The sideways view

Distributing information is a tricky problem for capitalism.

On the one hand, information wants to be free. For every person who values Stephen’s Sausage Roll at $20 but isn’t willing to pay $30, we are leaving $20 of value on the table. For every person who values a course of Solvadi at $40,000 but isn’t willing to pay $84,000 we are leaving $40,000 of value on the table. All in all it’s kind of a lot of value.

But on the other hand, producing information costs money. There is no capitalist equilibrium where drug companies discover drugs and then give them away for free. And as a society I suspect we already produce way too little information due to the difficulty of monetization.

So what to do?

The obvious approach is for producers to sell their information to the state, who then gives it to consumers, estimates the total social value, and reimburses the producers.

There’s just one wrinkle: how do you estimate the social value, without asking people how much they are willing to pay? And how do you ask people how much they are willing to pay, if you are just going to give it to everyone for free?

A scheme

(Note: Wei Dai has proposed a substantially similar scheme here.)

Step 1: Divide households at random into 10 sectors. Maybe balance the sectors so that they contain a similar number of households from different places, with different income levels, etc.

Step 2: You can voluntarily submit any product to a national Registry of Information Goods (RIG). If you do, you declare a marginal cost and a full cost. The idea is that you should be indifferent to selling an additional unit at the declared marginal cost, and you should set the full cost in the same way you would normally would. But you are welcome to pick these prices strategically using whatever method you want.

Step 3: When your product is added to RIG, you choose a sector. Households in that sector will have to pay the full cost. Households in other sectors only need to pay the marginal cost.

The allocation of goods to sectors is publicly available. If you are making a hepatitis C treatment you’ll want to put it into the same sector as all of the other hep C treatments, so that people in your target sector can’t save money by swapping to a hep C cure from a different sector.

Step 4: For every purchase in the target sector, the RIG pays you 9 * [full cost – marginal cost]. So you are only collecting 1/10 as many [full-marginal] payments as you usually would, but each of them is being multiplied 10x. Hopefully that means you are roughly indifferent to whether your product is in the RIG.

In order to get paid you need demonstrate a strong equivalence between products sold to the target sector and to the other sectors. You should not have any internal policies that depend on whether a customer is from the target sector or another sector, ideally you wouldn’t even be able to tell (e.g. if you have to interact face to face with the customer, the representative shouldn’t know which sector they are from). You can’t advertise more aggressively to people in the target sector. And so on.

Step 5: The RIG pays sellers an extra bonus to incentivize participation (and to grease the political wheels). Maybe they increase your payments by 10%, so that you are making about 9% more money than you usually would.

We’d want to be sure that this incentive payment was smaller than the value of the extra products they are giving away, otherwise the public may be justifiably skeptical. Moreover, if the products being given away aren’t significantly more valuable than the incentive payment, then we didn’t want to incentivize participation anyway. That is, if the number of products sold in the target sector was the same as the number sold in other sectors, then the RIG isn’t adding any value and so we don’t really want to provide an incentive payment.

The RIG should also reward targeting under-represented sectors, for example by adjusting the incentive payments so that goods in each sector receive the same total payment. This could help counteract some natural pressure for goods to clump in the same sectors, and also improve equity amongst households. At equilibrium it seems like we should be able to have pretty balanced sectors with minimal efficiency loss.

Step 6: The federal government levies a broad-based consumption tax to fund all of this. If the RIG is successful, then this may end up being a really big tax, think like 2-6%. But it would probably be accompanied by a larger decrease in average household spending, since RIG payments are offsetting free products given to households which they would have otherwise paid for.

If the rich and poor consumed information goods at the same rate then the net effect would be distributionally neutral. If poorer people consume much less information (I don’t actually know) then you may need an additional redistributive process to make this policy equitable–perhaps you would want to use something less regressive but more distortionary than a consumption tax, or you could bundle it with some other more efficient redistributive policy.

Some problems

People might prefer buy goods on the black market rather than paying for them. It may be  harder to clamp down on the black market when you are in fact giving away the product for free to 90% of consumers. We could try to use an approach like Steam, where you would have to basically break the system in order to move products between different customers.  It’s already pretty easy to steal many kinds of IP, and so it’s not clear this will actually make things any worse.

The black market situation is trickier for drugs, where controlling production is actually a pretty good way to limit access. We could have the same kind of prohibitions that currently exist on giving other people prescription drugs, but I don’t know if that works. If this is a critical problem for some products, they just won’t be put on the RIG.

It sounds hard to enforce equivalent treatment of customers in target and non-target sectors. A seller would be happy to give a 5x kickback to target sector customers. So sellers are going to be looking for info that would let them discriminate in favor of target sector customers, and target sector customers are motivated to help them.

To really cause damage this kind of discrimination would have to be a pretty broadly-based, so it might be possible to keep it in check with whistleblower bonuses and string operations (with violations causing a product to lose its RIG compensation for the preceding year). You could also make it illegal for customers to provide info about their real type, and count on customers to try to pretend that they are from the target sector. In the worst case you could limit RIG purchases to arms-length transactions through a common marketplace.

If firms can perfectly price discriminate then the RIG doesn’t add any value, because everyone is already buying the product at exactly their willingness to pay. If firms can’t perfectly price discriminate then I think the optimal strategy is to overstate the marginal cost and use the RIG to help with price discrimination, which slightly reduces social value (and also shifts some value from consumers to RIG sellers). This is a complicated topic but overall I don’t think it’s a big deal. Ideally the RIG would facilitate perfect price discrimination, which would be extremely efficient though may be unpopular based on shifting surplus value from consumers to producers.

This mechanism would probably only work for items that are consumed by the individual or household. It doesn’t really make sense for intellectual property sold to businesses or games that one friend purchases and all friends play.

I expect the largest political problems would be the significant cost of the program, the potentially disruptive effects on affected industries, and the complete incomprehensibility of the proposal to the majority of voters.