Arnold Kling  

Should We Forget the Production Function?

PRINT
Betty Yee's Outrageous Untruth... How I Fought Envy, Part 2...

That is the question that occurred to me after reading Tyler Cowen.


If the employers don't want you at the high wage, and don't want you at the low wage, what might your perceived MP be, temporarily or not?

MP means marginal product. It comes from a production function, where output is a function of labor input. Instead, think of Garett Jones workers, who build organizational capital.

Imagine you are the chairman of an economics department. When you hire, do you think in terms of a production function? Probably not. You think about whether you need senior faculty or junior faculty. You think about whether you want to build a capability in behavioral economics or add to your capability in labor market econometrics. How much does salary enter into your thinking?

If a less-than-ideal candidate will take the job for 90 percent of the salary of the ideal candidate, don't you still choose the ideal candidate? Probably if the differential gets as high as 50 percent, you consider the less-than-ideal candidate. But that is a pretty lumpy labor demand schedule.

Individual organizations do not possess production functions, in which output is a continuous function of labor input. Their labor demand is not a continuous function of the wage rate.

Perhaps you can argue that in the economy as a whole, all these individual lumpinesses wash out, and you get a nice smooth aggregate production function. But I would not necessarily count on that.

When I talk about complex, roundabout production and PSST (patterns of sustainable specialization and trade), I am trying to suggest ways of describing an economy without using the concept of a production function. Or, if you insist on a production function, I think you need to allow for large discontinuities in order to explain unemployment.


Comments and Sharing





COMMENTS (7 to date)
mdb writes:

one thing academics might miss is simply this, layoffs are a time to get rid of deadwood - without fear of lawsuit. Sure many good workers get laid off as well but they are not first on chopping block. If given a choice between a laid off worker for 10% less and an employed worker, there is less risk in the employed worker. If you hire a laid off worker at rate commensurate with the increased risk, they probably would not stay - even in this economy - so you lose all your training costs.

Les Cargill writes:

mdb: What makes anyone think that firms can successfully identify deadwood, especially given that Garret-Jones-organizational-capital-firms will be subject to internal public choice problems? Er, rent-seeking anyway. That's
my read on G. Jones, anyway - "organizational
capital" is a euphemism for rent-seeking against
the firm itself.

Firm formation right now is inadequate to even
support healthy levels of quits, much less real
competition. The knowledge problem applies most
emphatically to firms' internal governance, and
there's a lot of "too big to fail" out there.

Some of that deadwood isn't deadwood - it's "graduation".

Hugh writes:

Maybe at Foxconn, with its 450,000 workers, the idea of MP works quite well ("hire me another 10,000 assemblers for the iPhone V"!).

But I agree that in most organisations this is not really the approach, and a less linear model needs to be applied.

Kevin Dick writes:

The way I think about this is firms have two types of workers:

(1) Those that directly implement a production function

(2) Those that try to figure out what production functions to implement.

In a firm like Google or Apple (or a university department), (2) >> (1). In a restaurant or retail shop, (1) >> (2). At a startup (1) may actually be 0.

http://emergentfool.com/2011/01/09/thoughts-on-the-theory-of-the-firm/

http://emergentfool.com/2011/01/13/startups-small-businesses-and-large-companies/

http://emergentfool.com/2011/03/09/production-function-space-and-hiring/

Nathan Smith writes:

Yes, the economy is lumpy, and the neoclassical tradition abstracts from this. It is linked to specialization, from which the neoclassical tradition abstracts. In advanced economies, workers are highly specialized, to the point where no one else is qualified to do their particular job, even if that only means that no one else knows how the filing cabinets at Bubba Gump, Inc. are organized. Organizations often have n workers, and more than enough work for n workers, but not enough work for n+1 workers. In chapter 2 of my dissertation ("Complexity, Competition, and Growth") I show how "The Division of Labor is Limited by the Extent of the Market," using as a workhorse the non-neoclassical market model of Howitt and Clower (2000): http://www.sciencedirect.com/science/article/pii/S0167268199000876.

But I don't think there's any need to give up the concept of a production function. The lumpiness arises, not from the nature of production processes, so much as from the indivisibility of individuals. You can hire one, or two, or three workers, not 2.3, or 0.17.

Costard writes:

Fixed costs are anathema in a bad business climate, so the incentive will always be to improve productivity and use every resource - including labor - more efficiently. The emphasis changes to quality over quantity. So why not maintain a salary, if it allows the employer to "upgrade" to a more experienced or better-trained pool of employees, and avoid fixed costs related to additional hires? And if cutting salaries means a loss of talent to one's competitors?

It is entirely possible that an industry in this situation becomes saturated at any reasonable price, and that a set of specialists becomes (temporarily or otherwise) obsolete. In which case, the question is not merely how much potential employees are asking for, but with whom are they interviewing, and for what sort of job?

To put this another way: if food production on Mars doubles next year, Martians will almost certainly not eat twice as much, and it does not necessarily follow that they will pay less for food, either. They may choose to eat better, since they had already planned on spending X% of their income on food. In any event, there will certainly be food that they simply have no room for; and the process of finding some other use for this food, is one that takes time and price discovery.

Chris Koresko writes:

Arnold Kling: Individual organizations do not possess production functions, in which output is a continuous function of labor input. Their labor demand is not a continuous function of the wage rate.

Perhaps you can argue that in the economy as a whole, all these individual lumpinesses wash out, and you get a nice smooth aggregate production function. But I would not necessarily count on that.

My gut tells me you're right. If you average together a large number of independent production functions then you must get a smooth one. But there are factors in the production function that are not independent but highly correlated across firms. I'm thinking of regulatory thresholds, like the 50-employee limit above which ObamaCare kicks in.

Comments for this entry have been closed
Return to top