Thursday 26 March 2009

Economic Egoism

15 comments
Economics, like any other empirical enterprise, makes simplifying assumptions to make the challenges in question tractable. However the solutions provided are dependent upon those assumptions having been in some sense empirically validated, otherwise such simplifications might mislead more than edify.

One assumption is over how individual agents exercise choice. They are assumed to operate with purely self regarding desires or only selfish motivations. From this a theory of economic rational behaviour is derived, based on the combination of selfish motivations (ends) and instrumental (means-end) rationality, where everyone is seeks to maximise their benefits from the exercise of their choices. Selfish agents are rational maximisers. This assumption is sometimes known as the rational expectations hypothesis.

Now actual individuals vary from behaving as identical rational maximisers due a wide range of disparities in biology and culture - age, skills, education, influence, information and so on. However it is also assumed that these disparities cancel out, so that in an aggregate fashion, individuals collectively tend to be selfishly motivated rational maximisers. This also allows for non-selfish motivation, other regarding desires as noted in my last post. However these values are expressed, it is assumed that they tend to cancel out.

When people wish to maximise their available resources they seek to maximise their return whether due to labour, investments or some combination and to minimise the risk associated with that return. In order to gain more - a higher return - they might undergo more risk to achieve this. Working for a junk food shop offers low risk and related low return. Starting your own business offers a potentially higher return but with an associated higher risk (the business might fail). Generally most would not undertake more risky endeavours or investments without the accompanying expected return. One does not create a start up to earn junk food wages. There is purportedly an optimum risk and return ratio for any agent's particular skills, resources and requirements.

The economic challenge is to measure such risk to discover this optimum. A key benefit of rational expectations approach is the relatively simple mathematical characterisations of risk that should be expected, if this is true. No man is an island and so statistical averages can be derived due to everyone interacting with everyone else. That is the rational expectations hypothesis enables everyone to be considered as both making decisions independently and to conform to an identically distribution - due to, in aggregate, all behaving as rational maximisers as other behaviours cancelling out. The aggregate is the arithmetic average, the distribution is the normal, Gaussian distribution or bell curve and the measure of risk is the standard deviation around this average.

The above is just a sketch but whilst a crude simplification of this economic issue sufficiently captures key assumptions that are embedded in and used in economic reasoning not just in business but governments. Does this approach lead to empirically sound models of choice behaviour?

In fact other sub-disciplines within economics such as behavioural finance and in other disciplines such as cognitive psychology have more than cast doubt on the assumptions underlying and the conclusions resulting from the rational expectations hypothesis.

A large variety of studies have shown that we are exhibit a variety of consistent and predictable asymmetries in decision making and that we are affected in various ways by many heuristics and biases and all which are regarded as irrational and not self regarding according to the rational expectations hypothesis. Further these asymmetries, heuristics and bias can be systemic on an aggregate level and so rather than cancel out - with or without negative feedback - and can persist or even be amplified in positive feedback in group choice making actions.

The implication of such systemic biases is that this destroys the independence and identically distributed * assumptions that underlay the statistical assumptions of our most common risk measures. This makes our measures of risk at best different and at worst undefined and this means that the amount of effort and investment undertaken to generate an expected reward for a given risk might be wildly different to reality.

Another challenge to the rational expectations hypothesis is over the maximising assumption. Herbert Simon won a Nobel Prize in Economics for his work on satisficing- choosing what is "good enough" over a few choices compared to seeking the optimum value over many choices. Later psychological work has shown that satisficers are happier than maximisers.

All in all this work shows that the rational expectations hypothesis or economic egoism - as I titled this post - has little empirical support. We are not purely selfishly motivated, and even when we think we are we are "irrationally" affected by heuristics and biases in a systemic fashion that can be taken advantage deliberately and unwittingly by various interests.

An important fact is while much of this work is recent, it is not that recent, and is sufficiently well known within economics, even it has been all too often repeatedly ignored. This makes the recent financial debacles quite different to last major ones in the 1980s. Yet still it has occurred and is indeed still occurring and it is quite unclear as to its resolution.

