Friday, September 12, 2014

Economists Are So Cute (but so are badgers )

Economists would be so cute in their naiveté if they weren't so destructive. The other day I was reading an article (oddly, on umpires and the strike zone) and I ran across this quote "As economist Edward Lazear has shown, organizations become more productive when a job well done is rewarded with extra money and dumb mistakes are punished." The referenced article is "Incentive Contracts" http://www.nber.org/papers/w1917. The paper is from 1986. Normally I would not bother with something so old, but it did get a reference from the recent article.

Actually, Lazear does not show much of anything, but he does survey various incentive schemes. There are a variety of equations, including a partial differential equation based on nonsense, that are not used for anything other than to give the article a patina of respectability.

Here is my summary. Workers are lazy and in order to get them to do anything we have to provide proper incentives. The goal of the incentives is to increase worker effort. We can base incentives on effort or on productivity. Oftentimes it is difficult to get information on effort and productivity, so we may have to adjust labor contracts for the best available information. Oh, by the way, we also have to think about quality.

The general notion of incentive is quickly replaced by money and the article only discusses wages. The reduction of "incentive" to "money" is a common simplification in economics. Money has the nice quality that it is easily counted and fits nicely in equations. Most people work to support themselves and their family, but over two thirds of workers also work to "Feel useful/productive". http://www.pewsocialtrends.org/2009/09/03/iii-attitudes-toward-work/

The main payment schemes discussed are for sharecroppers, piece rate workers, and hourly wage workers. There is also mention of the value of pitting workers against each other so the most productive workers are disproportionally rewarded.

It is demonstrably true that compensation and incentives are important in the work place. The notions in the article are worth thinking about, but not within the framework presented. The intellectual framework for the pretended quantitative notions in the article is based on demonstrably false premises. The article include such gems as:
"Incentive contracts arise because individual (sic) love leisure. In order to
induce them to forgo some leisure, or put alternatively, to put forth efort,
some form of compensation must be ofered (sic)" 
This statement is true only when the work or the workplace involved is dehumanizing. Habitat for Humanity and other non-profit groups that rely on volunteers are simple counterexamples. Perhaps we should forgive the academic who thinks about work in terms of factory piece work and sharecroppers for simplifying the human condition to a form of slavery where monetary carrots and sticks are the only way to motivate.

Another quote gives the rosy view that the workers inevitably get the rewards of increased  output.
"Thus, (2) merely says that output, Q, must be paid entirely to the worker otherwise another firm could steal the worker away by paying more." 
Beware of arguments based on this type of logical sounding statement that bears no resemblance to reality. Simple evidence is the stagnation of wages over the past couple decades despite large increases in productivity. I did a couple blog posts on productivity. The most relevant one for this context is http://colin-quodlibet.blogspot.com/2010/09/if-we-are-to-believe-productivity.html.

The sad thing is that employers often believe this nonsense and instead of providing a more pleasant and humane workplace, they turn it into an unpleasant, metrics driven, competitive hell-hole.


Saturday, August 30, 2014

Conspiracies Without Cabals

Todays topic is human organization and the analysis of power. In particular, how we can have actions that look concerted, but which do not necessarily require collaboration between individual actors. That is we have things that may appear to be conspiracies, but without any organizing cabal of plotters.

We analyze situations at a level that is practical and with the tools that produce results. We cannot analyze complex physical phenomena in terms of wave functions because the mathematics are intractable. At the level of atoms and molecules we look at electron shells and bonding strength. At the level of materials we talk about stress and strain...

An analogous situation occurs when talking about human behavior. We have to decide which types of analysis are most productive for describing particular aspects of human society. Sometimes the appropriate level is examining individuals, their motivations and actions. At other times it may be more productive to look at larger scale organizations.

In the animal world, an example is a flock of geese. The flock travels long distances without a leader and without conclaves. At any given instant, the goose who is most certain about the proper direction influences the flock to move in that direction. No single goose knows the route, but at any instant one or another goose has a pretty good idea. The flock finds its way where the individuals might not. For analyzing travel, the flock is the correct unit.

We can view the political system as a means to resolve contention between individuals who have differing notions of man and society. Some people value loyalty and the preservation of known successful structures. Other people value fairness and adaptation to a changing world.

At a level up, we can view the US political system as a competition between two major groups, the political parties. The names and values of the parties change over time, but the structure of the US electoral system seems to ensure that there are only two parties (with the occasional splinter party). For analysis, we can look at the party itself, not the people in the party. That is, look at the machine. The parties have evolved over time, but for the US system, the important qualities seem to be: that there are two major parties, that the populace is partitioned into people who largely support one or the other, and that real consequences - power, money, and favor - flow to the inner circle of the party in power. At this level, the modern targeted marketing favored by both parties make sense. Elections are held. A party only gets power if its representatives get the votes in their district. Separating the electorate into very fine segments and appealing to a single issue that is important to a particular individual will sway votes and get the party into power. For the party, being in power is more important than any particular issue or group. The individual parties have expressed beliefs, the party platform. The platform is not any kind of coherent philosophy or reality tested approach. It is merely a collection of issues the core constituency cares about and that can be used to polarize the population and get votes. The party in power uses the apparatus of government to reward those who keep it in power. The favors generally flow to large contributors, often organizations not individuals. The base currency of reward is access. The result of access is most commonly legislation that favors financial interests (a group level result), but it can be personal (e.g. ambassadorships or other government jobs).

