When we talk about procurement or support costs in the 2040 decade, obviously that’s a long way from now. Any sort of projection of a cost like that is going to be very heavily influenced by seemingly subtle decisions about inflation indexes…
It is an empirical reality that the cost of – I’ll call it industrial inputs, by that I mean the labor in particular – the cost of industrial inputs has risen at a faster rate than societal wide inflation for a number of years. And that’s what the CBO builds into it’s projection, and that’s what the Navy puts in for at least some of it’s current year estimates and there is some inconsistency in how it is extrapolated.
But what is not knowable – we wish we knew it – is that going to continue? In an equilibrium sense it can’t possibly continue. We can’t possibly have industrial workers, their relative income, always rising relative to every other sector of the economy. In some sense it makes sense that this would return to societal inflation rates. And if that’s so, then it would lead to lower outyear cost.
By the same token we haven’t seen any evidence of these costs growing at societal inflation rates as opposed to the elevated rates that they’ve been at. That difference, which is almost a philosophical or theological difference, is under-girding outyear cost estimates. We don’t really understand what the world is going to look like in 2045.
That was the insightful Edward Keating on the Acquisition Talk podcast.
The increases in the cost of labor, material, and capital used in production does indeed seem to be a major philosophical question. These price increases are generally passed on to into the cost estimates for weapon systems. So, for example, if Navy shipbuilding input costs grow at 3 percent above inflation, then usually the next ship (if it is assumed to use the same amount of labor, materials, etc.) will also grow in price by the same 3 percent. Of course that’s a simplification due to learning/rate effects putting downward pressure on the estimate, as well as specification changes putting upward pressure.
However, it would seem that input prices and output prices shouldn’t be correlated. Productivity, of course, leads to higher output for the same, or even lower, input. Wages for computer scientists have gone up dramatically, but we derive far more value from software systems at today’s prices.
Alex Tabarrok chalks up most of the story to the Baumol Effect, or the idea that because we get so much productivity in the world of “goods” like computing, that the prices of goods will fall relative to the prices of “services,” like a four string quartet, or healthcare or education.
This explanation doesn’t sit well with me. Certainly we imagine that if we avoid a calamity in the next couple centuries, our technology in healthcare will be far in advance of our own today, curing many ailments. If we can imagine that technology in the future, then it is plausible to think about different cultural or institutional arrangements which could accelerate the development of that technology. And if we could select the most innovative arrangement, that would have the effect of making the healthcare “service” look more like a fast improving “good” with decreasing quality-adjusted prices.
But wouldn’t such a world also have even faster growing productivity in other sectors, always making healthcare display relative price growth?
One major point about how Edward described industrial worker compensation: He said it was always outpacing societal inflation and that the pattern couldn’t persist. Usually, however, we think about wage growth above inflation as return to productivity. The Employment Cost Index (ECI) has long outpaced the GDP Price Index.
But should contractor labor rates always be growing faster than the ECI? I can attest to the fact that major defense contractor rates have grown faster than the ECI for many years. Like Edward said, this cannot got on forever… or can it?
The trend in the fully-burdened labor rate the government pays a contractor for an hour of labor really can grow much faster than the ECI indefinitely. What cannot outpace the ECI for long periods is the direct labor rate which covers take-home pay. And indeed, in my analysis of 1921-3 cost reports, defense contractor direct labor rates have grown at about the ECI, perhaps a bit faster. That may reflect differences in the composition of the workforce, which in defense is getting older and perhaps more highly skilled than the rest of the economy.
Where we see the massive growth in contractor labor rates is in the fringe rate (retirement, medical, other benefits), in the overhead rate (managers, facilities, capital) and in the general & administrative rate (corporate, IRAD, B&P). The dollar cost of each of these rate categories has grown perhaps 5 to 9 percent per year over and above inflation.
Whereas direct labor costs are integrated into the market for labor which keeps prices in-line (and gives the administrative contract officers in government information on where wages should be), indirect costs are unhinged from the market. Indirect costs can be whatever the government is willing to pay. Often times, they are kept cosmetically lower by such tactics as shifting labor from overhead to direct (e.g., the foreman), thus increasing the base and decreasing the expense.
This gimmick, one of many, makes it appear like overhead rates are in control (just 100 or 200 percent of direct labor costs). It conceals the accounting changes. Whereas overhead in today’s tech firms have risen and indeed become basically the entirety of the operation (due to intangibles like software, data, etc.), defense contractors have gone the other way. Whereas 50 to 60 percent of a defense factory’s labor was tagged as direct labor in the 1970s, that figure today is closer to 90 or 95 percent. It’s not that more of today’s defense work is actually direct, but the accounting systems are sophisticated enough to track it and the government incentivizes it. In short, it is because government officials cannot gauge the value of the objects they intend to buy.
I am concerned with this divergence between contractor accounting and the realities of a technology business. It’s not just in treating defense labor as though the bulk of time was riveting or welding when in fact they are doing knowledge work. The outsourcing costs, which for primes has risen to something like 70 percent of total costs, are not really physical items like steel. As Edward explained:
It turns out for any military ship, the steel itself is really a very small percentage of the entire ship. It’s less than 5 percent. The way to think about a military ship frankly is that it’s a very large computer system that happens to be on water and has steel that surrounds it. But the steel input itself is really a very small percentage of the total cost of any modern ship.
That’s an excellent way of describing modern weapon systems. They are like a computer, where most of the cost isn’t really in the physical inputs like silicon — which is after all just sand — but it is in the conceptual design.
Leave a Reply