Arnold Kling  

Robin Hanson on Market Failure

Historical Counter-Factuals an... Mortgage Loans with Low and Hi...

At Cato Unbound, He writes,

Finally, consider next the many functions and roles of managers, both public and private. By being personally impressive, and by being identified with attractive philosophical positions, leaders can inspire people to work for and affiliate with their organizations. Such support can be threatened by clear tracking of leader forecasts, if that questions leader impressiveness.

If accurate forecasts can help business function more effectively, then any systematic tendency to lower the quality of forecasts should put a firm at a competitive disadvantage. Firms that impede forecasting should be driven out of business by firms that take it more seriously.

Implicit in Robin's view is a theory of market failure. The market fails to do its job of winnowing out ineffective practices.

In many recent posts, I have discussed differences between profits and non-profits. The difference that I keep coming back to is that competitive, profit-seeking firms have to focus on results. Non-profits instead have the luxury to focus on making donors feel good, without having to worry so much about results. But when it comes forecasting, Robin does not seem to think that profit-seeking firms are compelled to focus on results.

Comments and Sharing

COMMENTS (17 to date)
J Storrs Hall writes:

It's just an impression, but it surely seems like there are a lot more non-profits in the forecasting business.

david writes:

But there are no firms that can avoid employing human managers at some level (we're not quite yet at the singularity...). So if all human managers are threatened by accurate forecasts, all firms behave as such, too, even if subject to competitive behavior.

Competition only drives toward optimality, not impossibility.

jc writes:

What's more important to results?

(A) Somewhat more accurate forecasts than competitors, or (B) a cadre of workers that are inspired and have internalized the organization as a tribe worth fighting for?

Perhaps markets sometime favor the latter?

Grant Gould writes:

I don't think you need to hypothesize a market failure to accept Hanson's view. All he is saying is that the price of an accurate forecast -- reduced social status of managers and planners -- is greater than the benefit -- more accurate plans. It's only a market failure if there are managers and planners available who neither value status nor rely on it to obtain workplace compliance.

ChrisA writes:

Having been convinced by Robin's argument in the past to try using betting markets to improve forecasting in my company, my experience was that the cost (in management time and so on) of improving the forecast turned out not to be worthwhile and we dropped the idea.

In most cases, for instance oil markets, it turns out that where there is value to a forecast, there is already a market based forecast (the Nymex forward curve for instance). This is probably not a coincidence, if the decision is important to one firm it is probably important to many people and a market should arise to allow hedging.

Another example, a publicly traded firm also gets some pretty good feedback from the market place everyday about the impact of major decisions that affect the organisation. You don't need to set up a much less liquid internal market to replicate this feedback.

ziel writes:

It's because accurate forecasting is actually impossible. The best a model can do is provide plausible impacts under various scenarios - but no one can actually make a model powerful enough to predict real-world outcomes.

rafael writes:

rather than thinking in terms of "non profit" vs. "for profit" it might be more helpful to thing in terms of "fee for service/good" vs. "paid for by others". There are some fee for service non-profits who act like for-profits in large part because they have to provide value to customers who pay for the thing offered.

Robin Hanson writes:

What david, jc, and Grant said - there is a difference between a globally inefficient equilibrium and a locally inefficient action. A signaling equilibrium where managers must signal impressiveness might not be globally efficient, but each person might not be able to do better doing anything else than they are doing.

rpl writes:
The difference that I keep coming back to is that competitive, profit-seeking firms have to focus on results. Non-profits instead have the luxury to focus on making donors feel good, without having to worry so much about results.
This is nonsense. For-profit and non-profit organizations both have to focus on the same thing: getting people to hand over their money. How you do this depends on what it is you purport to offer in exchange for that money. In the case of charities, making donors feel good isn't a "luxury"; it's what entices the donors to donate in the first place. In some sense, feeling good is what they are selling. Actually providing it is "focusing on results."

Arnold, I think what is going on here is that you are somehow just now discovering that a firm's real customer is always the guy footing the bill. "Results" in a business context always means making that person feel like he got something valuable in exchange for his money. For-profit or non-profit status does not change this fact in any way.

ChrisA writes:


How does this work for a commodity business? Why does a company buying (say) bauxite care about how impressive the guy selling them the stuff is? They care about price and grade (I know from experience). So Arnold's point is very true for me, improving forecasts, at least for commodity companies, can't be an important driver of success otherwise the ones that do forecast well would drive out of business the ones that don't. I suspect this operates for most profit businesses as well, not just commodities. Again, my view, if the forecast is important there probably is already a forecasting market, if it is not important, then by definition, it is not important (profitable)to have a good forecast.

R Richard Schweitzer writes:

Being closer to 90 than to 80, I realize my views may be archaic, still: How is this subject one of Market Failure?

