When can you trust your intuitions?

In my last post I discussed the part of Daniel Kahneman’s ‘Thinking Fast and Thinking Slow’ that I like least. In this post I will to discuss the part that I most enjoyed reading.

At the beginning of his book Kahneman sets up the idea that the human mind can be thought of as being comprised of two systems. System 1 operates quickly, with little effort and no sense of voluntary control. System 2 allocates attention to the effortful mental activities that demand it.

When I read that I immediately began to search for links to Timothy Gallwey’s concept of Self 1 and Self 2. Gallwey is a sports and business coach and author of popular ‘inner game’ books. I have read nearly all of Gallwey’s books and have written about them previously on this blog.

Gallwey observed that when he was playing tennis he seemed to have two identities. Self 2 was playing tennis and Self 1 was constantly interfering by telling him how to play and trying to get him to conform to his instructions.

It struck me that Gallwey’s Self 1 might correspond roughly to Kahneman’s System 2 and that Gallwey’s Self 2 might correspond with Kahneman’s System 1. Anyhow I didn’t find the link until I read Chapter 22 of ‘Thinking Fast and Thinking Slow’ in which Kahneman discusses his collaboration with Gary Klein, who turns out to be an admirer of Tim Gallwey's books.

Klein and Kahneman collaborated in a study directed toward answering the question of when you can trust an experienced professional who claims to have an intuition. Kahneman’s scepticism about intuitions was shaped by observing failures of intuitive judgements by experienced professionals.  He observed that experienced professional e.g. clinicians, stock pickers and political scientists often had too much confidence in their intuitions. He suggests that this occurs because System 1 tends to produce quick answers to complex questions, creating coherence where there is none.

Klein’s optimism about intuitive judgements by experienced professional was shaped by studies of leaders of fire fighting teams who seem to be able to make good decisions in emergencies without comparing options or knowing how they are able to sense the best course of action to take.

Klein and Kahneman agreed that successful intuitive judgement involves pattern recognition. Two basic conditions are necessary for acquiring a skill in intuitive judgement: an environment with sufficient statistical regularity for patterns to exist; and an opportunity to learn these regularities through prolonged practice.

Examples of statistically regular environments include sports, games such as chess, bridge and poker, and professions such as medical practice, nursing and fire fighting. By contrast, the failure of stock pickers and political scientists who make long term forecasts reflects the unpredictability of the events they are trying to forecast.

This all makes sense to me. When I am playing golf I should learn to trust Self 2 (System 1) and when I am trying to understand economic issues I should employ System 2 (Self 1).

However, that is an over-simplification. It probably isn’t wise to rely entirely on intuition when selecting which club to use when playing golf and the intuitions of economists have probably been the source of many a useful hypothesis about relationships between economic variables.

I particularly liked the way Kahneman ends his discussion of the relationship between System 1 and System 2 in the final chapter of his book. He suggests that System 2 is who we think we are – it articulates judgements and makes choices. (That is presumably why Tim Gallwey labelled it as Self 1.) Kahneman goes on to make the point that while System 1 is the origin of most of what we do wrong, it is also the origin of most of what we do right. The judgements and choices made by System 2 often involve endorsement or rationalization of ideas and feelings generated by System 1. 

Does libertarianism rest on rational actor assumptions?

‘The assumption that agents are rational provides the intellectual foundation for the libertarian approach to public policy: do not interfere with the individual’s right to choose, unless the choices harm others’ – Daniel Kahneman, ‘Thinking Fast and Slow’, Penguin, 2011.

I feel as though I am being somewhat churlish in protesting about Kahneman’s comments on libertarianism, which amount to only a few pages near the end of a 400 page book. In my view Kahneman’s book deserves high praise and it has indeed been widely praised (for example, even in a post on his blog by  David Friedman, who describes himself as an anarchist-anachronist-economist). Having thought slowly about the matter, however, it seems to me that it is important to try to prevent paternalists from getting a free kick from the reasoning that Kahneman develops in this book.

