Problems with the TFP concept, from my new paper with Ben Southwood

Here is another excerpt from “Is the rate of scientific progress slowing down?”:

First, many scientific advances work through enabling a greater supply of labor, capital, and land, and those advances will be undervalued by a TFP metric. Let’s say someone invents a useful painkiller, and that makes it easier for many people to show up to work and be productive. Output will rise, yet that advance will show up as an increase in labor supply, rather than as an increase in technology or scientific knowledge. Similarly, a new method for discovering oil may boost output, but that will be classified as an increase in oil supply, even though it does properly represent a form of scientific progress. TFP is best at measuring scientific and technological advances that are superimposed on top of an existing supply of the other factors of production. If, for instance, you imagine a series of capital and labor resources at a factory, and someone develops a new formula for combining those resources more effectively, this will be picked up very effectively by standard TFP measures.

The more general problem is that many scientific and technological advances are embodied in concrete capital goods. Again, the TFP measure does best when the supply and nature of capital is fixed, and a new idea makes that capital (and associated labor) more effective. But what happens when the new idea is itself embodied in a concrete capital good? If a hospital equips its surgeons with iPads, or with Augmented Reality glasses, to make them more effective in the operating room, as a first order effect that measures as an increase of capital expenditures by the hospitals rather than as an innovation. Health and later output will increase, but will we really know if it is due to better ideas or just more investment? It will appear to measure as new investment. Capital expenditures and TFP are not so easily separated, whether at the conceptual or the practical measurement level.

Similarly, separating TFP from labor expenditure is not always so simple either. If a worker generates and carries forward a new scientific idea for producing more with a given amount of labor, that measures the same way as the worker being taught greater conscientiousness and producing more. Yet the former is (ideally) TFP and the latter is not, but both will count as an increase in the quality of labor input in the same way.

Here is my original post on the paper.  Here is the paper.

Comments

Attempts to create greener technology also face this problem. The externalities imposed by carbon dioxide are unpriced and therefore not subtracted from GDP. If a new technology allows same level of energy production with lower CO2 emissions, welfare rises, but there is no effect on GDP and therefore no effect on measured productivity. Since a good deal of energy (pun intended) has been devoted in recent years to producing green technology, this may be contributing substantially to the mismeasurement issue.

Capital flows, a bit constricted, but flows nonetheless.
Listen to Peter Thiel, it is all about rescaling. Keeping the economic complexity manageable. Driven by two things: 1) Known uncertainty and 2) Necessity of minimizing duplications.

The effect is to generate scale with stable queues in input everywhere.

I understand the capital argument, but not the labor one. Doesn't the second labor scenario result in less leisure time while the first leaves leisure unchanged? Inducing idle people to contribute work-hours into the labor supply seems fundamentally different than productivity. Perhaps the distinction is cultural productivity versus technological productivity.

TC has confused science with technology, again. Question: was the marketing of radium health tonics a scientific advance? It must have increased employment (in the short run). How about, oh say thalidomide as a morning-sickness medication. Progress? (Note, in case you live in a cave, that thalidomide is a WHO Essential Medicine). Any attempt to conflate scientific "progress" with societal or economic "progress" is doomed, because science simply doesn't value worth and happiness as goals. Is learning of the heat death of the universe progress? The Big Stretch? The Big Bang? Although, as a scientist, I do like the idea of requiring all economic progress to be scientific. Let's toss out any Economics that doesn't have randomized, double-blind, replicated data behind it. Anything that doesn't obviously can't be "progress".

Issue with TFP or with using TFP as measure of technological progress?

This sounds Lachmann-esque and I am looking forward to reading once granted access.

the scientific rate is in-fact slowing down. I tire to explain my idea of growth but the fact remains science is incorrigible. Yeah, it's a necessary condition of one to find his prevailing wisdom slowing down.. But who am I to say? I'm not anyone to say but I am someone that can say that the conditional clause isn't the future clause.

These issues are not probably important when we measure change in TFP which is a more relevant metric

Some think that the loss of manufacturing has hollowed out the economy making innovation harder to find. They say that's the real problem.

"If a worker generates and carries forward a new scientific idea for producing more with a given amount of labor, that measures the same way as the worker being taught greater conscientiousness and producing more"

Jesus Chr _ st, have you ever had a real job where you, like, made something?

- 'new scientific idea': you are trying to say a factory worker will adopt a new manufacturing PROCESS, which will often embody a new technology, or at least some improvement in engineering. Not some 'science breakthrough'.

- "the worker being taught greater conscientiousness and producing more" See my point above. This is just gibberish.

The oil discovery example does not work unless oil is one of the "factors" in the PF that TFP is the residual of. Scale economies will also show up as TFP.

"Let’s say someone invents a useful painkiller, and that makes it easier for many people to show up to work and be productive. Output will rise, yet that advance will show up as an increase in labor supply, rather than as an increase in technology or scientific knowledge. "
> Only if it's showing up, *and* this showing-up is measured continuously (e.g., on the books, if firm employs a full-time worker, then they have more or less sick days; a technology that makes them have fewer sick days will not change the reported labor input but make it more productive). Not if it makes them more productive. To the extent that laborers become more productive, this will be TFP.

"Similarly, a new method for discovering oil may boost output, but that will be classified as an increase in oil supply, even though it does properly represent a form of scientific progress."
> That's not right. Suppose the discovery means a given oil rig can find oil twice as often. Then it takes 1/2 the capital to generate the same intermediate input. This will turn up as TFP.

Thanks

Stupid question:
Given the conflation of capital and technology (those you listed, such that tech may be manifest as capital investment, as well as the converse, that capital investment in tech bootstraps further technology -- most salient in the development of machining); what's wrong with just using real GDP/capita as an indicator of technological progress?

This indicator will reflect:
- individual worker productivity
- labor supply per capita (reflecting changes in healthcare, human capital, institutions enabling greater labor force participation -- I think it's desirable to call improvements in all of those "technology" of one sort or another)
- more efficient capital allocation (either better technology to buy, or better mechanisms for allocating the capital)

The one downside is real GDP/capita also reflects more capital supply. This may not be that much of a problem:
- When capital buys plant, it is buying technology. Presumably capital supply increases with better tech to invest in. So this seems like technological progress.
- Might be thornier for cases when capital buys land, especially when the price of real property rises faster than inflation for reasons other than the value that might be extracted from the land is increasing due to innovative uses.

Did not RTFA. real GDP/capita the first metric used.

I see this argument for labor, but actually less so for capital.

"But what happens when the new idea is itself embodied in a concrete capital good? If a hospital equips its surgeons with iPads, or with Augmented Reality glasses, to make them more effective in the operating room, as a first order effect that measures as an increase of capital expenditures by the hospitals rather than as an innovation."

Say a hospital in the past bought pagers for nurses for $200. They worked fine, but the upgrading to a $200 ipad worked better. Having been replaced, the pagers have a depreciation rate of essentially 100% and are discarded. In this example, the "size" capital stock is essentially unchanged, but TFP has increased.

Comments for this post are closed