The function of markets (presumably)is not forecasting, which is "simply" the manner in which some participate in a market.

The function of a market is the distribution of goods, services, and results of productive activities. The actions of participants in determining the manner or extent of involvement (planting more or less corn, wheat, soy etc) is not a Failure of the market system.

Bill writes:
The difference that I keep coming back to is that competitive, profit-seeking firms have to focus on results. Non-profits instead have the luxury to focus on making donors feel good, without having to worry so much about results.

Arnold, as someone who is a Finanacial Director by day, volunteer leader by night, I think you have a rather naive view of non-profits. The non-profit arena is very competitive, with many causes chasing limited funds. Funders, foudations and even the average Joe who pay a $35 annual membership all expect to get their money's worth. I think sometimes, planning is even more difficult in the non-profit, as I can lose 50-75% of my revenue in any given year. Try planning a multi-year project under those conditions. Forecasting and developing contigencies are very important, especially when you're not rally allowed to hoard cash.

Pandaemoni writes:

In my personal experience with investment bankers, they mocked their own in-house economists ineffectiveness. More than one agreed that the only reason the bank needed economists at all was for marketing and public relations purposes. Not that their forecasts were "wrong" per se, but that their forecasts were usually minor variations on the already received wisdom.

Any economist who has radically different forecasts from that which the market generally expects, won't be the favorite of the rest of the people at the bank. That might change over time if the predictions proved to be consistently accurate, but that's not the real world experience.

Still, no one wants to think an i-bank does no forecasting at all, so economists are needed to make potential clients feel better.

Mark Seecof writes:

Prof. Kling,

Does your recent contemplation of these matters change your views of charitable tax deductions at all? Since, as you say, non-profits seem to be run more to please their donors than their putative beneficiaries, why should they get tax breaks which ordinary businesses do not?

To rehash an argument I offered before, it's not clear to me why a Christian Science Reading Room should be tax-exempt but not a Barnes & Noble.

Costard writes:

Human beings are not a fungible commodity.

There are - in that strange place, "the real world" - managers who are good at their job; and leaders who are not mere Hanson-esque caricatures of leadership. For whom there is every incentive to be held accountable, because they measure up quite well.

Which is precisely why you have competitions, contests and fairs. In addition to the fear of losing, there is also the hope of winning; and it is human nature to think better of one's abilities than one should.

jc writes:

Fwiw, in case anyone is interested...

Here are some basic results from two leadership meta-analyses, the first on within firm outcomes (and leadership antecedents) and the second on firm performance outcomes. (Some findings seem high to me, but hey, this is not my area.)

From Dirks & Ferrin (2002)

Mean Weighted Correlation (corrected) b/w Trust in Leaders and the following outcomes:

- Job Performance: .17 (k=21, N=5686)
- Organizational Commitment: .59 (k=40, N=9676)
- Various Org. Citizenship Behaviors (5): b/w .13 & .25
- Job Satisfaction: .65 (k=34, N=10631)

Mean Weighted Correlation (corrected) b/w Trust in Leaders and the following antecedents:

- Transformational Leadership: .79 (k=13. N=5657)
- Transactional Leadership: .67 (k=9, N=3624)
- Distributive Justice: .58 (k=15, N=3562)
- Procedural Justice: .68 (k=30, N=5972)
- Breach of Expectations: -.43 (k=5, N=1391)
- Participative Decision Making: .52 (k=7, N=1273)

From Nair (2006)

Various Study Correlations (corrected) b/w Leadership and:

- Financial Performance: .18, .29, .36
- Operational Performance: .11, .35, .44, .63
- Customer Service: .24, .43, .45, .57
- Product Quality: .27, .45, .47, .73

Jacob Oost writes:

My problem with Hanson's article is that it seems to be based heavily on assumptions and anecdotes, and the entire premise could in fact be nullified if one were to discover that forecast accuracy-tracking of the kind he talks about is actually quite popular in certain countries, or parts of the country, or in certain industries. *Then* the question would not be one of market failure, but rather "why is this group of people interested in forecasting accuracy but not this group?"

Another thing I have to say is that I do not immediately see the importance of forecasting needing to be all that accurate, so long as the corrective mechanisms of profit and loss are in place to guide resources to where they are most valued. Whether through forecasting or through trial and error the results ought to be the same over time.

Also, don't forget the Hayekian reasons forecasting could fail just like central planning.

Last thought: organizatons and businesses *do* pay for accurate forecasting. My linear algebra teacher was an industrial engineer or some such and he worked for a hospital modeling wait times for patients and doctors and room occupancy for one of the state's largest health organizations. He showed us some of the work he did, it was all software based and their models were updated continuously based on real-world data from the hospitals themselves so as to be more accurate.

Comments for this entry have been closed
Return to top