Much of Kahneman’s book is a discussion of research findings relating to biases in intuitive thinking. The view presented is that intuitive thinking (fast thinking) tends to be much more influential than we realize – it is responsible for many of the choices and judgements that we make. The confidence we have in our intuitions is usually justified, but they can lead us badly astray on issues that require deliberation (slow thinking).  For example, most people have particular difficulty in making judgements that require an understanding of probabilities. Kahneman is not optimistic that people can easily learn to recognize when they are in a cognitive minefield in which they need to slow down and question their intuitions. When people feel the stress of having to make a big decision, more doubt is likely to be the last thing they want.

My intuitions tell me that Kahneman may be too pessimistic about our ability to recognize when we are about to enter a cognitive minefield. It seems to me that many people have developed emotional systems that provide ample warnings when they are about to enter cognitive minefields. Since I am feeling such warning signals right now, however, my intuitions about this could well be wrong. I should confine my remarks to matters about which I can write with some confidence.

When I set out to write this post the plan in the back of my mind was to refer to some earlier posts in which I distance myself from the rational actor model employed by people like Gary Becker (whose theory of rational addiction is cited by Kahneman) and then to proceed to demonstrate that the classical foundations of libertarianism do not require the assumptions of that model. However, my early warning system suggested to me that it might be a good idea to check whether Becker actually bases his defence of libertarianism on the rational actor model.  It turns out that in the defence of libertarianism that I found, Becker actually distances himself from rational actor assumptions. (This is a post he wrote on the Becker-Posner blog in 2007 on the peculiar concept of libertarian paternalism - supported by Kahneman, but advocated originally by Cass Sunstein and Richard Thaler.)

Becker presents the view that I had planned to present more eloquently than I could, so I will quote him:

‘Classical arguments for libertarianism do not assume that adults never make mistakes, always know their interests, or even are able always to act on their interests when they know them. Rather, it assumes that adults very typically know their own interests better than government officials, professors, or anyone else ... . In addition, the classical libertarian case partly rests on a presumption that being able to make mistakes through having the right to make one's own choices leads in the long run to more self-reliant, competent, and independent individuals. It has been observed, for example, that prisoners often lose the ability to make choices for themselves after spending many years in prison where life is rigidly regulated. In effect, the libertarian claim is that the "process" of making choices leads to individuals who are more capable of making good choices’. 

Arnold Kling’s views on the implications of the cognitive biases documented by Kahneman are also worth quoting:

‘If social phenomena are too complex for any of us to understand, and if individuals consistently overestimate their knowledge of these phenomena, then prudence would dictate trying to find institutional arrangements that minimize the potential risks and costs that any individual can impose on society through his own ignorance. To me, this is an argument for limited government.

Instead of using government to consciously impose an institutional structure based on the maps of cognitively impaired individuals, I would prefer to see institutions evolve through a trial-and-error process. People can be “nudged” by all manner of social and religious customs. I would hope that the better norms and customs would tend to survive in a competitive environment. This was Hayek's view of the evolution of language, morals, common law, and other forms of what he called spontaneous order. In contrast, counting on government officials to provide the right nudges strikes me as a recipe for institutional fragility.

If Kahneman is correct that we have “an almost unlimited ability to ignore our own ignorance,” then all of us are prone to mistakes. We need institutions that attempt to protect us from ourselves, but we also need institutions that protect us from one another. Limited government is one such institution’  (‘The PoliticalImplications of Ignoring Our Own Ignorance’, The American, December 2011).

In responding to comments on his post, David Friedman has made a similar point that on balance Kahneman's work may actually favour the libertarian position that market decision processes are superior to political decision processes:

‘The arguments suggest that people are more nearly rational when they use the slow mind than the fast and, since the slow mind's attention is a scarce resource, they are more likely to use it the more important getting a decision right is. My market decisions are almost always more important to me than my political decisions, since the former directly affect outcomes for me, the latter do not. That suggests that people will be less rational in their political decisions than their market decisions.’

It is also worth noting that we do not have to choose between relying on our own individual thinking processes and relying on governments to guide us. In those areas of decision-making where we  may not be able to rely on our intuitions and deliberations we have family, friends, representatives of voluntary organizations of various kinds and paid professionals who may be willing to act as our advisers or our agents (as well as the social norms and customs mentioned by Kling). If I need an agent to make decisions for me, it seems to me to be preferable to appoint one to act as my servant than to appoint one to act as my master.

