Bitcoin's deflation problem
Add this article to your reading list by clicking this button
TWO weeks ago we published a Free exchange column examining whether Bitcoin could be considered a true money, and if not, why not. Mike Hearn, one of Bitcoin's most prominent software developers, responded to the column somewhat dismissively. I wrote an e-mail response to Mr Hearn, the gist of which I will reproduce here. He makes two broad criticisms. The first is that we have lazily repeated the argument that deflation will kill Bitcoin, which in his view has been debunked. And the second is that we are naive to think put much faith in official inflation statistics.
I think Mr Hearn may have misunderstood the piece's argument. It was not that deflation would kill Bitcoin. Rather, it is that deflation will prevent Bitcoin from becoming a unit of account, and that, in turn, will keep it from displacing traditional currencies. But Bitcoin could survive and indeed thrive without becoming the coin of the realm.
The issue, as the piece explains, is that deflation in the unit of account leads to unemployment, thanks to the fact that wages generally don't adjust downward. Mr Hearn suggests that the idea that deflation might be costly is controversial among economists. I must disagree; it really isn't. Economists would love it if he were right that deflation didn't matter—that money, in economists' parlance, is neutral. If wages adjusted quickly and cleanly then they could go back to applying really straightforward classical economic models and everyone's life would be simpler. But the data are very clear on this point; wages are "sticky", and so deflation in the currency in which wages are set is costly.
I understand there is a general lack of faith in official inflation statistics in the Bitcoin community, which is perfectly reasonable. What is less clear to me is the assumption that weaknesses in inflation statistics cause them to systematically understate inflation. There is good reason to expect that inflation is actually being overstated in some areas of the economy. Statisticians aren't very good at taking into account rapid improvements in the quality of technological goods, nor do they do a good job capturing when technology sends the market price of some goods (like encyclopedias, for instance) to zero. On the whole, I'm inclined to believe that various biases generally offset each other, so that the official inflation number is broadly reliable.
Why should I believe that? For a simple
reason: inflation may be hard to estimate but total spending in an economy, or nominal GDP, is much less so. Calculating nominal output does not require any value judgments about the comparability of certain goods, or much in the way of statistical wizardry, so we can feel reasonably confident that such figures, while imperfect, are not especially biased. Mr Hearn cites housing prices as evidence that CPI understates inflation. Let's assume he's right, and that actual inflation from 2004-2006 was in fact much, much higher than the official estimate. Here we run into a problem. Since we have good data on nominal GDP, a much higher rate of inflation implies that real, or inflation-adjusted, GDP was actually performing terribly during this period: that America was actually in severe recession from 2004-6. But during this period American unemployment was falling rapidly and the economy was adding jobs at a healthy clip, which is definitely not what happens during severe recessions. One could say that the jobs figures are cooked, but there are lots of private estimates of job growth, industrial production, and so on which also indicate that America was booming at the time, not busting. When the data all point in one direction like that, I tend to trust it.
One final point: I think it is a mistake to view deflation or zero inflation as the idyllic, uncorrupted state of a monetary system. Money is not a natural thing. It's a technology that society deploys to meet certain needs. Different monetary systems imply different things for the price level in an economy, with vastly different distributional consequences. To say that one set of distributional consequences is right while another is not would be wrong; they are what they are and society chooses which it prefers. There are costly side effects to -2% inflation or 0% inflation just as there are to 2% inflation or 15% inflation. My personal view is that the societal consensus behind low but positive inflation that prevails at the moment makes a lot of sense.
More broadly, a hard supply cap or built-in deflation is not an inherent strength for a would-be money. A money's strength is in its ability to meet society's needs. From my perspective, Bitcoin's built-in deflation means that it does a poorer job than it might at meeting society's needs. Maybe I will be proven wrong. We shall see.
Technology: We, robot
Book clubs: Reading "Capital": Chapters 5 and 6