A failure to communicate on UBI

As some science fiction would have it, in the future we will have things mostly taken care of for us. Although there will always be limits, money will no longer be an issue, or at least it will be a distraction in life. But is this really the direction the world is heading? It’s hard to see what job couldn’t be replaced by a robot in the span of a career now, or what competitive advantages humans will even have left in couple decades. This discussion has sharpened around a proposal for Universal Basic Income (UBI), which as you might guess means receiving a check from the government every month just for being a human.

Over the past year or so, as the topic has trended, I’ve been following discussion of UBI on a number of forums and discovered a fairly wide gulf between two positions whose proponents might be characterized as “technologists” and “economists.” The technologists are pessimistic about the ability of their projects to solve inequality without specific government policies. The economists, on the other hand, insist that things will return to normal equilibrium after the financial crisis and accompanying global imbalances recede. What’s been frustrating is that the terminology used on both sides is so different that there seems to be a lot more disagreement than there really is.

I won’t try to recreate these arguments in full, but if you’re interested in UBI and the discussions surrounding it you’ve probably heard their basic contours. The technological argument is that exponentially increasing computing power will eventually result in superhuman intelligence, and various versions of this story assign it benevolent or malevolent motives. We are approaching a moment like the dawn of humanity in natural history, where all previous assumptions will be thrown out the window — except that it’s the dawn of the robots. Much of this discussion focuses on cyber security, but there is also an unmistakable economic element.

Economists, for the most part, are not having any of this. Think about physicists’ reaction when young earth creationists suggest that maybe the speed of light could have changed over time. Maybe it could have. Maybe nothing is real. But as a scientist, your job is to find the fundamental constants that underlie everything, so economists look to the past as a guide to the future. Here you will see counterarguments referencing the Industrial Revolution and other periods of technological change: 95% of farming jobs have been destroyed through automation and other efficiencies, yet we don’t live in poverty.

The process of creating new sectors to replace the old ones destroyed by innovation is not at all straightforward, however, and there is still plenty of room for technological fudge factors. The economists are right to pick on the quantitative thinking implied by concepts like “technological unemployment.” According to the marginal worldview, everything has a price.The resulting situation is better understood as “technological deflation.” Although some textbooks refer to this process as “benign deflation,” this is a freshman-year simplification and it’s easy to show using the basic models how it could turn into something worse.

I think the real academic economists realize this possibility, but are a bit shy in public. In fact, a lot of people on this side of the debate are not academic thought leaders, but people in industry and talking heads on TV who may not understand the implications of their model when historically unusual conditions are inputted (with respect to anyone highly qualified who may disagree with any of my points.) People making the same argument I am in public have been laughed out of the room, giving the impression that UBI is something that’s far outside of the mainstream of economic thought.

The most fundamental problem in macroeconomics is the zero bound. If interest rates go below that, savers will generally hold cash rather then lend capital. (Some central banks have experimented with negative rates recently, based on the idea that large corporations can’t keep physical cash in warehouses — but there’s clearly still a tight limit on how far that can go.) The problem will all this is that rates sometimes ‘want’ to go lower. The following equation shows the simple logic that loans must be paid back according to projected future price levels, not today’s:

nominal interest rates = real interest rates + inflation expectations

So if the magnitude of deflation expectations is higher than real interest rates, nominal rates hit the zero bound. It’s important to note here that this principle holds no matter the original reason for deflation — whether we choose to call it good or bad.

The following chart (the only one I’ll need here, I promise) shows the relationship between aggregate supply (AS), aggregate demand (AD), price levels, and output (GDP).

Interest rates that are too high, as in the zero bound, don’t feed enough money into the system, reducing spending power, moving AD to the left and down. When we talk about “benign” technological deflation, on the other hand, that means moving the AS to the right, or equivalently down. Output increases — which is good — while prices fall. The problem is that “good” AS deflation can lead to “bad” AD deflation through interest rates.

Often this AS deflation isn’t problematic, because technology affects both the supply and demand sides, leaving prices where they were originally but increasing output. As our basic needs are taken care of, we start to demand new scarce goods. But think about what this entails: Creating new demands is the holy grail of entrepreneurship. It requires that consumers feel secure enough in their financial positions to start having higher standards, and presumably that the economic benefits of the technology be spread to most people, rather than concentrated among a few. To underscore how mysterious this process is, even talking about “technological inflation” sounds strange, but a rightward movement is equivalent to an upward movement in the AD curve. This is exactly the shift that needs to happen for new industries to form.

