affected differently. If you are a homeowner or you are unemployed, you may benefit a lot from the presence of a new employer (higher house prices, lower unemployment). If you are a renter and you are already employed, the increase in rents may overwhelm any benefit you get from the city's improved finances. Racists may dislike the new employer if it brings in lots of non-white employees, or may support it if its employees are mostly white. Perhaps the city's new voters are younger than the average resident, changing the city's politics for better or worse. On some level I think you have to ignore these effects or you will get bogged down, but you might worry that incumbent politicians (the ones negotiating the tax rebate) are not neutral as between their constituents, systematically helping certain residents and hurting others. (But I mean... on some level this is an ineradicable problem that isn't confined to tax rebates.)

These objections are relatively difficult to address by law. The second is perhaps the easiest in theory. You could pass a law requiring all negotiations to be conducted in a particular way, with public scrutiny and with rigorous bans on any profit by the public servants negotiating on behalf of the city or state. In practice, though, it may be very difficult to detect the payoff made by the corporation, and in general I'm skeptical that U.S. anti-corruption laws are particularly effective.

To address the first objection, you might require all tax rebates to be contingent on the actual state of the world. For instance, if the promise is that the corporation will provide 1,000 jobs, then you would require it to demonstrate that its payroll is at least that large before giving it the rebate. If the corporation ends up under-performing on its promises, then it will face no legal penalty (shit happens, you shouldn't penalize the company for unanticipated shortfalls), but on the other hand it will not get any of the negotiated rebate. (This means of course that the government can't give cash up front, which is probably a prudent limitation to impose in any case.)

The third objection can only really be addressed with a highly intrusive federal law. It is generally difficult to root out "victimless crimes," that is, crimes where both participants have motives to seek each other out (e.g. drug sales). If you ban tax rebates, cities will provide tax credits. If you ban those too, they will provide loan guarantees or project insurance or free services. Or they will simply cut their taxes (are you really going to pass a federal law freezing local taxes in place?).

One fairly radical proposal would be to impose a federal tax on corporations replacing local and state taxes. The revenues would be ceded to the states, counties, and municipalities that would receive them in the status quo ante. You would harmonize tax rates everywhere and refuse to vary them for particular projects. Corporations would then make their location decisions based on other factors, and there would be no "race to the bottom." Corporations would stop being able to internalize their positive externalities, and states and cities would get more tax revenue.

But! "Harmonize" is a bit of a euphemism, since the federal government would now be setting tax policy for every municipality in the U.S. Some cities would find their tax revenue dwindling if the federal government chooses a lower rate than they were previously charging. Other cities would lose their employers as the only basis on which they could compete was low taxes.

And anyway Congress would never do this. But without doing it, I think tax competition is inevitable, and so the best you could do would be to require it to be done in a transparent way, with special ethical rules on the government negotiators.

Wednesday, February 13, 2019

Harris and Trumpy Politics

Everything I wrote about the Liz Warren ancestry controversy can be applied in miniature to the ridiculous controversy about Kamala Harris and what music she listened to while smoking pot in college. I really do mean "in miniature"—it is obviously not remotely as potent an issue, it just has similar attributes.

In short, during an interview Harris said she smoked weed in college and seemed to indicate that while she was smoking weed she listened to Snoop Dogg and Tupac. The problem is that her college years preceded their first albums.

As the Times piece I linked to explains, it's not clear that Harris made the claim being attributed to her. She was asked what kind of music she likes in general, and before she could answer another interviewer interposed another question, specifically asking, "What were you listening to when you was high?" The first interviewer then suggested Snoop, which Harris confirmed, adding Tupac as well. She could easily could have been ignoring the part about being high, or she could have smoked pot later in life while listening to Snoop and Tupac.

But look. No one is going to listen to this defense. I mean it's worth trying, if you want to lay out the facts. But Harris herself is going to look (mildly) ridiculous if she tries to argue the point.