Finally, we should also recognize that when governments make paternalistic laws to criminalize stupidity they don't necessarily stop people from behaving stupidly. They may just add to the problems of the people they are trying to help.

How should we measure progress?

There seems to be a lot of talk about progress, or lack of it, at the end of each year. I tend to get caught up in this even though a year is far too short a period to measure the kind of progress that most interests me.

Two years ago I wrote a post entitled ‘What is progress?’ This was the first post I had written with the ‘progress’ label on it. At the time I intended to read several books and articles relating to the concept of progress and then to write something more definitive about the meaning and measurement of progress.

Since then I have read several books and articles about progress – from an historical perspective and looking towards the future – and have written 38 posts related in some way to the concept of progress. However, I don’t think I have written anything as definitive as my first post on this subject.

The main point I made in that first post about progress is that if progress is to have any meaning from a public policy perspective it must mean movement toward a good society or movement from a good society to a better society.

A fairly obvious response that might come to mind is that it could be just about as difficult to define what we mean by a good society as to define what we mean by progress. As things happened, however, I had just spent a few months in 2009 thinking about the characteristics of a good society.

I had reached the conclusion that just about everyone should be able to agree that a good society is good for its individual members. Such a society would enable its members to live together in peace. It would provide its members with opportunities to flourish. It would also provide its members with some security against threats to their flourishing. I had also come to the conclusion that these characteristics of a good society are measurable.

It follows, or so it seems to me, that the best way to measure progress is to bring together relevant indicators of the peacefulness of societies, opportunities for flourishing (including consideration of economic, environmental and social capital indicators) and security (including consideration of security against misfortunes such as ill health and unemployment).

The approach I am suggesting is similar to that followed by the Australian Bureau of Statistics (ABS) in its ‘Measure of AustralianProgress’ (MAP). The difference is that ABS offers a smorgasbord of social, economic and environmental indicators which could, in principle, cover everything that anyone has ever suggested might have some relevance to the question of whether life is getting better. I think attention should focus on indicators that nearly everyone would agree to be closely related to important characteristics of a good society.

I strongly support the ABS’s approach of recognizing that progress is multidimensional and refraining from any attempt to combine indicators into a single measure. It seems to me that so called ‘genuine’ progress indicators which reflect the value judgements of individual researchers relating to such matters as income distribution and environmental values are useless. The relative importance of progress in various dimensions must remain a matter for public discussion and judgement by individual citizens. If a collective judgement is required about the priority that should be given to various dimensions of progress, we have constitutional processes including elections and parliamentary processes to perform this task.

Since the combining of progress indicators must involve individual value judgements, why not just ask individuals to make an evaluation of their own lives (on a scale of 1 to 10), combine these evaluations in some way and use this as our measure of progress? There are several problems with measurement of subjective well-being in this way, as discussed elsewhere on this blog. As I see it the main one is in ensuring that respondents have an appropriate benchmark in mind for measuring progress when they make their evaluations. If you ask people to assess their own lives relative to ‘the best possible life’ as in the Gallup surveys, the results of successive surveys cannot provide a measure of progress because perceptions about ‘the best possible life’ can be expected to rise as a result of progress. If I am climbing a ladder that is attached to a helicopter, my height above sea level depends on the height of the helicopter as well as on my ability to climb the ladder.

So, I think our measurement of progress should focus on widely accepted criteria that are relevant to the question of whether we are making progress toward more peaceful societies that offer greater opportunities and more security. There is also a more fundamental question, however, of whether the institutional drivers of progress – for example, institutional factors leading to productivity improvements - are also moving in the right direction. Perhaps I should write more about that next year.

Have important factors been omitted from the HALE index of well-being?

The aim of the Fairfax organization in sponsoring the development of the Herald/Age - Lateral Economics (HALE)  index of the well-being of Australians seems to have been to publish a broad indicator of social progress in the hope that this will help people to avoid viewing GDP as ‘the supreme indicator of our wellbeing’.