One factor in this has been the interaction between economic egoism and ethical egoism. Each position has been drawing support and justification from the other. However this and the last post together indicate that there is no empirical support for either type of egoism. Neither can the lack of empirical evidence in either area - even noting that they have similar but distinct notions of selfishness - cannot jointly support either egoism.

Clearly though selfish motives, of course, do exist, some people can act with extreme selfishness, with utter disregard for others, as the latest financial meltdown has nfortunately demonstrated. The points here do not contradict this but rather indicate that such arguments for economic and ethical egoism, ones that have been used to support and justify regulations and policies that have contributed to the mess we are in, did not and do not have the empirical support that was assumed.

* [In fact Stable Levy or Pareto distributions with undefined or infinite variance can be the result of iid variables. This paragraph is focused on the measurement of risk issue]

15 comments:

Larry Hamelin said...

The points here do not contradict this but rather indicate that such arguments for economic and ethical egoism, ones that have been used to support and justify regulations and policies that have contributed to the mess we are in, did not and do not have the empirical support that was assumed.

Examples?

Martin Freedman said...

Not sure what you mean by examples and this was only a preliminary post to introduce this theme in my blog.

Anyway I already mentioned Simon with respect to satisfycing over maximising. Off the top of my head Schiller, Gigerenzer, Tversky, Amos et al have done much on the asymmetries of decision making and systemic biases. Search for "Heuristics and Biases" for reams of cognitive psychology work that can relate here.


All go to show that the hypothesis of economic selfishness - rational maximisation - is not emprically supported.

The implications of this affects the distributions of agent preferences and market dynamics in general (where rational expectations translates into the iid assumption), they are not iid (independent, identically distributed), and the central limit theorem does not show that these tend to normal distributions and so related statistics most simply standard deviation as a measure of risk are invalid.

It is an open question as to what the actual distributions are and related risk measures could be. Off the top of my head again Mandelbrot, Fama, Hsieh, David Lo and others have lead the way here.

Martin Freedman said...

typo David Hsieh, Andrew Lo

Martin Freedman said...

If I misread your examples question another type of answer would be Greenspan and other Randiods - amongst others - influencing policy and regulations etc.

Larry Hamelin said...

There are two claims: 1) People act according to rational self-interest; 2) people acting according to rational self-interest leads to a self-regulating economy.

I'm not persuaded that systematic deviations from iid have contributed to the current financial crisis. The large finance capitalists seem to have rational comprehension that downside risk would be substantially delayed (as Keynes notes, "In the long run, we're all dead") and fall to the taxpayers, not the shareholders.

I'm not trying to defend economic "rationality"; Economists define "rationality" very narrowly; such that it's "rational" to sell my grandmother into slavery for $1 if she cuts me out of her will.

But I don't see that systematic deviations of human behavior from the assumption of economic rationality has had much impact on the crisis.

Keep in mind that for the large capital owners, this "crisis" is -- like all depressions -- an *opportunity* to acquire more assets and more economic control. Everything is working just fine for the bourgeoisie.

Larry Hamelin said...

The idea that rational self-interest leads to a self-regulating economy is utter nonsense, and contradicted by the simplest economic models.

Martin Freedman said...

In answer to your first comment before this.

Here is just one example of how iid was a problem, noting this was not the main thrust of my post.

Recollecting 18 year old knowledge when I examined mortgage backed securities, the problem was to turn a set of mortgage debt into a bond with which could reliably pay its coupon. The issue were early redemption and foreclosure. Three bonds were built from three tranches of a debt, with differing coupon rates and risk which could be made to correspond to normal bond credit ratings.

Simplistically high early redemption, low foreclosures were put in one tranch, high foreclosure, low early redemption were put in another and the rest in the third. Each tranch had a different risk profile (as well as coupon rate due to differential expected early payment/non-payment) to determine a suitable equivalent credit rating.

Now there was no way to do this unless iid or equivalent assumptions were used. In fact there was no way to carve up the risk over a robust range of market conditions with their relations to the job market consequential capacity for mortgage debtors to repay reliably. Independency varied and there was no single risk measure to capture this - in extreme markets behaviours become correlated and not independent and positive feedback occurs.

