Why Isn’t Silicon Valley in San Francisco?

Posted by on . 11 comments.

Randall Parker leaves a comment:

If density is so important then why is far more high tech in Silicon Valley as compared to San Francisco? The ideal level of density needed for high tech is probably below the ideal density for commuter rail.

And gives me the opportunity to correct a lot of potential misconceptions about what I’m arguing. First, it’s worth pointing out that San Jose is pretty dense. Its weighted-average density — the density at which the average person lives — isn’t as high as San Francisco’s, but it’s on a par with that in Chicago.

But this is somewhat beside the point. The locus of high-technology work is in San Jose rather than San Francisco precisely because density is so important. Because proximity to the dense cluster of tech firms in the Valley is so advantageous, it’s very costly for any individual firm to leave the cluster. Perhaps the firms in the cluster would all be more productive if they were grouped together more densely, but there’s no way to move them to San Francisco except by a massive coordinated effort; the advantages of density hold firms around San Jose. And of course, the fact that the cluster began there in the first place is just an accident of history — a product of the location of Stanford University and the military facilities that supported much of the early technology research in the area.

The argument is not that economic activity will seek out the densest place in the country to do their work. It’s that, other things equal, an area will be more productive as its density rises. And presumably there is a role for skill complementarity; if we increase the density of San Jose by adding engineers to the local population, productivity will likely change in a different way than if we add plumbers or accountants, or than if we added engineers to Manhattan. Density isn’t magic; it’s simply a measure of the market, labor, and human capital to which local firms have good access.

Parker seems to be arguing that Silicon Valley currently exists at the ideal level for high tech industrial work. I doubt this is true, though there’s no way for any of us to know. High tech centers around the world exist at very different densities, and the density of Silicon Valley has itself varied over time as the area has grown. I wouldn’t begin to propose an effort to try and determine ideal densities and push areas toward those levels. Rather, I’m observing that in general there is a relationship between density and productivity, and this means that artificial limits on density (which the Bay Area has in spades) are very costly. The first, best way to address this would be to remove existing barriers to development. This is often a challenge. An alternative that has been successful in many cities is to develop local transit networks and adjust land use around stations to accommodate higher densities.

San Jose is already dense enough to support some rail travel. But we can look at the heart of Silicon Valley and observe a few things. First, the cost of housing indicates that there is excess demand for homes in the area. Second, we see that many scientists and engineers commute into Silicon Valley from places all around the Bay. It seems likely that a substantial share of long-distances commuters into the Valley are forced to live farther away from their place of work than they’d prefer thanks to high housing costs. The use of local transit to facilitate denser development would increase local housing supply, slow housing cost growth, allow more people to live in the region, and allow a larger share of the regional population to locate at its heart.

Comments

  1. John says:
    February 5, 2011 at 5:06 pm

    It’s also true that the value of density is related to some purely political considerations as well. San Francisco is highly zoned and for that reason is probably overpriced. I live in New York and I’m amazed at how even in lower Manhattan, the market wants to build a lot more than they are allowed. Zoning restrictions, even in already dense areas, make relocation artificially overpriced.

    It’s also the case that Silicon Valley would likely become much denser without zoning, belying the “ideal density” argument. I’m sure if that happened tech firms wouldn’t suddenly flee the area.

    Also, of course, in New York, “Silicon Alley” is in southern Midtown, which is considerably denser than San Francisco. Those companies don’t just move out into the woods or New Jersey.

  2. Jim Russell says:
    February 5, 2011 at 5:20 pm

    The tech corridor in Boston is another similar story, with the firms on the outskirts of the city instead packed into the core.

    I see a couple of misconceptions in play. First, the issue is density of firms. Proximity to innovation (and venture capital) is not a matter of where people live. It is a matter of where people work.

    Second, you can observe a relatively high density of firms in the more spacious suburbs. There is the added bonus of less legacy costs out in the greenfields. It is density without the high real estate costs.

    Third, the density in Silicon Valley is impressive. You have all these niche clusters packed into small parts of this region. You can map (I’ve seen these maps) the specialization.

    Lastly, venture capital (like knowledge) doesn’t travel very far. The 20-minute rule is a geographic restriction on the diffusion of funding. If you want money for your startup, then you have to cram yourself in with all the other established firms.

    Silicon Valley is the greatest example of how much proximity matters:

    http://www.nytimes.com/2006/10/22/business/yourmoney/22digi.html

  3. Evan says:
    February 5, 2011 at 7:51 pm

    There are a large number of tech firms in the city, but even there, speaking to your point, it’s almost all in a couple of neighborhoods (perhaps not coincidentally, within walking distance of the 4th & King Caltrain stop, which is the train from the south bay).

    Lately, the larger firms down in the valley have been having to hire on more and more transport to keep people happy. The valley is a terrible place to live in terms of amenities, so now that basically anyone who wants to be competitive has to run its own bus line from the city to its drab valley campus, I could see the center of gravity eventually shifting up to the city. I’ve noticed that firms in Los Gatos (up in the hills above the valley to the east, most crucially very hard to get to from SF) seem to have open positions for a much longer period of time than places in the valley, who have positions open for longer than people in the valley with busses, who have positions open longer than people who’re in the city.

    It would not surprise me at all to find that salaries rise with search times, although I think that the social cachet of working for firms like Apple might distort the results a little.

  4. Evan says:
    February 5, 2011 at 7:55 pm

    I guess what I lost sight of in that long rambling comment was that to someone who lives near the valley and works in tech and knows a lot of people who do the same, that Parker is pretty profoundly wrong, in terms of some ideal density level.