In contrast to some previous attempts to create ‘genuine’ progress indexes for Australia, which seem to have been aimed at maximizing the weight placed on possible negative spillovers associated with economic growth, the authors of this index seem to have adopted a fair-minded approach. However, I still have some concerns about the methodology adopted. I discussed one of those concerns in my last post – namely that it would be desirable for the index to take into account changes in uncertainty about the economic situation if it is to be taken seriously as an indicator of short term changes in well-being. In this post I want to identify important factors that have been omitted from the HALE index that might affect its use as an indicator of longer-term changes in well-being.

It seems to me that the most important factors affecting individual well-being are social capital (respect for person and property, quality of governance, individual safety, inter-personal trust) national security (peacefulness of the international environment, relations with other countries, security threats) physical and financial capital (financial wealth, housing, infrastructure, indebtedness, economic security) human capital (skills, health, personal relationships and emotional well-being) and natural capital (natural resources, environment). The relatively importance of different factors must ultimately be a subjective judgement, but this does necessarily mean that all important factors are taken into account when people are asked to rate their satisfaction with their lives. For example, there is empirical evidence that even though personal safety is obviously fundamental to individual well-being its contribution to measured life satisfaction is negligible in Australia (see, for example, a study I have undertaken using the Australian Centre on Quality of Life data set). One possible explanation is that most people feel so safe living in Australia that safety concerns do not even register in their minds when they are asked about their life satisfaction.

The most obvious omissions in the HALE index are social capital and national security. Those factors are unlikely to affect well-being much from year to year, but their impact over several decades could be substantial. For example, looking back over the last 40 years, there has arguably been a substantial improvement in the well-being of Australians as a result of improvements in relations among countries in the Asia-Pacific region.

Some less obvious omissions in the HALE index may also be important. The starting point of the index, net national income, reflects some flows of services from human, physical, financial and natural capital and one source of change in capital stocks (net investment in physical capital). Subsequent adjustments to take into account changes in environmental capital and human capital are presumably aimed at measuring changes in capital stocks more comprehensively to obtain a comprehensive income measure (based on the Haig-Simons definition of income i.e. consumption plus change in net wealth). I use the word ‘presumably’ because change in human capital from improvements in school education is measured in terms of the estimated effects of an improvement in current PISA scores on long-run GDP, without any discounting to take account of the passage of time required before improved PISA scores could possibly be reflected in the human capital of members of the labour force. It seems to me that, rather than fluctuating widely depending on literacy and numeracy skills of the current crop of school children, the value of human capital stocks probably changes gradually over time as people with differing skill levels enter and leave the labour force.

The market values of some forms of wealth obviously fluctuate fairly widely from year to year, but this is not taken into account in the methodology used in calculating the index. Changes in the value of financial capital and housing are ignored in calculating the index. This raises the question of whether the effects on well-being of such unrealized capital gains and losses are as great as for changes in current income. My feeling is that they are probably not as great. Investors are likely to view capital gains and losses in a different light to changes in dividends. Home owners who obtain unrealized capital gains on their homes would probably not generally feel that there has been much change in their well-being – their home still provides the same services to them as it did previously. Their lives probably remain largely unchanged unless, of course, they run down their liquid assets of borrow funds in order to spend their capital gains.

However, this brings me to what seems to be an important omission in the HALE index - it doesn’t make any allowance for changes in debt levels. Well-being is more closely related to net wealth than to total physical and financial assets. Looking at Australia as a whole, debts cancel out to a large extent – the liabilities of one person are the assets of another – but they do not cancel out completely. Changes in net foreign debt levels may have important implications for the average well-being of Australians. Changes in interest rates on foreign debt should also be taken into account because they influence the extent to which current income is available for purposes other than debt servicing.

Research by the Australian Centreon Quality of Life several years ago shows that people who have difficulty in repaying debt tend to have lower subjective well-being than those who do not have such problems (Survey 11, Report 11, August 2004). In the light of current debt problems in many developed countries it seems remarkable that happiness researchers have not given a great deal more attention to the effects of excessive debt on personal well-being.