This was all hidden in p values and confidence levels which systematically understimated risk, all due to theoretical arguments based on rational expectations hypothesis.

Now this is all 18 year old knowledge and I have not refreshed myself on this topic. No doubt many sophisticated arguments and new risk models have been introduced but what has happened the last couple of years was - specifically sub-prime mortgages and their funding -looks like a pretty clear confirmation of those issues I identified a long time ago - which was why I refused to work in mortgage-backed securities and not take the money and run - as obviously many others did. Of course I did not realise nor predict the gravity of the error perpetuated to such a level has occurred.

Nonetheless this aspect of the crisis was entirely avoidable given these facts.

And this brings me to the second of your comments above. I agree.

Larry Hamelin said...

I think you're using "independence" equivocally.

In your original article, you referenced that deviations from rational self-interest were independent: some people might, for example irrationally avoid risk, others would irrationally seek it, leaving aggregate, mean risk-taking behavior closely tracking rational expectations.

In your latest comment, people's rational self-interest can be mutually dependent: i.e. changing circumstances can influence rational self-interest in a systematic way.

I'm curious to know if systematic deviations from rational self-interest (i.e. the majority of people acting irrationally) in an economic sense can be shown to have non-trivial macroeconomic effects.

(My question is not at all rhetorical, by the way. I really am interested in knowing.)

Larry Hamelin said...

Note too that I t think the premise of rational self-interest is different from the premise of perfect information. Although it might well be a systematic deviation from rational self-interest for a lot of people to systematically irrationally believe they had more information (or information of higher quality) than they had reason to believe.

Martin Freedman said...

"In your original article, you referenced that deviations from rational self-interest were independent: some people might, for example irrationally avoid risk, others would irrationally seek it, leaving aggregate, mean risk-taking behavior closely tracking rational expectations."
Correct. There are numerous empirical studies showing that the assumption of rational expectations is questionable. This also questions as to whether "rational self-interest" as defined by rational expectations is actually "rational", or is this really a semantic issue. That is not to say that there are and can be "irrational" systemic biases in the exercise of choice by people.

The general issue is that rational expectations is used to justify iid and the resultant normal statistics which makes the problem of risk appear tractable and mathematically. And this relies on mutual and reciprocal support from ethical egoism but neither have sufficient empirical grounding and share the same presuppositions which was my underlying point in this pair of posts (and indeed, as you brought up, the shared suppositions are due to psychological egoism which suffers the same semantic and empirical issues).

"In your latest comment, people's rational self-interest can be mutually dependent: i.e. changing circumstances can influence rational self-interest in a systematic way."
This was an illustration of the underlying use of rational expectations to justify iid and normal distributions. If I am reading you correctly, what you mean by "rational self-interest" here is not derived from the a priori assumption of rational expectations but looks more like the a posteriori actual behaviour of people and labelling that "rational self-interest"?

I am saying in numerous ways that short term and long term dependencies can occur but they are not stationary and suitable analysis and the law of numbers can sometimes wash out these dependencies and lead to erroneous conclusions and, in this case, under-estimation of risk (Other methods ignore such assumptions as rational expectations and work off empirical distributions and non-parametric models - there are other issues to address there. I guess much of the later work in mortgage-backed securities did make such arguments but I don't know). Systemic psychological bias that can be amplified not dampened by feedback is one such causal factor in this.

"I'm curious to know if systematic deviations from rational self-interest (i.e. the majority of people acting irrationally) in an economic sense can be shown to have non-trivial macroeconomic effects."
I am not sure the question is well-formed. Are there systematic deviations from rational self-interest" as defined by rational expectations then there is plenty of evidence to show this is the case. I will in due course write a post or series of posts addressing this. Are there, regardless of how rational self-interest is defined systemic biases that can be detected macro-economically? This is a question over short- and long-term dependencies and non-independence and non-stationarity in distributions and there is a large literature in quantitative finance on this topic. I mentioned Mandelbrot, Fama and others in a previous comment. Is any this empirical data sufficient to derive a better parametric model than rational expectations - well that is a holy grail in economics. I do no think there any agreed solution but there are some candidates, when I have time - who knows - this would be a good topic to pursue.