And there's another important feature here, which I forgot to mention in my post about Warren. This is eminently mockable on Twitter, and once people get into the "dunking" and riffing mode, anyone pointing out the precise details of the interview is going to look absurdly uptight.

In fact this thing people do on Twitter, where they feed on each other's jokes and show off their feeble wit, is one of the most annoying things about the site. I can't count the number of times I've seen waves of jokes premised on a tendentious or outright false understanding of the news. And I think the fact that it's all couched in a sort of cynical, ironic, humorous pose makes it much more "persuasive," in a sense, than earnest tweeting.

Anyway the other similarity with Warren is that it's the kind of issue that no rational person would give a shit about (even the existence of this blog post is quasi-ridiculous), but I think it amounts to a kind of threshold issue for some people—is Harris phony? I think it's the kind of thing that can make someone laugh and then distrust Harris forever.

Again I really don't want to make this out to be a big deal. It's not. I just hate this aspect of politics and I hate how well this stuff works for Trump.

Tuesday, February 12, 2019

Getting Angry for No Reason

I refuse to read the whole article, but this piece in the New York Times on capturing carbon dioxide from the air seems dumb as hell. Here's a passage:

The technicians had in front of them 12 large devices, stacked in two rows of six, that resembled oversize front-loading clothes dryers. These were “direct air capture” machines, which soon would begin collecting carbon dioxide from air drawn in through their central ducts. Once trapped, the CO₂ would then be siphoned into large tanks and trucked to a local Coca-Cola bottler, where it would become the fizz in a soft drink.

This is just so dumb. It reminds me of the way people used to describe electric cars as "carbon free" because they don't literally release carbon dioxide into the air. Of course if those electric cars are recharged with electricity from coal power plants, they are hardly "carbon free." People were taking the term far too literally.

Similarly here. If you remove a bunch of carbon dioxide from the atmosphere and then use it to carbonate Coke, guess where that carbon ends up? The question is not whether the technology is capable of literally removing the carbon dioxide from the air, the question is whether that's a more useful (i.e. less carbon-intensive) way to produce carbon dioxide than existing methods.

This is especially dumb because we already produce a lot of carbon dioxide as a byproduct of industrial activity. For instance, fermenting beer creates a lot of carbon dioxide, and a handful of breweries capture it and compress it for use carbonating their beer. This carbon dioxide is "free" in the sense that almost all of the energy used to produce it was going to be used anyway to make the beer, so the incremental environmental cost of capturing the carbon dioxide is simply whatever machinery has to be installed to capture and pressurize the carbon dioxide from the fermentation.

The same is true of all kinds of commercial and industrial activity. Now conceivably the technology profiled in the New York Times article is superior in some way. But if so it's not because it removes carbon from the air (only to put it back into the air when people open their cans of Coke), it's because it does so more efficiently than other methods of producing carbon dioxide for commercial use.

Thursday, February 07, 2019

Fire from the Left

A friend reminds me of an important addendum to my previous post. I think it will matter quite a lot whether the leftists decide that they prefer Trump or the Democratic nominee.

I am exaggerating slightly, but I think it's fair to say that a lot of prominent leftists were ambivalent at best about the 2016 election. Many of them spent the election season vigorously attacking Clinton and making sure that everyone was aware that the Democratic Party in general and Clinton in particular are elitist, game-rigging corporate shills who despise working-class people. Clinton delivered her nomination acceptance speech to loud boos and jeers from Sanders delegates. I think it's hard to overstate how much better off Obama was as a result of a united Democratic Party.

I am not presently arguing for the nomination of a leftist candidate (I'm not sure who that would even be—Sanders, I suppose, though I suspect that a lot of loyalty to him would evaporate the moment he started saying nice things about the Democratic Party). I'm just observing that this fissure has been a real problem in the past, and the Glenn Greenwalds of the world will almost certainly do whatever they can to get Trump reelected. It's a dynamic that will be tricky to navigate.

