Thursday, October 17, 2013

Limits to Central Planning, Paralysis

This essay is based on a couple of conversations I had with Mark Nissen many years ago.

All errors are mine and Mark cannot be held responsible for any of my errors.


Imagine the purchasing department for a huge organization.  It buys huge quantities of toilet paper, paint, socks, electrical wall name it.

The dollar amount of nearly every purchase is over one million dollars.  Further, imagine that the organization's process is to have 4 signatures on all purchase requests over one million dollars.  Each signature represented a level in the org chart.
  • A grunt would prepare the purchase request. 
  • A first level supervisor would review and approve if acceptable
  • A second level supervisor would provide oversight, review both the purchase request and the performance of the first level supervisor
  • Ditto for third level supervisor
  • Ditto for the fourth level supervisor who had +$1M approval authority

The problem was that the purchasing department could not buy items fast enough.  You think you have stress when you run out of your kid's favorite cereal, try running out of toilet paper on a cruise ship in the middle of the Indian Ocean.

The initial analysis of the issue quickly homed in on the number of signatures required.  Think of each supervisor as a filter.  If each filter bounces 30% of the purchase requests, then the process efficiency is 0.7*0.7*0.7*0.7 or approximately 25%.  Each time the grunt launches a request into the approval system, there is only a 25% chance that it will get through without it being bounced back.

Worse yet, the bounced paperwork stacked up in the grunt's in-basket.  The stack grew larger.  Each piece of paperwork grew more stale.  And as the paperwork grew stale the number of details that might need refreshing also grew.

Organizational resistance

The organization reviewed the assessment and initially rejected it.

A few of the grunts and first level supervisors thought it had the look, feel, smell and taste of the truth.  So they quietly started collecting data.  They quickly determined that the situation was worse than originally assumed.  Most "approvers" auto-calibrated to a 50% approval rating.  That meant that only one purchase request in sixteen made it out the end of the pipe-line and was let for bid.

The auto-calibration at 50% should not be a surprise.  Tversky, Kahneman and others demonstrated that humans are reasonably well calibrated at 50%.  Get much off top-dead-center and our human brain (a computer with very limited digital processing capability) starts to transmit highly erroneous conclusions.

Posed another way:  An approver is expendable if they approve everything.  They justify their existence by bouncing proposals.  That is how they "add value".  They prevent risky actions by vetoing them.  Showing too high of an approval rate and you might get repurposed into a job with more stress and more job content.

The solution

The solution, in this case, was to define certain (most) types of buys as commodities.  Toilet paper and paint are low risk buys as long as they conform to established commercial standards.  Guard rails were put in place where large price jumps or no-bids might trigger higher level oversight.

Motivation (sidebar)

The most popular reprint in the history of Harvard Business Review was written by Frederick Herzburg (in 1969) and it involved motivation.  One of Herzburg's examples involves a department that answered customer complaint letters.  Clerks wrote the responses and supervisors edited them with their red pens.

Productivity was very low.

The solution, according to Herzburg, was to make the clerk's responsible for their own quality.  The (a) supervisor might audit the clerk's quality on a random basis.  The thinking was that a clerk who was incapable of writing an acceptable letter would be moved out of that department.

Productivity and moral soared.

Allocative Efficiency

Allocative Efficiency is one of the hallmarks of Capitalism.

Allocative Inefficiency is one of the hallmarks of centralized systems.

There are countless stories of collective farms in the U.S.S.R. that had oil filters but no engine oil for their tractors.  Other farms, 60km away might have oil but no oil filters.  Another farm might lack air filters but have plenty of oil and oil filters.

Planting season in northern climes like the USSR is short and there is little time to waste.

The tragedy is that the miss-shipments could have been easily remedied.  All it would take is a small truck, 20 gallons of fuel and "a market".

"A Market" is a venue where information regarding relative needs and surplusses can be hashed out.

Central Planning abhors markets because it threatens their existence.

No comments:

Post a Comment

Readers who are willing to comment make this a better blog. Civil dialog is a valuable thing.