The sideways view

As far as I can tell:

• The largest component of a Rolex’s price is its elaborate time-keeping mechanism, which may cost around $1000.
• This mechanism has no effect on the aesthetics of the watch (since you can’t see it).
• This mechanism is not a useful way to tell time. A quartz watch will keep time more accurately with less trouble.
• However, it is hard to fake the exact appearance of a Rolex, and in practice a Rolex is easily distinguishable from a cheap imitation.

This is a textbook example of burning resources, and then wearing a hard-to-fake signal that you burned them. Someone who sees the Rolex knows that you have resources to burn (and perhaps that you are able to keep up with fashion).

While writing this post I searched for someone explaining expensive watches, to see if they offered any rationalization beyond the costly signaling. Google’s first result is Richard Holt’s an expensive watch is money well spent, which makes my point better than I could:

By spending more, you might not get a watch that tells the time any better, but you will get something that has had many, many hours of research, design and painstaking work…

As a society, I think it would be much better if we didn’t spend our resources making thousands of copies of nice watch mechanisms, and instead did basically anything else. From my perspective, the main question is: how do we get from here to there?

Scope and scale

Worldwide, we spend a few hundred billion dollars a year on jewelry. That’s something like 1/500th of our total output. We spend a much larger fraction of our effort buying other expensive and fashionable things, doing impressive and difficult activities, being generous in inefficient but visible ways, etc.

Signaling often serves a useful social function—for example, it’s useful for employers to know who can actually do the job, or for friends to know who actually cares about them. These benefits are usually captured by the signaler. But costly signaling almost always creates negative externalities as well. That means we probably have too much of every kind of signaling, not just Swiss watches.

It’s easiest to see certain overt signaling, like wearing a nice Swiss watch that serves few plausible non-signaling functions. I’m sympathetic to the theory that most signaling is strongly optimized to be covert and deniable and to have a plausible cover story. I would not be at all surprised if someone told me that the deadweight losses from signaling swallowed 10% of society’s productive capacity.

By way of illustration: suppose that when anyone makes a Facebook post they can spend a little bit more time making it witty/clever/wise. People who write wittier Facebook posts stand to benefit when others think better of them. But this game is mostly zero sum. The existence of wittier Facebook posts does not increase our overall estimate of the cleverness of society—it does not increase the sum, over all people, of their apparent cleverness. (At least not to a rational observer.) So when Bob writes a clever post and his apparent cleverness goes up, other people’s apparent cleverness must go down. Under these conditions, a benevolent social planner would have people spend less time trying to look clever.

Robustness of the problem

Costly signaling is an interesting kind of cost: if the cost doesn’t hurt, then people will just signal harder. So it’s unusually hard to reduce the costs.

This is especially true when the virtue-to-be-signaled is the signaler’s willingness-to-pay. But even in other cases, where it is possible to reduce costs, the dynamics of signaling tend to make it unusually difficult. (Though see Katja Grace for a discussion of cases when it is easy.) For example, if you made it easier to be witty, I expect that everyone would up their game and that quality of life would not increase much if at all.

A common response by idealistic young people is to lament gaudy rich folks’ desire to keep up with the Joneses, and perhaps to push for a culture which is aggressively dismissive of positional goods and signaling more broadly.  When I squint, I see the same phenomenon in the hipsters’ hatred of consumerism, everyone else’s hatred of the hipsters, and so on.

I think that eliminating costly signaling is a hard change, perhaps just as hard as making it cheaper. My default expectation is that if you take one mechanism for signaling off the table, another will crop up to serve the same function. If you clamp down further on overt signaling, then signaling will become more and more covert and obfuscated.

In fact, we already have pretty strong norms against overt signaling. It’s not clear if such norms improve our situation—we might well be better off if people literally put their wealth on their name tag rather than spending on the Rolex.

So, what can we do? If we can’t eliminate costly signaling, and we can’t make it less costly, is there any way to fix the deadweight loss?

