Sometimes I have strong opinions about things like the importance of interest rates in various real estate and economic trends, which differ from the norm, and I will die on those hills.
This is more like a hunch. I have a skeptical reaction to reports about market conditions where projects “won’t pencil”. I’m not sure it’s a very dependable signal of market trends. Does it mean that costs are too high? Other suppliers are being too aggressive? Demand is low?
Perceptions are downstream of models
Before I get into it, I want to back up and take a 30,000 foot view of the issue. The basic problem is that our models determine how we describe the facts and how we interpret them. Frequently, conversations that feel empirical are really not empirical at all.
Let’s say there is a bad harvest. Ask a medieval priest or the president of the Farm Bureau what caused it. They will both engage in empirical conversations. Did we make the right sacrifices during the harvest festival? Were pagans spotted nearby potentially cursing us? Have we been engaged in sinful behavior? Was the fertilizer specified correctly for this year’s conditions? Is there a genetic adjustment that strikes the right balance between yields and resilience?
Those are each empirical conversations within their own worlds.
I feel this on almost all conversations about monetary policy. I follow the market monetarist approach. Monetary policy doesn’t have to be associated with interest rates at all. The relationship between interest rates and nominal economic activity is complicated. Centering interest rates in our conversations about the nominal economy is entirely a rhetorical choice. But everyone thinks every word they speak about the Fed and interest rates is an empirical statement.
Especially when it veers into territory like, “The yield on the 10 year note rose 0.25% this week, suggesting that the market expects the Fed to tighten the policy rate in future meetings.” That’s just stuffing reality into a rhetorical straightjacket.
Similarly, with builders, costs are the language they use. In the case of monetary policy, I would say that speaking in terms of interest rates adds confusion. But, for builders, costs is definitely the language they need to use. The cost language adds clarity within a specific operating context. But, outside that context, it gets muddy.
So, I think it leads to confusion when builders try to address issues at a macro level.
Demand spikes and sucks up all the local capacity for building. Interest rates increase. Supply chain disruptions slow construction. Demand collapses, leaving builders with unsellable inventory.
The micro-level descriptions for all of these contexts will be “New projects don’t pencil.” I have an annoyingly long earlier post about this for subscribers.1
So, when producers say that the problem is that costs are high or that projects don’t pencil, they are a few steps ahead of the medieval priest. Their facts have some causal connection in the physical realm. They are also a step ahead of the macroeconomic debaters. Their facts have clear causal effects on operations. Higher costs really do make it harder for projects to pencil.
But, they are still engaged in what they think is a purely empirical conversation that is actually purely rhetorical. It’s just pounding the square peg of an operator’s model into the variously shaped holes at different levels of economic reality.
What can we infer when projects don’t pencil?
Figure 1 might help consider this question. There are 2 different demand curves and 3 different supply curves in Figure 1, and I have highlighted 4 spots where different scenarios might settle.
In some scenarios, more homes will be started than in others. Scenario 1 is the standard, functional scenario we should hope to generally be in. It is a city with enough demand to induce some amount of new construction, with supply conditions that allow construction to grow at reasonable costs. That scenario describes most American cities, at most times, in the 20th century. And, whether a project pencils or not is almost entirely a product of demand expectations.
That is the world of “location, location, location”. At the level of the metropolitan area, there is some rate of construction growth that will maintain moderate rents. And, within that market, developers are performing the important economic task of accounting for all the idiosyncrasies of the local economy to figure out which units at what location that growth should take the shape of.
That is the typical paradox of development. Some of those decisions will be better than others, and the more accurately targeted projects will perform better than others. But, there are also elements outside the control of developers. Demand shocks can push potential growth up or down. That can change the proportion of projects that turn out not to perform well. And the greater the production of the other developers at any given future demand level, the more losers there may be.
That’s the normal, functional construction market. “Location, location, location” and “Everyone else is building too much.” Some projects won’t pencil, and they would pencil if the production of other developers was low enough to keep rent expectations higher.
Scenario 2 isn’t much different in practice. For scenario 1, think Atlanta in 2002. For scenario 2, think Cleveland in 2002.
Scenario 3 is Cleveland today (or Canton). There are supply chain problems that are pushing up the cost of inputs, but demand for new construction is pretty low in any case.
Scenario 4 is most cities today. In some of them, like Los Angeles, it doesn’t matter what happens to each individual cost. Some other cost will rise to keep supply constrained. Quantity is capped where it is by local policy. If inputs were cheaper and land was cheap, then the queue of developers would just grow longer until the de facto cost of construction was at whatever cost level demand would require to keep the quantity where it is.
In other cities, costs are high because of inputs. They are on the “High Cost Supply” curve. In this scenario, it is most clear, I think, to think in terms of the supply curve changing. If supply chains loosen up, the supply curve will shift down until the price equilibrium is matched with the new total number of units that can be completed.
You might say that costs are high on the “High Cost Supply” curve because of high interest rates. But, high interest rates could relate to high demand (either real or nominal), a shift toward investment seeking equity returns versus fixed returns, or a shift between different sectors that utilize different proportions of equity and debt, or any number of combinations of various changes of these sorts.
My point is, can “Jobs aren’t penciling” tell you anything about which of these scenarios you are sitting on and what sort of change to expect? Can it even tell you whether the trend will be to more or less construction, or higher or lower prices? I’m not sure it can tell us much at all.
How many projects aren’t penciling?
Finally, one additional point we might take from Figure 1 is, how many projects aren’t penciling? At all times, and in all places, there are an infinite number of projects that don’t pencil. Sitting here, right now, if I started to list all the projects that I personally wouldn’t be able or willing to complete, I would never stop compiling. It is infinite. And that’s just the projects that I won’t build.
There are always a fixed number of projects that will pencil and an infinite number that won’t. So, really, what we notice - what we think we are measuring - when we say “Projects aren’t penciling.” is “Projects that I think I should be able to complete aren’t penciling.”
It is plausible that moving from point 2 to point 1, the number of projects that builders consider but that don’t pencil will increase. In fact, it is plausible that the proportion of projects that builders consider but that don’t pencil will increase.
I think it is highly likely that moving from “Easy Supply” point 1 to “High Cost Supply” point 4 will be accompanied by a significant increase in projects that don’t pencil, because they are all projects that were expected to pencil under normal supply conditions.
A large number of projects that won’t pencil is frequently associated with expectations of declining starts. But, if the market is in condition 4, quantity will almost certainly increase.
In our current condition, quantity will almost certainly increase as capacity loosens up. Projects will pencil as the supply curve moves down. Costs, interest rates, and queuing for inputs or project approvals will all adjust to move toward the new capacity level.
I think expecting starts to decline from here because high input costs or high interest rates mean that projects don’t pencil is not the best analytical position to take.
There is good news in that post. It is from July 2024, and I noticed upon re-reading that I expected rent inflation to return to a level about 2% higher than general inflation as things settled out. But, rent inflation has continued to moderate and now appears to be running at less than 1% above general inflation. In the long run, we should all - producers and consumers - hope for moderate and stable rent inflation.
The lack of empiricism in the housing discourse lends itself to these sorts of logical errors. Micro phenomena like project level returns do a poor job explaining aggregate variables, but housing is stuffed full of micro analysis claiming to be macro
I like this. If interest rates were all that constrains builders and building, some financial guru could work up a multiplier for costs of materials, land, labor and loans that would match the actuals in a locale, and then the effect on demand in that locale. Is anyone doing that?