A Potential GDP Explainer
What it is, and why it's important (to all of us)
Can you recall a time when you were “in the zone”? Perhaps your “zone” was marked by sleeping well, getting up early, exercising regularly, with everything flowing at work and, well, just about everything in your life? Remember that time? Yes? That was your “Potential GDP” state of existence.
Then, sometime after, maybe you fell out of the zone and found yourself in a rut? You felt unmotivated, full of doubt, a wary eye cast towards a seemingly bleak future? You were below your Potential GDP. You were in a recession. Eventually, however, you rebounded from that rut—you charged yourself up with motivational podcasts, excessive coffee consumption or 600-calorie lattes, took on your first half-marathon and perhaps tried for something new in your career or personal life . . . all at the same time? That was your economic boom. You were “running hot,” running above your Potential GDP.
The Macroeconomy is like that, too.
The “zone” for the macroeconomy is Potential GDP—it’s a level of output where macroeconomists think the economy should be, given the state of our technology, the amount of equipment we have (“capital”), and the quantity and quality of our labor force (education, skills, and so on). It’s a concept, found in every macroeconomic textbook that exists.
Why is Potential GDP so important for all of us? It is a concept around which all macroeconomic policy revolves. For example,
Why is the Federal Reserve planning to push up interest rates in 2022? Because they think we’re above Potential. Like a person “burning it at both ends,” the economy is doing too much, all at once.
Why does Congress and the White House rush to pass “stimulus” spending bills when the economy has stalled or contracted? Because we are sliding below Potential GDP, falling into that economic rut.
Macroeconomic policy is centered on, and succeeds or fails by, Potential GDP.
While the term “potential GDP” isn’t often mentioned specifically in mainstream news coverage of the economy, it shows up implicitly when any writer or commentator mentions the “output gap.” The idea of Potential GDP is crucial to the “gap.” More on the “gap” in a minute.
First, let’s take a look at an estimate of Potential GDP, provided by the Congressional Budget Office (CBO). Here is the (somewhat long) definition from the CBO:
“Potential output—the trend growth in the productive capacity of the economy—is an estimate of the level of GDP attainable when the economy is operating at a high rate of resource use. It is not a technical ceiling on output that cannot be exceeded. Rather, it is a measure of maximum sustainable output—the level of real GDP in a given year that is consistent with a stable rate of inflation. If actual output rises above its potential level, then constraints on capacity begin to bind and inflationary pressures build; if output falls below potential, then resources are lying idle and inflationary pressures abate.”
Like I said, Potential GDP is where the economy should be, assuming the things the CBO mentions.
(Note, any researcher could generate their own estimate of “Potential”—an example of which I show at the end of this post. This one by the CBO is the readily-available “go-to” estimate. For the so inclined, the CBO also provides a detailed overview of their projections).
Here is the CBO’s estimate of Potential GDP (alongside actual real GDP):
The CBO provides the estimate out to 2031 (the orange line), with the actual GDP data (the blue line) ending with the third quarter of 2021. The CBO’s estimate for 2031 is 23.8 trillion dollars. Also highlighted on the figure is the third quarter of 2021. The estimate of Potential GDP for that quarter equaled 19.8 trillion dollars. While, actual real GDP equaled 19.5 trillion dollars in that quarter.
Wait a sec? Why is there an estimated value for Potential GDP in a quarter for which we already have an actual real-live estimate of GDP? The reason is that Potential GDP—the entire series estimated back to 1949 and out to 2031—is the result of a forecasting model. That forecast is generated by “fitting” a line to the historical data, and then projecting that “fit” out to the unknown future. This strategy is at the heart of forecasting, as I discussed a bit in a previous post (Hacking the Fed’s Forecast of GDP).
Here’s the gist: We do not know for sure what our Potential GDP really is—it’s an educated guess. And, the best source of information for guessing where we’ll be in 2031 is where we’ve been and where we are today.
Why is this concept, and estimate, important? As I mentioned above, Potential GDP is the “center” around which our ideas of “stabilizing” the macroeconomy revolves. This helps us understand at least a couple of things that you may have wondered about:
Why does the Federal Open Market Committee (FOMC)—the Fed’s policy committee—meet every six to seven weeks? Is it just to terrorize investors?
Investors may think it’s all about them, but like any one-sided relationship, investors care more about the Fed, than the Fed cares about them. Rather, it is because the Fed knows better than anyone that with Potential GDP they are working with a “moving target.” Or to use another cliché, they are steering a ship on a heading that is not certain. Hence, the FOMC meets frequently in case they need to quickly change course. I should add, too, the FOMC is not just using the CBO’s estimate of what the “center” is—they are using an arsenal of data and forecasts as noted in Hacking the Fed’s Forecast of GDP.
As a quick aside, the estimates of Potential GDP highlighted in the figure above imply the CBO expects the U.S. economy to average about 2.0 percent growth per year between now and 2031—which is only slightly higher than what the Fed expects for the “longer run.”