Trump's Taunts

I was having a conversation with some friends and the subject of Trump's attacks on Elizabeth Warren came up. As it happens I think those attacks tell us a lot about Trump, whose fondness for tagging opponents with unflattering nicknames is a signature move (to the point that people actually mocked him when his nickname for Nancy Pelosi turned out to be... "Nancy").

One of the earliest examples was "low energy" Jeb Bush. In all likelihood Bush was doomed to fail no matter what, but "low energy" seems to have been very effective. I think this was for a few related reasons.

First, it is basically true. Bush's campaign was boring and his manner on the debate stage was bumbling and ineffectual. (Bush fans can argue, rightly I think, that he actually has a good sense of humor and a fair amount of sharpness. The same is apparently true about Gore. In neither case did it matter much.) Bush was also running a campaign straight out of the GOP's "post mortem" on the 2012 election—upbeat, welcoming to immigrants, kumbayah. Trump sensed a much darker mood in the Republican Party's base and found a way to establish a sharp contrast.

Second, "low energy" is vague and impossible to refute. What does it mean? Nothing exactly. It connotes a lot without referring to any specific thing.

Third, relatedly, to confront it directly is to invite more mockery. "Don't call me low energy!" What a low energy response that would have been. A friend asserted, convincingly, that if Bush had walked across the debate stage and punched Trump in the face, he may have salvaged his candidacy (or at least ended Trump's). The likeliest outcome—Trump running away—would have drawn scorn from precisely the right quarters. If Trump didn't run away, Bush actually landed a solid blow (this part is key), and Trump didn't hit back effectively (also key), it would be almost as good. (The move would have drawn condemnation of Bush, but mostly from people irrelevant to the Republican primaries.)

Anything verbal response, though, would simply demean Bush, as Marco "Trump has a small penis" Rubio learned. Trying to shift between dignity and Trump-esque vulgarity is a recipe for disaster.

Fourth, again relatedly, there is a strong psycho-sexual element to the taunt. It's not just that Bush isn't a good public speaker. It's that he's unmanly. (Again, to repeat my second point, this lack of virility is hinted at but not stated outright.) Trump's specialty is serving up provocative sexual and racial innuendo without quite being specific enough for anyone to pin it to him. (He famously said that an Apprentice contestant would "look good on her knees" or something like that... on her knees because she was begging to be allowed back on the show, of course! How dare you suggest any other meaning?)

You see the same dynamic with Trump's "spill the beans" tweet about Heidi Cruz, except that there's no sign of truth in that case.

I want to emphasize that this only works with a particular audience, and it alienates a different (probably larger) audience of reasonable people. But thanks to luck or cunning, the tactic was well-suited to bring a lot of Republican primary voters and less-educated white general-election voters into Trump's camp while alienating a lot of Democratic voters who were never going to vote for him anyway.

Now consider Elizabeth Warren. Trump seems to have gotten the idea of calling her "Pocahantas" in 2014 (note that Trump's style of retweeting is simply to copy-paste the text into his own tweet):
He returned to the term in 2016 (he may have been using it before then, but there's no sign of that on his Twitter account):

This attack has it all—it has an element of truth (more on that in a minute), it's vague, it's almost impossible to refute without losing dignity (as we've seen), and it has a strong element of racial antagonism.

Now the dilemma here is simple. Warren's behavior was problematic from the viewpoints of at least two constituencies (members of Native American tribes and whites who are resentful of affirmative action), both involving strong feelings related to racial identity. In other words, the attack is aimed at a true political vulnerability (unlike, say, "Cryin' Chuck Schumer" or "Sneaky Dianne Feinstein," Trump coinages that don't touch on anything of political significance). So in other words, it was a potent attack even if she maintained a dignified silence about it.

Moreover the attack benefited from strategic ambiguity. Was Trump mocking her for having Native Aemrican blood or for not having it? Was he claiming that she lied about her ancestry or that she used her (genuine) ancestry for illegitimate purposes?

