On Perks

Every day, companies around the world execute a low-risk, high-return arbitrage: they buy the time and attention of some of the most highly-compensated professionals in the world — programmers, product managers, data scientists, quants, investment bankers, and lawyers — for the cost of a plate of pad thai. Given how much wealth is generated by companies that comp or subsidize employees’ food, and given that for each of those companies, getting smart people to work a little longer is a core competency, free food arguably belongs on the list of important technologies that have a visible impact on GDP growth.

I’ve recently started getting lunch with friends who work at big tech companies. From the perspective of the technical employees and managers, a profitable tech company is a socialist paradise, with monopolistic ad pricing instead of North Sea oil. You get your badge, walk past the fitness room, the nap room, the meditation room, and the lactation station, then get all the free food you want.

Restaurants are a hard business, unless you locate your in fabulously expensive commercial real estate and charge zero dollars.

Free lunch, though, is not where the action is: the economics of free food are driven by free dinner. Lunch just amortizes the fixed cost of food prep facilities. Free food in general is a great deal for companies:

  1. It encourages people to work much longer hours. Dinner is a natural reason to go home, and if good dinner is available back at the office, it can potentially stretch into another couple hours of productivity.
  2. On-site food encourages cross-pollination between departments and job functions. At a small company, everyone knows what everyone is doing. At a 500-person organization, you have no idea — but if someone on the backend team has tacos with someone from the mobile team, they might realize they’re duplicating efforts, or hatch a new idea for a feature.[1]
  3. At a minimum, it saves a little bit of search time and wait time. If you were going to work until 10, dinner that takes five minutes rather than thirty still saves you five minutes.
  4. Free food lets a company impose lifestyle creep on otherwise frugal employees — especially on frugal employees, for whom “free” is extra-enticing. A low personal burn rate is a force multiplier for an early-stage startup founder, and by getting people accustomed to on-demand vegan sushi BigCos force employees to downgrade their lifestyle when they strike out on their own.
  5. Edit: a reader notes another reason: perks have favorable tax treatment compared to employees buying their own food. Or at least used to.

It’s also important to note what generous perks discourage: outside friends, and having a family.

This is not a sinister plot from Big Corporate; I’m convinced that companies that offer free food are calculating cynically, but companies that offer onsite laundry and egg freezing are doing their best to provide a generous benefits package to employees. Rather than take the conspiratorial view, I take the Darwinian one: companies can offer benefits that encourage family formation, or that discourage it, and the biggest and most profitable companies tend to offer benefits that, on net, discourage it

Even company-sponsored drinking events are an example: it’s not a plot to make the firm inhospitable to anyone who a) has kids, and b) actually likes them. It just has that effect as a second-order consequence of keeping people out late and sending them home buzzed.

By definition, humans respond to incentives best when we approximate the behavior of Homo Economicus, and H. Economicus is an unmarried orphan who doesn’t have any friends and works nonstop until he dies. Many popular fring benefits encourage people to defer family formation and maintain social ties to coworkers rather than outside friends. Homo Economicus has never left work early to go to his kid’s recital, or worked from home because someone got hand, foot, and mouth disease.[2] This is great for ramping up work hours, making employees more responsive to financial incentives over the prospect of leisure time, and generally helping the employer think of you less as someone with 40 hours a week and more as someone with 168 of them.

Even the benefits that aren’t directly related to family stuff tend to encourage long-term singledom. Onsite laundry, rapid-response janitorial personnel, and similar benefits are a way for a company to do your chores.

Compared to other people who share a dwelling, spouses divvy up chores with lower transaction costs. People marry for many reasons — love; lust; the realization that they’ve found The One; the realization that they’ve found one of the N such that N is the number of potential marriage partners past some threshold, where N is shrinking fast enough that the high search time for finding another N implies that their current partner is the best they’re ever likely to do[3] — but one key reason urban professionals marry is that they’re sick of roommates. My wife and I maintain a cleaner apartment and eat better home-cooked meals than either of us did when we were single, and the low transaction costs of allocating chores are a big reason. (Love is another explanation, but that’s not very rigorous.)

It’s cynical and reductionist to think that marriage is merely driven by the economics of the household, but it’s a fair description. The institution of marriage, at least, evolved around what it was good at, and that counts. You-singular have a unique and touching story of how you met your true love, and when you tell it it brings a tear to the eye of anyone with a soul. But you-plural are a meat-robot maximizing total utility with bloody-minded ruthlessness, albeit constrained by imperfect information and sometimes-flawed mental heuristics.

Even paternity leave selects against family formation, by privileging households with two roughly-equal earners. In a household with a primary earner or single earner, a long guaranteed leave period just means it’s harder for that person to get back to work and get the next promotion. But if you’re dividing up labor, you might find that there’s a division of labor between full-time career work and nontraditional employment, particularly as you move up the income distribution. If we want a maternity- and paternity-leave system that actually encourages maternity or paternity, give each parent the option to cash in some of their time off and get back to work.

While it’s clearly good for companies, it may not be good for society if the smart and conscientious people who work at these companies are delaying or even avoiding having families. And in my experience, many people in their 40s (not to mention their 60s or their 80s) don’t look back on their 20s and 30s and wish they’d done less to ensure a steady supply of grandkids. However, there is no equivalent to the B Corporation for maintaining traditional family structures, and companies are quite good at responding to incentives, as long as the incentives are there.

Lavish perks shape the social structure of upper-middle-class and above workers, in counterintuitive ways. And if Paul Fussell was right, your social class consists of imitating what you think the class above you does. It would be pleasant for the average worker if the first-order effects of generations perks trickled down the class strata, but we should take care that we don’t form a system in which people work long hours at challenging, fulfilling jobs, and then die alone.


Don’t miss the next story. Sign up for my occasional email newsletter.


[1] It’s striking that tech companies have cafeterias while banks and law firms give employees food delivery allowances. Some of this is doubtless due to the more mercenary nature of these industries — the correlation between equity comp (you do well when the company does well) and bonuses (you do well when you do well) is no coincidence. There’s probably a regulatory reason as well: a lawyer or investment banker gabbing about work is way more likely to violate ethical or legal standards.

Everyone knows a good company should outsource everything that’s not a core competency, but as the value of employees’ time rises, getting more effort out of them becomes a core competency. In finance and law, it’s straightforward to do this with bonuses; in tech, it’s harder to measure individual contributions except at the extremes; a good product is, in mathematical terms, a product: engineering times product times marketing equals success. So the costs and benefits of perks in tech tend to be more diffuse.

[2] It’s not as bad as it looks, but what it looks like is leprosy. I got it from my 1.5-year-old, who also shared it with her big sister. Fun!

[3] This is why economists marry.