This is getting into economic theories of value and certainly the labour theory of value is a failed hypothesis here but then I am no advocate of marginalism either. I have no answers here except we need to do better than either. That is one of my interest in Fyfe's "desire fulfilment" as a theory of generic value but I do not have time to explore this sufficient well in economics. (It might be fun debating this so we can learn off each other's thinking. More something to discuss in an open thread than for me to make a definitive post on my blog for now - since I can't say anything definitive on this yet).

Larry Hamelin said...

And this relies on mutual and reciprocal support from ethical egoism but neither have sufficient empirical grounding and share the same presuppositions which was my underlying point in this pair of posts

Actually, you made a stronger claim: that the assumptions of economically rational behavior were contradicted by empirical evidence.

Are there systematic deviations from rational self-interest" as defined by rational expectations then there is plenty of evidence to show this is the case.

What's the actual evidence? Remember, the deviations have to be systematic, i.e. everyone has to deviate in the same way. Otherwise the deviations cancel each other out.

Are there... systemic biases that can be detected macro-economically? ... [T]here is a large literature in quantitative finance on this topic.

If there's enough literature to make the point obvious, then why are you talking about it? If not, I'd be interested in hearing about the actual details of Madelbrot, et al.'s research.

It's not very informative to substantiate a position on a controversial topic by saying, essentially, "Get an economics degree and you'll understand." I'm not asking for a dissertation, just a little more detail than "Mandelbrot".

(On the other hand, it would be legitimate to say (if true) that the rational behavior hypothesis is simply ridiculous to serious economists, and those economists who use it are as dishonest as a scientist who talks about philogiston, the luminiferous ether or intelligent design.)

Martin Freedman said...

"Actually, you made a stronger claim: that the assumptions of economically rational behavior were contradicted by empirical evidence.
"

On reflection your observation is correct and I think this is quite correct from my past knowledge of this topic. I would be surprised if anything has changed, more likely there is better evidence that rational expectations is dubious.

"It's not very informative to substantiate a position on a controversial topic by saying, essentially, "Get an economics degree and you'll understand." I'm not asking for a dissertation, just a little more detail than "Mandelbrot"."
Good question and you are providing plenty of seeds for new posts. Now this is about analysis of empirical distributions and dependencies that one would not expect if rational expectations and no systemic bias were correct. However I have little time to pursue this to the level that I (and you) want, for now. I will be expanding on this theme the next few months.

For now I was most interested for now addressing the supposed mutual support between economic and ethical egoism. I can't find anything empirically substantive, can you?

"On the other hand, it would be legitimate to say (if true) that the rational behavior hypothesis is simply ridiculous to serious economists, and those economists who use it are as dishonest as a scientist who talks about philogiston, the luminiferous ether or intelligent design."
That is possibly not so far from where I could go with this, once I have my references unpacked and time to go through them and check for any new data that could contribute. This is a work in progress. For now AFAIK rational expectations is not rejected by all serious economists (as of 15 years ago, dunno about today) but there is most definitely a decent set of (mostly younger) economists who do reject this.

When I can I will pull out my old papers on this and update them and publish them here(they were focused on the related Efficient Markets Hypothesis).

However I am limiting my total blog time to 5 hours a week so it will take while to get there.

Larry Hamelin said...

However I am limiting my total blog time to 5 hours a week so it will take while to get there.

I will wait with worms on my tongue (bated breath).

Martin Freedman said...

OK I updated the post with some links which took longer to check out than writing the post. Nothing against what I asserted but I added a footnote for clarification of a point, that will be dealt with in a future post.

I have no time to do new original posts this week anyway, so this will have to do for now.

I noticed that wikipedia is pretty bad in covering financial risk and alternate distributions with respect to finance. Interesting. Gives me something to blog more about in the future but first I need to unpack my books so I can find my references (one of my many tasks this week).

admin said...

govt jobs alerts | central gov job |job central government | government job for india | govt job for bihar | govt of jobs in india |