There are lots of concepts in macro that sound sensible in theory but often fail to usefully describe reality: the Phillips Curve, MV=PY, and Barro equivalence are a few I came up with off the top of my head.
Another example is "potential GDP," which is supposed to relate real growth to the rate of inflation. I did a longer debunking of the usefulness of the concept last week on Alphaville, citing, among other things, the fascinating new paper on "equilibrium real interest rates" by Ethan Harris and Jan Hatzius. (The concept of "an equilibrium rate" is also a bit daft, for similar reasons, but if you want more on that you should read the full post.)
Anyhow, @georgepearkes recently wrote a post chastising me for saying that potential is "almost certainly bogus". His argument is that "potential" is a theoretically sound concept even if it is so hard to apply in practice that "it shouldn't be given too much weight". Moreover, he goes on to describe my statement as equivalent to claiming that "there are no resource constraints in an economy."
I completely disagree with George's characterization of my position. Our world is filled with all sorts of constraints. And I probably am more likely than many to think that there are harsh limits on the rate of sustainable growth. I still think that the theoretical underpinnings of "potential GDP", at least as usually understood, including by me and George, are almost certainly bogus.
The issues here are:
- Which constraints matter?
- What happens when we hit those constraints, and why?
The standard answers to these questions from believers in "potential GDP":
The constraints are the labor supply, the capital stock, and the productivity of both labor and capital.
If real growth exceeds "potential" for too long, the result is an acceleration in the general level of prices. Generally speaking, this is because "insufficient" unemployment empowers workers with "excessive" leverage over their employers, so they will demand bigger income gains than can be justified by their productivity. To preserve profits, companies will simply raise prices, leading to the dreaded wage-price spiral.
Since George readily admits that it's really difficult to usefully anticipate what will happen on the supply side, let's focus on the theoretical argument, which has lots of problems. I talked about a bunch of them in the post I linked to up top, but here are a few more:
The linkage between wage growth and consumer price growth is relatively weak. One important reason is that, contrary to what mid-20th-century economists used to think, labor's share of income isn't a law of nature. Moreover, we don't really understand what determines the income split between workers and owners over time. (There are lots of believable theories about globalization, politics, etc, but nothing that conclusively answers the question.)
Everyone reading this knows Milton Friedman's line about "inflation is everywhere and always a monetary phenomenon". Now, I haven't the faintest idea how to precisely define and quantify what he meant by "monetary phenomenon", but I do know that his point was that real economy variables don't have a bearing on the trend level of the price index. No wonder Friedman came up with the idea that the unemployment rate will end up being whatever it will be, no matter the inflation rate, and therefore it isn't worth attempting to boost inflation for the sake of reducing joblessness.
You might disagree with Friedman, but the big acceleration and subsequent deceleration in inflation in the US wasn't caused by a big downturn in the jobless rate followed by a big upswing in the jobless rate. Here's a comparison of the annual change in the core PCE deflator against the year minus year change in the unemployment rate. If anything, it almost looks as if unemployment increased at the same time as inflation accelerated, perhaps because of the impact on risk premia and the cost of capital. (Or not, heck if I know.) There's a reason why "stagflation" and the "misery index" are part of the lexicon. And of course the big disinflation of the 1990s was associated with falling, rather than rising, unemployment, while the massive drop in employment in the crisis wasn't matched by a commensurate slowdown in inflation.
Getting back to George's original point, why do we assume that the result of hitting a "resource constraint" is faster consumer price inflation? It could just as easily be a recession, or simply a slower future growth. There has to be a mechanism that creates the inflation separate from the real constraint. Maybe it's inappropriate monetary policy, maybe it's something else. Put another way, how does real output keep growing faster than its theoretical speed limit, thereby theoretically producing inflation?
And if you believe in this story, why wouldn't you pay the price of permanently faster real growth by tolerating faster and faster inflation? Something in the "potential GDP" narrative doesn't add up.
So what might actually matter? I'm going to go out on a limb and say that the constraints we should really be focused on are financial, rather than real.
I said before I don't know how to define and quantify what Friedman meant, but one reasonable way to think about inflation is that people buy things with nominal spending power. That's basically income plus credit but other things matter too, including changes in savings behavior and unrealized capital gains. Prices are a reflection of how much spending power people have (and want to use) relative to how much actual stuff is available to buy.
Of course, the supply of stuff isn't static. Producers attempt to make more stuff when more nominal spending power flies in their direction. Since real GDP is basically the amount of stuff that gets made (yes, a lot of "stuff" is technically services, but you get my point), an increase in nominal spending power generally leads to some combination of higher prices and a higher level of real GDP.
(I'm including the prices of assets in "prices" but the forces that affect what gets spent on things that show up in the PCE deflator and what gets spent on assets are complex and tricky to measure. One reason I don't think consumer price inflation shouldn't be given too much weight as a signal of imbalances but that's for another time...)
The key question is where the extra nominal spending power comes from. There is a limit to how much you can boost your nominal spending in the absence of income growth. Sure, you can dis-save and comfort yourself by counting on the unrealized capital gains in your illiquid portfolio, but there are limits to this process. (I won't pretend to know how to define or measure these limits, but it shouldn't be too hard to imagine what happens when the constraint gets breached.)
Another way of looking at this is to consider the perspective of a producer rather than a consumer. In a world where people suddenly have a lot more nominal spending power at their disposal, it makes sense for you to build out capacity to service the heightened appetites of your customers. The problem is if you invest on the assumption that the growth you're observing in spending power is sustainable rather than, say, ever-increasing levels of indebtedness. After the constraint is breached we end up with an overhang of excess capacity.
(Producers here can also be thought of as asset issuers, with the consequence being "excessively low" prices.)
Borio et al at the BIS wrote a great paper a few years back arguing that you need to incorporate information about financial relationships into any useful measure of potential GDP, although they were quick to note that their improved measure is just as fraught with measurement issues, etc, if not more so, than the conventional definition of "potential".
There are plenty of constraints on the sustainable growth rate of GDP. But the problem isn't that we run out of "real resources" and pay the price by enduring faster inflation. The danger is that the excessively rapid growth in real GDP is driven by unsustainable financial relationships.