It usually does move, eventually. But I want to emphasize here that there’s nothing in theory that says how long this process will take, how far it will lag AS, or even that it must take place at all. We can’t even measure when it does occur (or at least separate technology from other factors like monetary policy.) Here’s what I want the economics people to internalize: that any factor which boosts AD (without knocking back and raising AS prices more than desired) can be considered a ‘technology,’ just like the more familiar technologies that are aimed towards AS. So the next thing I want to talk about is why that can be so hard.

The central bank can always manufacture inflation — it’s just a question of how far it is willing to go. Its first line of defense is to lower interest rates. In that case, it signals its willingness to lend money to the banks, but the banks don’t necessarily have anywhere to lend it from there. Suppose then that they do lend it to productive companies. The companies themselves would have to spend it on projects that will pay back over time — either new equipment, or new hires, which can be a risk in themselves. As I mentioned above, if interest rates go all the way down to zero before this succeeds, then this process doesn’t work anymore.

The second line of defense is to buy securities, inserting money directly into the market. This (originally) unconventional monetary policy still has all the potential problems of interest rates, of getting the money from treasury bonds to consumers’ pockets and convincing them to spend it. In addition, responsible central banks are extremely skittish about the quality of securities that end up on their balance sheets, so the supply of safe enough assets is limited. Furthermore, both of these solutions involve debt that has to be paid back later, creating potential instability.

When all else fails, then, the central bank can stimulate demand simply by giving citizens money directly: UBI. Milton Friedman coined the term “helicopter money” for this situation, but when economic policy makers have used the term, the press has treated it like a joke. Google it and you will see that it is often assumed to be a term of abuse, even when used in its exact intended context. Despite this, serious economists have been talking about the idea in the past couple of years like never before — although not necessarily in the context of technology and AS.

I want to be clear that helicopter money is slightly different and less radical than many UBI proposals. Its goal is the revival of a consumer economy, rather than guaranteeing individual well-being. It doesn’t restructure existing social programs, and it is not unconditional either, rather reflecting macroeconomic conditions. But if the pessimists are right, those macro conditions will keep recurring, making these two proposals equivalent.

Now, after having fully outlined the technologist argument in economic terms, comes the question of whether it’s actually right. The main point of skepticism comes from the productivity paradox: “ “You can see the computer age everywhere but in the productivity statistics.” Productivity means taking GDP figures and filtering out certain easily identifiable causes. The remaining special sauce is considered to reflect the underlying level of technology in a country (although technically it’s anything that could affect growth, like changes in business practices, government, education, or more.)

But wait a minute — GDP figures are based on actual transactions with both a buyer and a seller. Most of this discussion naturally revolves around the impact of technology on supply; very little of it has to do with its effects on demand, like inequality. Take for example, the trend towards offshore outsourcing of the 90’s. This is clearly an example of AS deflation, as it allowed for cheaper production. But unless some other industry takes its place, this also reduces onshore production, and it therefore doesn’t contribute to GDP growth, statistics of which are then fed into productivity. When the reasons for the productivity paradox are discovered, whatever they are, they will almost certainly have something to do with the demand side.

As for the claim that change is happening faster than before — here’s some interesting research from the Bank of England showing that interest rates are as persistently low as they have been in 5,000 years.

Now, I haven’t shown that this is due to AS deflation — although I have shown that it would be impossible to isolate that as a specific cause anyway given the way these statistics are calculated. The fact that the zero bound problem is quite real, however, regardless of the cause, suggests that we will need to think about even less conventional monetary policy sometime in the near future. Most importantly, this chart gives some basis to this sense that there is something different about the way the world is heading now from where it has ever gone before.

Now the technological singularity argument also has some points that need picking over, particularly with regard to anti-monopoly laws. Situations of rapid change disfavoring workers tend to empower the most powerful individuals in a society, which has inspired legal changes in the past. It’s unclear if the singularity argument means to assert that having multiple competing intelligences won’t resolve any potential problems, or that anti-monopoly laws won’t succeed in forcing competition in the first place. Some points raised by writers on AI have the flavor of computer whizzes who have just rediscovered problems that philosophy and the social sciences have been discussing for centuries already. (How will the superhuman intelligence know what’s actually in the best interests of the human race?)

This is a new and confusing era, and there are a lot of strains of thought going around who aren’t necessarily talking to each other. Economists worried about deflation risks don’t have technology on the front of their minds, because the supply side can’t be measured independently from demand. Those worried about technological unemployment, on the other hand, seem oblivious to a lively parallel current debate among wonky (though not so much popular) economists. I’m not taking a strong position on anything myself — but could you two just talk to each other?