One more thing you may have wondered about:
Why does the FOMC feel the need to raise its interest rate “target,” or take other measures to slow the macroeconomy, when the clear implication of doing so will lead to job loss? Are they sadists?
No, they are far from sadists, I assure you. Rather, it is because Potential GDP is the concept underlying the “output gap”. A positive gap means actual, real-time GDP is above potential; a negative gap means we are below. And here’s the important part: A positive gap corresponds to rising inflation; a negative gap means falling inflation. With respect to the former, if the Fed thinks the gap is getting too wide in the positive direction, they attempt to close it by slowing down real GDP growth. In that sense, the economy is too far above Potential, and they view that as a bad thing for long run growth. Hence, some of the jobs and/or economic activity that are part of that widening output gap must be sacrificed.
Those are just two aspects of macroeconomic policy that Potential GDP helps us understand. There are many, many more things we can dive into. For example, the output gap is typically measured in terms of a deviation from a growth rate, as opposed to dollar values as shown in the figure above. And, as asserted in the previous paragraph, there is an intimate relationship between the rate of inflation and the output gap. I tackle that relationship in a future post.
This post, however, has gone on long enough, so I will end here. For those interested in the forecasting aspect of Potential GDP, in the postscript below I provide an example of how to generate a forecast like the CBO’s—that is, I try to mimic their estimate for Potential GDP as I tried to do for the Fed’s forecasts. This exercise can help one understand, at least a bit, what I meant by the statement I made above: that the best source of information for guessing where we’ll be in 2031 is where we’ve been and where we are today.
Postscript: Nerd Corner
Here I attempt to replicate the CBO’s estimate of Potential GDP. I do so as an exercise for anyone interested in learning more about forecasting.
To mimic the CBO’s estimate of Potential, I played around with estimating a “trend model”—which is probably the simplest forecasting model there is (it is the first model one would see in a forecasting class, typically). My trend forecast is a regression model with real GDP on the left-hand side of the equation and a trend variable on the right-hand side. You can extend that model to include a quadratic term or any number of polynomials you wish. Here I estimate a quadratic-trend model, where in addition to the trend variable on the right-hand side, I include that trend variable squared. Like any forecasting model, the general objective is to fit the historical series as well as possible, then use that “fit” to project forward (with some caveats; more on that in a bit).
After doing so, my fit and forecast looked as follows relative to actual GDP and to the CBO’s estimate of Potential GDP:
One can see that the trend model essentially “draws” a line through the average of the historical series. Including the quadratic term in the model allows the “fit” to match (more or less) the non-linear trajectory of GDP.
The CBO’s forecast is a little more sophisticated. As you can see it follows or matches more so than my model the undulations in the real GDP series over time. That is because the CBO is using a forecasting model with different inputs than the simple trend model I am estimating here. Nevertheless, notice that the forecasts from the trend model and the CBO’s model are not too far off from one another. Let’s hone in on the picture some:
You can see that for most of the forecast horizon (2022 through 2031) my fitted model predicts levels of GDP slightly higher than the CBO’s projections. By 2031, my model predicts GDP will be about half of a billion dollars higher than the CBO’s model.
Okay, so what? The results underscore an important aspect of forecasting. Any “baseline” forecast is ultimately a function of the history of a series. Yes, the CBO incorporates various assumptions about labor force productivity, capital accumulation and so on. But, ultimately any forecast is tied to the history of the series. After all, that is the best information on where we stand today as a macroeconomy.
Of course, “past performance does not guarantee future performance.” That is why you should think of the trajectories (from my model and the CBO model) as “baseline” forecasts. Then, it is up to you as the forecaster to figure out what possible deviations may be coming that will lead the economy away from that baseline. Such an effort will usually involve assigning probabilities on the likelihood of various events happening, among other assumptions you would have to make.
Why do I point this out? I suppose it’s to encourage a better understanding of “how the sausage is made,” when it comes to forecasting the macroeconomy. In the least, anyone out there without access to fancy statistics software can replicate what I’ve done here easily in Excel. First, use the “Insert” -> “Chart” options to plot a line (I am assuming you have downloaded a series into Excel, obviously). Then, right-click on that line and “Add trendline.” You’ll have the option to estimate a simple trend model or add a polynomial as I have done here (the “quadratic-trend” term), among other options. In other words, you can attempt to do your own “hacking” of others’ forecasts.
One caveat to this discussion is you should always be aware of “over-fitting.” That means I have configured a model to match the historical data so closely, that I have essentially built into my forecast random events that happened in the past. By definition, random events are not forecastable—so perhaps my fitted model is leading me astray as far as past performance being correlated with future performance. For anyone acquainted with machine learning concepts, “over-fitting” is a common concern. For those interested, you can do a deep dive into the topic here (and this author uses polynomial models, like I used here, to motivate his explanation).