I'll pause here to note that once again different audiences would see these attacks in different ways. Plenty of people were horrified at this attack, either because they assumed the worst (mocking her for having Native American blood) or they thought it was a disgusting tactic in any case. It's a classic example of Trump maximizing his appeal to racially resentful white voters, a tactic that worked brilliantly in 2016 but has unclear prospects for 2020.

In any case, Warren apparently decided to address her response to the most plausible interpretation of Trump's remarks, counting on a sort of "gotcha!" dynamic. And in fact, before Warren's response, if you had forced Trump supporters to write down the substance of Trump's charge and the circumstances that would vindicate Warren, a lot of them would probably have written something like: "She said she had Cherokee blood but she doesn't, and only DNA proof to the contrary would vindicate her."

But, and I can't emphasize this enough, no one was forced to write anything down. Trump and his followers ruthlessly exploited the strategic ambiguity that he had always maintained, with devastating effect. Warren's "gotcha" blow didn't get even close to landing, and she took fire from pretty much everyone and was forced to retreat in humiliating fashion. (As a side note, it's a classic issue where getting into the weeds is both (A) necessary if you want to clear things up and (B) devastating from a political messaging perspective. It's very much like Clinton's emails in this regard. Just an absolute political meat grinder.)

I have no idea how to deal with this on the level of political tactics (e.g. in thinking about our Democratic primary choices). I'm inclined to say that Clinton's emails were a bigger problem because Trump got institutional support from the FBI and the mainstream media that likely wouldn't be forthcoming from anyone this time around (seemingly Warren has made peace with the Cherokee tribe, and the mainstream media hates the nickname). But I don't know! I also don't know how effective it is to think about particular weaknesses of the Democratic hopefuls when surely Trump will latch onto something with the help of his media enablers. (To be clear, though, "Cryin' Chuck Schumer" shows how weak Trump's attacks can be when he doesn't have any obvious angles of attack.)

I'll conclude by observing that Trump has made no real innovations in policy, and it's a massive category error to speak of, say, a "Trumpist" foreign policy except in the the most basic of senses. But he did introduce true tactical innovations to Republican politics, and I believe this was an under-appreciated source of his popularity with Republican voters. They wanted a leader who fights dirty, and they got one. It shouldn't have worked. Voters shouldn't have been vulnerable. But they were, and here we are.

Tuesday, January 29, 2019

Bailouts

Just a quick observation that the word "bailout" has become one of those meaningless words that simply connotes a negative view of a situation. Saying "I oppose bailouts" is pretty much tautological. When someone I don't like benefits from government policy, it's a bailout. When someone I like benefits from government policy, it's the government serving its proper role.

I suppose this was inevitable because of the way English works, but I am going to find bailout rhetoric annoying for the rest of my life and I wish it didn't have to be this way.

Friday, January 18, 2019

The Last Vestige of Racism

Over the course of my lifetime it has become far less acceptable to be racist in public in the United States. And even when I was young, overt racism was already looked down on for the most part—there was plenty of it, and it often passed without objection, but it was far less acceptable than it had been 20 or 30 or 40 years before. Of course I realize Trump has unleashed a new wave of racism, but that's confined mostly to the conservative movement. Relatively few liberals say racist things in public, and when they do they're generally forced to apologize quickly.

I have to say, though, that there is one glaring exception, and it's distressingly common on both ends of the political spectrum. This is racism against Puerto Ricans. I can't count the times I've seen people refer disparagingly to "Puerto Rican people" (usually abbreviated as "PR people"). "Did some P[uerto ]R[ican] person draft this?" "You sound like a P[uerto ]R[ican] person right now." "Oh great another P[uerto ]R[ican]-drafted statement instead of full disclosure and an honest apology."

As I understand it, the "joke" is that Puerto Ricans often use English in a somewhat stilted or indirect way, deflecting blame and issuing non-apology apologies. It's a weirdly specific racist thing to say or to believe, and I hope it goes the way of all other flagrantly racist tropes.