In this analysis the characteristics of the groups (parties, lobbying groups ...) rather than individuals is important. To my mind, this fits the facts pretty well. At an individual level, any participant may be internally persuaded that they are working toward future personal rewards, but sometimes the odds don't seem very good. During the presidential primaries of 2008 Andrew Young, an aid to candidate John Edwards, tried to protect the candidate by claiming to be the father of Edwards' illegitimate child. The notion that this lie would somehow advance Young's future personal prospects seems delusional. It looks more like a subsumption of personal interests to the aims of the campaign (the group).

Analysis of group behavior is difficult because groups are constantly appearing and disappearing and they form and break alliances. For example, there is a firearms industry that sells weapons to individuals. At one level we can examine each company as an organization promoting its own aims and competing with other firearm manufacturers. In the political sphere they have banded together and sponsor the National Rifle Association (NRA). The NRA is beholden to the manufacturers for much of its funding, but it also has a large number of individual contributors. The demands of the rank and file may move the organization away from the aims of the manufacturers. The NRA also gets direction from the specific individuals who lead the organization. The leaders of the organization have some autonomy, but they must operate within a range that is acceptable to the funding constituency.

The difficulty of analysis is increased by the natural human tendency to anthropomorphize the world around us. Every day on the news there is a story talking about the stock market rising or falling, and the reasons behind the movement. The "reasons" are invariably nonsense based on the feelings of "Investors". "Investors" or "The Market" is everyone who bought and sold stock that day. This is anthropomorphized into a thinking, feeling uber-being. On the news you will hear comments like "Investors were frightened by the newly released jobs figures, causing the market to go down". These comments reflect the personal feelings of news commentator and are not based on any rigorous polling of people making trades. If we did poll, we would undoubtedly find a huge range of reasons for buying and selling (the fund needed cash, rebalancing a portfolio away from stocks ...). Even rigorous polling would give us no sensible reason because most trades are not made by humans. Most trades are made algorithmically by computer programs so complex that no individual can, without extreme effort, determine why any particular trade was made. In the "flash crash" of May 2006 the stock market lost, then regained, 10% of its value in less than hour. After five months of analysis the SEC released a report trying to explain why the market fell on that particular day. The answer had nothing to do with the magical "Investors" or "The Market". The SEC explanation included a lot interaction between algorithmic trades (including high frequency traders). Almost a decade later, the reasons are still debated. If you hear a discussion about the stock market that includes human motivations for movement, it is almost certainly fantasy, often in service to the biases and aims of the organization that pays the commentator. That does not mean the commentator is deliberately lying. They are reflecting the views of those in their circle of "experts". Those experts in turn are employed by an organization with an organizational world view tuned to support the aims of the organization.

Analysis at the group level is not hopeless though. There are common strands that allow us to discover some rules and explain actions. Unfortunately, as in many complex systems, often we can only explain in hindsight. It is only in hindsight that we can see how the various forces actually played out. In hindsight, things that were uncertain when they unfolded appear to be more inevitable. This feeling of inevitability can, in turn, lead to the notion that there was deliberate intent by some small set of individual actors (the cabal). All of us have the intent trying to shape the future, but our individual intent is usually tempered by our position within society, which shapes what we think and believe, and by the organizations through which we act.

Partly because of our stated value of respecting individuals, US society is structured to support organizations rather than individuals . In the political sphere, we do not vote for a party, we vote for an individual in a winner takes all system. To get the resources necessary to win an election, individuals almost always have to join a party. The party has a ready place on the ballot for its candidates, the party has a pool of money and a network of people who can be mobilized to work for the candidate. To get this support, the individual must align him/herself with the aims of the party.

In national office, officials "represent" millions of people. They use "public opinion" to help shape their message (but perhaps not their votes). Public opinion can be gauged by polling, and by analyzing comments and requests from constituents. If you have ever written to a national elected official you can see how your feedback is handled. The "personalized" response from the official is a set of canned paragraphs on topics you mentioned. It is clear from this response, that you letter has been analyzed at the level of "concerned about issue X" or, at the most specific, "supports position Y". Based on that, the paragraphs are chosen. These days, the analysis of your letter and the response may be automatically generated without human intervention. I have asked my elected officials to let me talk to someone about this process, and been ignored. The input to the official is presumably equally coarse, basically a count in a spreadsheet that indicates how many constituent wrote in "support of position Y". That is, the responses are aggregated in a very simplistic way. In such a world, organizations have oversized influence because they can mobilize people to affect the count in the spreadsheet of public opinion.

This makes political life a competition of organizations. Powerful individuals can increase their power by creating "grass roots" organizations capable of generating pressure on elected officials by changing the counts in the spreadsheets. Thirty second ads are used to mobilize additional letters and calls. The smartest powerful people go one step further by controlling media to affect the way issues are framed. But, whenever an organization is created, even the most powerful individuals cannot force the organization to completely reflect personal beliefs. Inevitably, other individuals in the organization influence the direction as well.

When we see the results of this process, we may assume there is a conspiracy of individuals (the Trilateral commission, Davos participants ...). It is demonstrably true that actual public opinion has no effect on the legislative process in the US. But this does not mean there is a conspiracy with secret meetings, it is the result of the wars between competing, powerful, moneyed organizations. These organizations form alliances and work together when it is convenient. As alliances are created, individuals will meet, but I believe an appropriate level of analysis is the organization. Individuals create and manipulate organizations for personal gain. I contend individuals strongly influence organizations, but only occasionally control the outcome of the balance between organizations. Individuals generally ride the larger organization trends for individual advantage.