Looking-glass productivity

Over at the NBR (ungated), I get annoyed at the persistent Wellington tendency to get productivity backwards. Government's role isn't to run studies figuring out which firms are productive, which ones aren't, and to then try to arrange for the one to teach the other. Government's role should be figuring out and removing the barriers to entry and the barriers to competition that keep business away from the productivity frontier. 

A (lengthy) snippet:

Some folks take the wrong lesson from intermediate microeconomics – or never took the course in the first place. I worry that too many of them staff Wellington’s bureaus.
Every decent second-year university paper in microeconomics teaches students there are two equivalent ways of getting to efficient outcomes but one of them is much easier to implement.

Imagine an omniscient and benevolent central planner existed who knew our abilities, the productive capabilities of every firm in the market and exactly what each of us values. That benevolent planner could ensure efficient outcomes where no one could be made better off without making someone else worse off in the process. It is hard to do things that way because the knowledge the planner would need to make those decisions is impossible for the planner to obtain.

Fortunately, the welfare theorems in every intermediate microeconomics course show us that, if competitive markets are working well, we get there automatically. Prices coordinate individuals’ plans, so we wind up at the same kind of solution that the benevolent planner would have chosen. There is no need for a bureaucracy to figure out which firms are the most efficient at teaching the others how to work better as the discipline of markets takes care of it.

All of that left me rather irritated during the Productivity Commission’s “Productivity Hub” seminar last week on productivity in construction.

Housing New Zealand and BRANZ had commissioned work to review the literature on the productivity of housing construction before scoping future work on estimating the productivity of New Zealand’s housing construction sector.

It’s a big topic that matters. There have been conflicting studies on whether construction productivity has been stagnant or improving, and whether there is a long tail of low-productivity firms in the sector. Housing New Zealand is set to embark on a substantial bit of house building and will want to build the most houses possible on its budget.

Economic modelling

The work proposed using New Zealand’s administrative data to find the construction industry’s productivity frontier. Firms producing the greatest output value for their combined sets of inputs define the frontier in this kind of stochastic frontier analysis work; firms using the same amount of inputs to produce less overall value are inside the frontier and are considered less productive.

The seminar focused on important technical issues like the merits of Cobb-Douglas as compared to translog functional forms and the merits of stochastic frontier analysis over conventional linear regression techniques in figuring out which firms are most productive.

But I was getting itchier and itchier over the course of the seminar – as were a few others. It was not the researcher’s fault at all – she proposed taking the right approach to the question as defined. The problem rather was the framing of the question and its policy consequence.

Trabant efficiency

Imagine going back in time to 1982. You’ve been hired by the East German government to examine the productivity of the country’s automobile construction sector. You’ve been asked to use the latest statistical techniques to figure out which of Trabant’s suppliers are working efficiently and which are not. The government wishes to do this so it can send experts into the efficient plants to figure out what they are doing, so they can help the less efficient suppliers to get up to speed.

Stochastic frontier analysis was not really available then but that was hardly the main problem. Even if they got every one of the Trabant plants, and their suppliers and their suppliers’ suppliers running as efficiently as the most efficient East German plant, they would have mis-specified the problem. They would still have a Trabant at the end of the production line, and the value of a Trabant was less than the value of all of the bits of plastic and metal and rubber and labour that went into producing those things – and it certainly ran less well than the Volkswagens produced across the border in West Germany.

Instead of commissioning a study that might figure out which plants are efficient, tearing down the Berlin Wall and allowing free trade would let markets figure it out. Efficient plants might survive; inefficient ones would put their machines and workers to better use elsewhere.

New Zealand reached a similar conclusion about its local automotive industry during the reforms of the 1980s and 1990s. Economist Steve Landsburg talks about the Iowa car crop. Growing corn, selling it to Japan, and getting cars back, is a more efficient way of building cars than a lot of ways America has tried. New Zealand’s paddocks, forests and office towers provide our car crop more efficiently than did the Petone plant.

The construction productivity work specified the problem back to front but was hardly the first to do so. An academic study might point out which firms seem more productive than other firms but it can be easy to get those things wrong – especially when New Zealand’s construction firms have operated within a system designed to stymie construction productivity.

Regulations nest

Zoning has never allowed building to any reasonable scale, so the industry developed for smaller scale developments and bespoke projects. A broken land use planning system has given us a construction industry that may be relatively efficient within that system but it is not the one needed for the scale of the task at hand.

A nest of regulations and perverse incentives in council consenting makes it difficult to use far less costly but higher-quality construction materials from trustworthy places like Vancouver or Seattle or Tokyo. Costs here are consequently much higher; productivity is the ratio of the value of the outputs to the value of the inputs. Building houses costs (very roughly) twice as much here as it does in Texas, leaving land costs aside.

And competition from foreign suppliers is hampered not only by the Overseas Investment Act but also by the morass of local regulation that local firms have learned to navigate but foreign firms have not.

Seeing the productivity problem as one of finding the most efficient firms and cajoling the others to behave more like those Stakhanovites  (Soviet Union workers who took pride in their ability to produce more than was required, by working harder and more efficiently)gets the problem back to front. It is trying to work the welfare theorem from intermediate microeconomics the wrong way around.

I think the NBR's editors added the definition of Stakhanovite. I didn't know that was no longer a term I could assume commonly known. Alas.