    There’s a lot of empty commercial real estate in SF, and if the city government there ever got its head out of its ass and really started enabling rising densities and better housing supply, you’d find that there would be a rapid swing to SF as the local center of gravity.

    It’d also be possible for Oakland or Berkeley to do more or less the same thing, although it’d take them longer.

  5. Mixner says:
    February 5, 2011 at 8:29 pm

    It’s that, other things equal, an area will be more productive as its density rises.

    Yes, but you keep ignoring the fact that this increase in productivity is very small (according to one of your sources, a 100% increase in density yields only a 2-4% increase in productivity), that its cause may simply be that as density increases people tend to work longer hours, and that it may be consumed entirely by a higher cost of living.

  6. Matt Austern says:
    February 5, 2011 at 9:46 pm

    Minor correction, which doesn’t really affect the main point of the article: Silicon Valley isn’t really in San Jose. There are some well known companies in downtown San Jose, yes, but more are in places like Mountain View, Palo Alto, Cupertino, and Sunnyvale.

    You could say that Silicon Valley is part of suburban San Jose, but that would be misleading. There’s a solid stretch of development between San Francisco and San Jose (and on the other side of the Bay, between Oakland and San Jose), and Silicon Valley is in the southern part of it. There’s no real dividing line that would make it sensible to think of Menlo Park as a San Jose suburb and San Mateo as a San Francisco suburb. More sensible to think of the Bay Area as a single economic unit, with varying levels of density.

  7. Christopher says:
    February 5, 2011 at 10:52 pm

    Agree with Matt. But with a nod to Ryan.. Silicon Valley is really the Santa Clara Valley. And while San Jose is in that a lot more suburban locations further north from that are there too. Those areas much closer to Stanford and the research business parks they built near Menlo Park. The other cities of Silicon Valley were once filled with military research, fishing and manufacturing. The kinds of semiconductor manufacturers that gave the area it’s name. Overtime those industrial areas have given way to office parks. The Bay has been filled into push out fishing for more office parks and planned cities. All of it interestingly, though, follows the Caltrain line much closer than the East Bay’s 880. I don’t entirely why this was true other than than the 880 corridor has remained much more industrial. Like the entire East Bay. But the peninsula has San Francisco at one end and some early train suburbs like Palo Alto, Burlingame, Atherton and Menlo Park. Who’ve long been the places for wealthy suburbanites who wanted to be near horses farm that once dotted the Peninsula. And generally have more space in the ways that Chicago’s north suburbs or the North Shore of Long Island developed.

    Anyway, I remember when I first moved to SF and there was an article on how Mountain View had upscaled it’s downtown and it was described as the only blue collar city with an average 6 figure income. Many of those suburbs only 50 years ago were places to live and work in the factories or at the bases. They were working class towns.

    San Francisco has more recently made it’s place as the home of design side of silicon valley and now also biotech. It really just doesn’t have the space for the large sprawling campuses tech companies seem to want. Even here in NY metro. Silicon Alley is downtown, but more tech-focused companies are in NJ. They employ a lot of people and need more space at lower cost than even a skyscraper can offer.

  8. Ben Ross says:
    February 6, 2011 at 11:04 am

    No one has mentioned military contracting. There is surely a much smaller return to density in military research, since you’re not allowed to talk with outsiders about what you’re doing. In addition, security is easier in isolated buildings. Much high-tech started out on military contracts and then spilled over into civilian applications. There’s a lot of inertia in spatial agglomeration of industry, so an initial concentration of high-tech in the suburbs would tend to stay there even as things switched to civilian applications.

    An empirical study of this would be useful. Military contracts were certainly important in the early years of Silicon Valley. In Boston, my impression is that Raytheon and MITRE, both military contractors, were among the first to move to Route 128, while Arthur D. Little stayed in Cambridge, Stone and Webster and Chas. T. Main stayed in Boston, and Polaroid kept its research in Cambridge for a long time while building production facilities on Route 128. Whether that’s a general pattern in Boston I don’t know.

    Another point – it seems to me that the advantages of urban work accrue more to the employee (in a more pleasant life) than to the employer (in economic production). The ability to attract better employees is a bigger factor than the increase in the productivity of those employees as a consequence of the job location.

  9. Ben Ross says:
    February 6, 2011 at 11:26 am

    Okay, I looked into it a little more and it’s not just speculation. Here’s the conclusion of someone who actually researched the question: “Military funding was critical for the rise of Silicon Valley from the very late 1930s to the early 1960s.”

    http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2007/09/30/MNDTSEMSJ.DTL&ao=all

  10. MW says:
    February 7, 2011 at 11:17 am

    Tech might be drawn to the suburbs because of the cost of office space. The industry has fewer advantages to being downtown then others. For many other industries, prestige is a major factor to where they locate (On the other hand, tech is evaluated on results). All that being said, tech might locate downtown if land-use regulations were relaxed and rents were lower.

  11. Wad says:
    February 14, 2011 at 8:34 am

    Ben Ross provided the key as to why Silicon Valley developed within San Francisco’s hinterlands, rather than in the city itself.

    The Silicon Valley we know today was a decades-long collaboration between government, defense, education and a web of private support industries.

    The software companies, and later the dot-coms and venture capital, are later generations of the 1930s-1960s period.

    The Santa Clara Valley received the growth because at the time, San Francisco and the East Bay were fully urbanized and had productive industry. Land values were simply too high to allow for the early technology companies to acquire facilities. The South Bay, meanwhile, had lower-value agricultural land or wilderness, and were fairly remote from cities. Espionage had to be a big concern at that time, too.