A warm-up

Suppose that you lived in a world where it was typical to spend $5000 on a diamond engagement ring.

What would happen  if scientists published a method to synthesize identical diamonds for 10% of the cost?

By default, I expect that synthetic diamonds would come to be considered tacky—even if people can’t tell the difference visually (or can only tell with great effort), there are other ways they can learn how much a ring cost. In social circles where expensive engagement rings are typical, buying a cheap ring would send a bad signal.

But the synthetic diamonds present an interesting opportunity for an enterprising government ready to act quickly. Here’s a possible scheme:

• Impose a 900% tax on sales of synthetic diamonds (and a 900% import duty).
• Make a credible commitment to raising taxes to maintain rough parity between synthetic and natural diamond prices. Preferably harmonize this across the developed world.
• Spend the proceeds on popular government programs (the shiniest parts of education and public health, etc.)

After implementing these policies, the synthetic diamond rings not only look identical, they have the same price tag. They are also the ultimate in “ethical diamonds,” since the money is directly helping people. Indeed, buying a normal diamond is kind of like saying “fuck you” to the person whose education you might have funded.

Of course, manufacturers of synthetic diamonds should be happy with the result, if in fact the result is a shift from real to synthetic diamonds. So they should be more than happy to help implement such a tax.

The result is that people keep spending exactly the same amount on diamond engagement rings. But now instead of the resources being spent on diamond mining, they are being spent on socially useful projects.

Generalizing

While it’s hard to make costly signaling less costly to the individual,  it’s not necessarily hard to make it less costly to society. The best kind of costly signaling is when the value is being efficiently transferred from the signaler to someone else.

In the diamond ring case we imagined doing this near-perfectly. In reality almost nothing is so crisp, but we could still try to push in the right direction (disclaimer: I have zero idea if any of these are sensible proposals):

• Tax positional goods, with proceeds used to fund the kinds of things rich people want to be seen as supporting.
• Normalize and celebrate conspicuous philanthropy as a substitute for conspicuous consumption.
• Provide additional trappings that make it easier to signal philanthropy. For example, high-profile designers/artists could designate particular items as philanthropic, giving away most profits from sales of those items. Or we could normalize “ethical” versions of positional goods that are accompanied by large donations (e.g. large carbon offsets to accompany extravagant houses).
• Try to publish more unbiased estimates of on-the-job performance, especially for young employees. Eventually they could displace more traditional early-career signals like educational attainment that plausibly have big signaling components. There are plausible (signaling) stories about why firms would underinvest in such measures.
• When there is a socially salient opportunity to help a friend, rather than spending N minutes helping them, help give them a credit for N minutes of help-at-any-time. They can redeem the credit immediately, or if it’s not a great opportunity then they could save it for a more efficient opportunity.

For the rationalists: rather than disdaining signaling because its signaling we could selectively disdain the forms of signaling that are socially inefficient, while remaining aware that the signaling must go on. As far as I can tell, a subculture that wants to do this (without radically breaking with human nature or the rest of society) needs to embrace or at the very least tolerate slyness and hypocrisy, and to maintain a mutual understanding of the norms they are trying to cultivate without explicitly articulating it. They would need to play the games that people normally play (most of the time), with a smile and a wink rather than a philosophical commitment to calling people out on their bullshit (most of the time). Hopefully they could end up with a subculture that was no less hypocritical than normal, but much more efficient.

(Independently, it may or may not be worth it to try to maintain a not-hypocritical subculture. I think that’s a radically harder project which the rationalists are not (yet?) qualified to implement.)

Conclusion

Losses from costly signaling probably consume a significant fraction of global productivity, and the dynamics of costly signaling make it especially difficult to reduce the  costs to the signaler. But that doesn’t necessarily stop us from decreasing the social cost, by ensuring that the signaler’s loss is someone else’s gain. I’m not sure if this idea could actually significantly reduce signaling costs, but it seems worth thinking about and having in mind.

(Discussion on LessWrong.)