Wednesday, January 16, 2019

On the Horns

A quick followup to my previous post. What I hope I made clear is that U.S. cities are caught on the horns of a dilemma. If it were (politically) easy to build more housing, they could easily solve their affordability problems (at least in the short term). If it were easy to build more transit, they could easily solve their affordability problems (again, at least in the short run). You only need one of those things to be true in order to address housing affordability. The problem is that in the U.S. it tends to be difficult to do either of those things, and so we are left with very expensive housing in our most productive cities, with all that follows.

Well-Located Housing and the Tokyo Option

This post is meant to make a simple observation about cities, which I hope to expand later.

The high cost of housing in many cities is a function of supply and demand. But supply and demand of what? I think the answer is basically "housing from which it takes about x minutes to get to high-paying jobs." The point of formulating it this way is to capture the intuition that what we mean by "housing" isn't just raw number of units, but rather units that are usefully located.  For the rest of the post I'll use the term "well-located housing" to refer to housing from which places of employment can be reached relatively quickly.

There isn't a sharp threshold here—housing that is within 15 minutes of jobs is obviously better than housing that is 30 minutes from jobs, but in most big cities either of those would qualify as "well-located housing." However, you might want to attribute more value to the 15-minute housing than to the 30-minute housing, and there are various ways to do this. For instance, to determine how much "well-located housing" a particular development will add, you could divide the square footage by the average commute time to the city's employment centers. Or you could divide the square footage by the commute time to the nearest employment center, on the logic that people will generally choose to live near work. Or you could sort housing into tiers (1-10 minutes from work, 10-20 minutes from work, and so on), and attribute more value to the low-commute-time tiers. Or maybe square footage isn't the right measure and you should care more about number of bedrooms, or something. I think my argument would work for a wide variety of approaches, so I won't worry too much about the precise definition of "well-located housing."

The point I want to make is that there are several different ways to increase the supply of well-located housing. One way is simply to permit the construction of more housing in areas that are already well-located. This is the focus of the YIMBY (yes in my backyard) movement, which favors relaxing legal limits on housing construction in urban and suburban areas.

But it would also work (in a sense) to take existing housing and make it better-located. Imagine a city that is built along a river and that has a central business district adjacent to the river where most good jobs are located. Also imagine that the nearest bridge across the river is 5 miles away and there is no ferry. In the status quo, the housing immediately across the river from the city is not well-located (unless there is some other employment center nearby, and let's assume there's not), so its existence does not meaningfully contribute to the supply of housing for the city. Building a bridge across the river near the central business district will dramatically lower the commute time for residents of the housing across the river. In our framework this amounts to an increase in the supply of well-located housing and, other things being equal, a drop in the price of well-located housing.

Of course there are complications. First, the "new" well-located housing (maybe the better term is "newly well-located housing") may be located in a different political jurisdiction from the city where the central business district is located, with important consequences that are beyond the scope of this blog post.

Second, there will be winners and losers from the construction of the bridge. As with any increase in supply, the incumbent owners of well-located housing will suffer as prices fall (and their tenants will benefit). Meanwhile the owners of housing across the river will enjoy a windfall, while their tenants will pay higher rents but enjoy shorter commutes, with mixed effects on welfare (for instance, retirees will pay higher rents but might not care much about short commute times). On net the metro area should benefit (depending, of course, on how expensive it is to construct the bridge), but the distributional effect may be large and for many people may completely swamp the average effect. (To put it another way, there may be a lot of losers from the policy even if it is quite beneficial on the whole.)

Third, if the bridge dumps a lot of new drivers into an already-congested downtown, it may lengthen commute times for the rest of the city, so the net effect on well-located housing may be ambiguous.

