Should we have calibrated the American Rescue Plan to ensure that throughput would stay within the physical capacity of ports? If that seems like a bizarre question to you, it’s worth asking, as our contributor Hassan Khan did:
so if I understand The Discourse correctly: if the Port of LA/LB had more capacity and auto manufacturers hadn’t canceled all their chip orders prematurely at the start of the pandemic, there’d be no complaints about the stimulus. Is that right?— hk (@hassankhan) October 18, 2021
While we all agree that constraints to economic capacity exist, some sectors can have substantial spare capacity while others have none. The question “capacity constraints macroeconomic policy should optimize for” can’t be answered by the standard measures of “potential output” that dominated policy debates in February.
If we are going to take the supply side seriously - something that may have to happen to support the kind of demand-side policy required for maximum employment - we need to get serious about measuring it. Individual shortages and bottlenecks can have highly nonlinear effects on aggregate output. Granular measures are equally critical for crafting policy responses to the challenge of binding capacity constraints, since economic growth involves both encountering and overcoming capacity constraints.
In February, as discussion over the size of the American Rescue Plan was raging, we published a piece outlining how little conventional estimates of “potential output” contribute to an understanding of the real limits faced by a growing economy. Some at the time worried that already-enacted stimulus payments had put us on a trajectory above that implied by potential output measures, and that any further stimulus would burn off into inflation without aiding the recovery. These worries have since grown, putting a damper on attempts to pass an infrastructure bill proportional to the economy’s needs.
However, many of the present inflationary drivers provide further evidence for our original argument that an aggregate measure of “potential output” explains almost nothing unless it can be broken down into inter-linked sectoral measures. We argued that a highly aggregated measure based on the trend of realized GDP over time, or the distance between current unemployment and the “natural rate of unemployment,” would explain little about where the economy would hit temporary limits as it expanded.
Since then, the growing economy has hit a number of snags, especially in terms of port capacity and automobile production. Ships have backed up so far outside the port of LA that one of them snagged a pipeline, and President Biden stepped in to negotiate an agreement whereby the port would run 24/7 in order to clear the backlog. Automobile production has collapsed despite having ample spare capacity in a strong demand environment. The culprit? Shortages for niche low-tech microchips further upstream in the automobile supply chain.
This opens out onto some bigger questions. What are the relevant economic constraints? Where do they bind? Which constraints matter and which can we ignore? How much is it worth to fix a particular constraint?
“Potential output” estimates largely avoid most of these questions and effectively extrapolate along a trendline. The model can’t identify dynamics that we see every day in the news. As we noted in our original piece, establishing actual measurements of capacity requires digging into input-output relationships. But even with these in hand, there remains the question of how best to respond to the presence of identifiable constraints.
Using an aggregate measure locks us into a decision to give up at the first constraint, even though the best option from the perspective of economic development may be to work through those early constraints. While this does not require going all the way to wartime mobilization, it does show how many more degrees of freedom there are in a model of capacity that centers actual input-output relationships over mere trends in aggregate data.
As we explained in our piece “Beyond the Phillips Curve: A Dynamic Approach to Communicating Estimates of Full Employment,” different sectors have different timelines for adjusting capacity to meet changing demand or structural bottlenecks. In some sectors like construction or trucking, capacity may be limited by labor shortages. In others, like automobile production or the sale of consumer durables, it may be limited by the volume of goods attempting to travel along a particular supply chain. The key point is that alleviating different bottlenecks will have different timelines and different requirements in terms of real resources and labor for different industries. The longer it takes to add capacity, the thornier the political and economic challenge of managing existing inventory and capacity.
We also have to think seriously about which bottlenecks are most important. A shortage of aluminum might simply be resolved by investing in restarting an existing plant, as Alcoa is doing in Brazil, but it could also prove more challenging if key inputs, like electricity, are bottlenecked, as is the case for a plant in the Netherlands, where a shortage of natural gas shortage has made electricity less available. For bottlenecks where it is longer and harder to procure more of what you need and requires substantial investment or reorganization, ugly choices around rationing and price increases are more likely to emerge.
For example, oil shortages became apparent in the late 1960s but it typically took over a decade to move from initial investment to increases in final production. Spending time and resources to resolve bottlenecks does have a political cost, but today’s shortages - especially those caused by disruptions to shipping - will likely be sorted out by the end of the holiday season.
If we understand these issues purely through the lens of “potential output,” we abandon clear opportunities for economic growth and tighter labor markets. If output is constrained by bottlenecks, models that rely on estimates of “potential output” can only revise the maximum level of output downward, while not telling us anything about how to achieve a better trajectory. Doing this requires investigating where and why growth is being held back, and addressing the issues there.
If we instead think of these bottlenecks as plastic and subject to alleviation through targeted investment - whether by the government or private industry - the situation is substantially different. We face a menu of choices as to the likely costs and benefits of addressing specific bottlenecks so as to allow the rest of the economy to grow to a size that it reaches its own constraints. If we value tight labor markets and a fair deal for working people, this is the strategy we must take: widening the bottleneck is preferable to rationing demand. If we decide not to widen bottlenecks as we hit them, we forgo longer run gains that flow from learning how to expand capacity through increased investment in physical capital, a better-trained workforce, and better system processes.
Sector-specific inflationary snags and difficulties are indications of where investment must be made, not an indication that the economy should be shrunk to a small enough size that it can fit through the narrowest bottleneck it currently faces.