A couple weeks ago I saw this on Twitter:
My immediate reaction was, “This sounds pretty dumb.” After all, printing money doesn’t generate new wealth.
After a few seconds I revised my opinion to, “This is probably a terrible idea, but for reasons I don’t understand.” What I realized was that while printing money doesn’t generate wealth, it certainly redistributes it to the government (which is what taxes do). By printing money, the government would decrease the value of every dollar, thus devaluing all assets priced in US dollars but increasing the total value of assets held by the government.1
So why did I still think it was a terrible idea? Well, the global financial system is a really complicated system that I’m nowhere close to understanding. Messing with it would almost certainly have bad side effects.
Indeed, a few minutes later I realized one big problem with the government printing money instead of collecting taxes: printing money is equivalent to a flat tax on wealth. It’s not obvious how one could change the system to make the tax not be flat; it gives the government essentially no flexibility over the taxation system.
Now, it’s conceivable that this difficulty could be resolved. But even supposing that I came up with a clever way to resolve this problem, I would still think that printing money instead of taxing people is a terrible idea. This is for the same reason as earlier: there are probably lots of other problems with the idea, even if I don’t immediately see them.
But now suppose that I asked several economists what they think of printing money instead of taxing people, and they all said that there’s a clever implementation of this idea that hasn’t been tried but would bring about substantial economic benefits with no foreseeable side effects. Would I then support the idea?
Well, maybe…
***
I’m a pretty big believer in deferring to expert consensus. After all, if the people who have thought about a subject more than anyone else all share the same opinion, then it almost always makes sense to adopt that opinion. After all, if you were to learn what those people know, you would likely share their opinion. For example, if you were to learn that historical linguists believe that Tagalog (the most spoken language in the Philippines) and Malagasy (the most spoken language in Madagascar) share a common ancestor, you should believe them even if you don’t understand how they could have come to such a conclusion.
So why would I be hesitant to support the government printing money instead of charging taxes, even if I learned that economists think that this would have substantial benefits and no obvious costs?
Jacob Falkovich writes about the global financial/economic system:
Finance is … a complex, interconnected ecosystem. [One] lesson we should have learned from 2008 is that banks, hedge funds, insurance companies, and government agencies depend on each other in critical and opaque ways. Eliminating a class of [institution] from the financial system is like eliminating some annoying insect from the jungle: it could make things a little bit better, or it could bring the entire thing crashing down unexpectedly.
The global economic system is a very complicated object, and because of that it is fragile. Making big changes to it have a considerable chance of ending in disaster. For an analogy, suppose a NASA engineer observes a way to substantially change the design of a rocket to make it more efficient. There’s a decent chance that the engineer is right, but NASA would surely test the redesign before sending people on the new rocket, even if the test is expensive. This is because rocket science is pretty hard (it’s the canonical hard thing!) and even if the engineer convinces everyone that the design change is safe, there’s a decent chance that something will be overlooked. Similarly, economists might all agree that making some fundamental change to the economic system would only have upsides, but without first testing the change they cannot possibly be confident in this.
This is why I might not get behind a big reform idea even if the economic consensus supports it. It would probably take more: for instance, if the idea were first tested by smaller countries and there weren’t any major problems, I’d be more inclined to support it. Even then, I’d only support the idea cautiously: the United States economy isn’t necessarily comparable to that of (say) Poland.
More generally, I think human civilization is in a basically good state, in the sense that making random mutations in its structure is more likely to harm than help it. While making a huge change that experts support isn’t a random mutation, it is likely to have unpredictable consequences that may as well be random — and for a big enough change, those consequences may themselves be very large. If a policy is likely to have important unpredictable consequences, then the upside of the policy needs to be very large for the policy to be worth it.
For this reason, there are policy ideas that fall into an awkward epistemic position for me:
- I (and/or experts) think the policy has substantial potential to improve the world.
- I (and/or experts) don’t see any specific downsides to the policy.
- I’m against implementing the policy anyway.
I propose calling this provisional opposition: I provisionally oppose a policy if on the surface it seems good, but I’m scared of it because of potential unknown unknowns.
The reason I called provisional opposition “awkward” is that I might reasonably have the following debate with someone on a policy P that I provisionally oppose:
“Eric, I support implementing P. Do you?”
“No, I don’t.”
“What don’t you like about P? Think about all the good it could do.”
“I agree P could do a lot of good, and actually there’s no particular problem I have with P. It’s more that there could be a big problem, and if there were one, society might not be smart enough to figure it out.”
It’s kind-of unclear how the discussion of P is supposed to proceed from here. We might just have a meta-level disagreement about people’s abilities to precisely and accurately model the world — and such disagreements seem pretty hard to reason about and resolve.
***
Perhaps the clearest example of a policy that I provisionally oppose is dispersing sulfuric acid into the atmosphere to combat global warming. The idea is that just as carbon dioxide causes the greenhouse effect, sulfuric acid causes an anti-greenhouse effect, reflecting solar rays back into space. It turns out that the plan passes basic sanity checks: on the surface the idea appears safe to scientists. However, David Keith, the person who came up with this scheme, “doesn’t want to implement it anytime soon, if ever.” Keith correctly believes that “much more research is needed to determine whether injecting sulfur into the stratosphere would have dangerous consequences.”
It’s possible, though, that no amount of research will convince me that this is a good idea. The Earth is an even more elaborate and delicate system than the economic system, and messing with it could easily produce unintended consequences, pretty much no matter how much the plan is studied. I might be convinced to eventually support it if a lot of research happens and also global warming looks to be extremely dire (to the point where it might cause the deaths of tens of millions of people in the near future). Before that point, such geoengineering plans would not be worth the risk, even if climate scientists don’t know of any specific downsides. Another way to think of it is that if you considered the probability distribution of how much good I expected the geoengineering plan to do, the median of the distribution would be positive but the mean would be quite negative. (And the mean is what matters.)
The geoengineering plan is a pretty extreme example, though. More typically, if I provisionally oppose a policy, I could still be convinced to support the policy based on the results of studies and pilot programs. Here are some examples of policies that fall into this category for me:
- Open borders (free worldwide movement of people with minimal border controls)
- Having an income tax rate near 70% for very rich people
- Implementing a $1,000/month universal basic income
- Making education status a protected class (make it illegal for employers to ask about education level)
I don’t want to give the false impression that I’m just broadly opposed to reform. There are several big reforms that I am in favor of, where I think the pros really do outweigh the cons. But I hope I’ve convinced you that it’s quite reasonable to oppose a policy even if you don’t see any concrete harms that could come from it.
1. This obviously omits lots of details, but I think that this is true to a first approximation. Also, while this plan would increase the inflation rate, it wouldn’t do so by a huge amount.↩
Unexpected Values 
This seems to be a specific example of the general idea of decision making under inertia in which you only take a competing action over the status quo if you are sure the competing action is more beneficial in all states of the world. There’s an interesting literature about the topic in decision theory started by Bewley, might be worth checking out if you’re interested.
LikeLike
That seems interesting –essentially you take an action if it’s a Pareto improvement (over all states of the world). Ultimately I would reject this in practice as being too conservative (not in the political sense — in the “sometimes not acting when you ought to” sense). But it’s perhaps a reasonable guess about what a broad group of people — no matter their ethical axioms — would find acceptable.
Another issue I have is that sometimes there’s no clear difference between decisions such that you can call one action and the other inaction. Or maybe you’re deciding between several approaches, and doing nothing would be worse than all of them.
LikeLike