You can substitute almost any transportation infrastructure for the bridge and the logic will stay the same, with some differences depending on the particular form of transportation chosen. For instance, in a large city grade-separated rapid transit tends to make nearby housing much better-located. I bring this up because YIMBYs are often surprised at the low density of the Tokyo metropolitan area. Tokyo is, famously, the most populated metro area in the world, with tens of millions of people "well-located" relative to its job centers. But Tokyo accomplished this not by building (very) densely, but rather by building an excellent train system that created a sprawling megalopolis of well-located housing, much of it single-family and relatively little of it taller than three or four stories. (It is not an entirely fair comparison, but you would struggle to find a single ward of Tokyo that is denser than the neighborhood of Park Slope, which itself is not particularly dense by the standards of New York City.)

Unfortunately the "Tokyo option" is not really available to very many cities in the U.S., if any. This is because (A) construction costs are ludicrously high for new trains in the U.S. (and, relatedly, our transit systems are notoriously poorly run), (B) our political institutions are not designed to encourage mass transit (in particular, our transit-oriented cities tend to be boxed in by affluent suburbs that make Tokyo-style sprawl difficult or impossible to achieve), and (C) we are to some degree "locked in" to our existing housing stock, which is generally not transit-oriented. However, I think YIMBYs who are motivated by a desire to reduce housing costs should consider transit construction/improvement as a complement to/substitute for upzoning. Where density is already high (as in much of New York City), the lowest-hanging fruit probably consists of transit improvements. Transit improvements offer the possibility of increasing the stock of well-located housing while avoiding a political backlash from NIMBYs who, as I've argued previously, tend to be influential in urban politics.

(As an aside, I have read conflicting accounts of land use regulation in Tokyo. Some YIMBYs seem to believe that zoning is quite strict, while other YIMBYs seem to believe that Tokyo has few if any zoning restrictions. In a future post I will try to resolve this question.)

I want to bring up one more option to increase the supply of well-located housing, which is to build lots of little employment centers rather than a few big ones. This is the approach favored in the sprawling car-based cities that have grown in the southern and western United States, and I think urbanists tend to underrate its effectiveness. It is true that we observe agglomeration effects in lots of industries, where a firm's productivity is a function of proximity to other firms in the same (or related) industries. This would tend to favor development of a few large business districts, which in turn favors (above a certain scale) the use of mass transit and dense residential construction to achieve a lot of well-located housing.

But agglomeration is often industry-specific. The movie studios in Los Angeles don't particularly need to be near its financial industry, or its aerospace industry, or its beer breweries, or whatever. By developing lots of small job centers, each of which is easily reachable by car from the nearby housing stock, a city can do a pretty good job of capturing agglomeration effects while delivering well-located housing.

Of course there are complications. A car-based city is one that will be hard to navigate for anyone who can't drive (for financial or other reasons). However, I don't want to exaggerate this point—I imagine that once you take into account the cost of housing, most sprawled-out sunbelt cities are more affordable than New York City even accounting for the cost of a car.

Another complication is that once you have specialized job centers, "well-located" becomes specific to the industry in which the individual works. A brewer may not need to live near a movie studio for work purposes, but if she is married to an actor she may want to anyway. If the brewery cluster and the movie studios happen to be across town from each other, it creates a problem. Also, a worker who switches industries may also need to switch houses, which involves significant moving costs. This would be rarer in cities where jobs are all clustered together, because in that case switching jobs should not affect your commute much.

Finally, the car-based model probably doesn't scale very well. Beyond a certain point, traffic congestion starts to degrade the "well-locatedness" of most/all of the housing in the metro area. Car infrastructure is noxious (no one wants to live too close to the highway, and parking lots use up expensive, well-located land and are unpleasant to walk past). Moreover, car-based development is (to a certain degree) incompatible with evolving toward a transit-oriented, pedestrian-friendly city structure. In other words, up to a certain city size car-facilitated sprawl is probably the best policy, but as the city continues to grow it finds itself stuck in a pattern of development that no longer makes sense. Unfortunately it is hugely expensive and politically difficult to reorient the city toward a development pattern that is suitable for a large population. For this reason I think it's worth considering transit-oriented development at a relatively early point in a city's lifetime, although I admit this is a very tricky issue.