The standard explanation for why a software engineer in San Francisco earns twice what their counterpart in Austin earns goes something like this: the city is expensive, so companies pay more to compensate. It’s intuitive. It’s also largely wrong. The causality runs in the opposite direction, and understanding why reshapes everything you think you know about how tech compensation actually works.
The real driver of elevated tech salaries in expensive cities is not the cost of living. It is the concentration of other high-skilled workers, specialized capital, and institutional knowledge that only exists in dense clusters. Companies pay more because the talent they need is there, and the talent is there because the companies are there. Housing costs are a consequence of that loop, not its cause. This matters enormously for anyone trying to understand where tech compensation is heading, especially as remote work scrambles the geography of the industry. The strategy of hiring overqualified engineers fits neatly into this logic: when you’re already paying cluster premiums, you might as well buy as much talent surplus as possible.
The Agglomeration Effect Nobody Talks About
Economists have a term for what happens when skilled workers cluster together: agglomeration. The basic idea is that proximity between talented people generates compounding returns. A senior engineer in San Francisco can have lunch with a potential co-founder, attend a talk by a researcher working on adjacent problems, and run into a recruiter from a competing firm, all in the same afternoon. That informal knowledge transfer has real economic value, and it doesn’t exist in a vacuum.
Enrico Moretti, an economist at Berkeley, has spent years quantifying this effect. His research found that when a city adds one high-tech job, it creates approximately five additional local jobs in non-tech sectors, ranging from lawyers to baristas. The multiplier effect means the entire local economy reconfigures around the presence of tech workers, which bids up the price of everything, housing most visibly.
The wages are high not because rent is high. Wages are high because the expected output of any given engineer is higher in a dense cluster than it is in isolation. Companies capture this by paying more to attract workers into those clusters, which concentrates wealth, which drives up real estate, which creates the illusion that pay is compensating for costs.
Why Remote Work Didn’t Flatten Salaries the Way Anyone Expected
When the pandemic forced mass remote work adoption, a reasonable prediction was that tech salaries would converge. If a company no longer needed you in San Francisco, why would they pay San Francisco rates to someone living in Boise? Some companies moved aggressively in this direction. Others quietly reversed course.
The reason convergence stalled is instructive. Companies that tried to impose location-adjusted pay discovered something uncomfortable: their best engineers had options, and those options were still concentrated in expensive cities. The workers most capable of demanding non-adjusted remote salaries were precisely the workers companies could least afford to lose. The result was a messy, inconsistent patchwork of compensation policies that satisfied almost nobody.
This connects to a deeper truth about how tech companies actually think about talent. The cluster premium is not purely about geography. It is about access to a specific labor market, one where the workers are already pre-sorted by the filter of having chosen to live somewhere intensely competitive. The city functions as a selection mechanism, not just a location.
The Competitor Pricing Problem
There is another force keeping salaries high in expensive cities that rarely gets discussed with the seriousness it deserves: competitive signaling between companies. When Google pays a software engineer $300,000 in total compensation in Mountain View, it is not simply responding to the local cost of living. It is also preventing Meta, Apple, and a dozen well-funded startups from recruiting that person away.
Salary in tech is partly a retention weapon and partly a market-distortion tool. Companies that can afford to overpay relative to any rational productivity calculation do so specifically to raise the floor for everyone else. This is the same logic that governs how platform companies make competition structurally impossible in their product markets, applied to the labor market instead. Outspend the competition until only the best-capitalized players can participate.
Smaller startups trying to recruit in San Francisco or New York are not competing against a rational market rate. They are competing against the deliberately inflated compensation packages of companies with hundreds of billions in market capitalization. The high cost of living in those cities provides political cover for this dynamic. It lets everyone describe predatory compensation competition as a simple supply-and-demand story about housing.
What This Means for the Next Wave of Tech Hubs
Cities like Austin, Miami, and Denver spent the pandemic years marketing themselves as the next Silicon Valley, primarily on the basis of lower taxes and cheaper real estate. The pitch was essentially: come here, pay less, live better.
The agglomeration research suggests this pitch has a ceiling. You can attract workers who are tired of California’s costs. You cannot easily replicate decades of institutional knowledge, informal networks, and the specific density of specialized capital that makes a cluster self-sustaining. The new hubs are growing, but they’re growing by pulling in workers who already developed their networks elsewhere, which makes them partially dependent on the original clusters they’re supposedly replacing.
This also explains why the most innovative companies often hire people who have never worked in tech. In a tight cluster market where poaching established engineers is astronomically expensive, finding talent outside the traditional pipeline becomes a rational economic strategy, not just a feel-good diversity initiative.
The Honest Takeaway
If you are a software engineer deciding where to live and work, the conventional advice (go where the jobs are, accept that you will pay more for housing) is not wrong. But it obscures the actual trade you are making. You are not paying for square footage. You are buying access to a network with compounding returns, paying a toll to enter a market where your labor is valued differently because of who else is in that market with you.
For companies, the lesson is equally clarifying. Paying above-market salaries in expensive cities is not charity toward employees burdened by rent. It is the rational price of participation in the most productive labor markets on earth, markets that are expensive precisely because they are productive.
The causality matters. Get it backwards and you will keep making the wrong predictions about where tech compensation goes next.
