Home > Uncategorized > Neoclassical clunkers. How economists’ ideas about risk aversion are duping pension pundits.

Neoclassical clunkers. How economists’ ideas about risk aversion are duping pension pundits.

(Part of) the Dutch pension system will, if everything goes according to plan, soon be replaced with a new neoliberal system based upon, among other things, measured neoclassical ‘risk aversion’. However – economists are not yet able to measure this – which will lead to big problems. let me explain.

The present system consists of:

  • a social democratic element (the ‘first pillar’), a kind of not means tested basic income for everybody above 67 financed by taxes.
  • Next to this is the corporatist ‘second pillar’, largely based on Christian social thought, which consists of non-government non-profit sectoral or, sometimes, company based pension funds which pay funded pensions. These are financed by mandatory pension savings by workers and their employers.
  • The ‘third pillar’ is based on classic liberalism and consists of voluntary pensions provided by private financial companies and funded by contractual savings.

The idea is to transform the collective second pillar part into a would be individual system, organized along neoclassical lines. How does this work?

The collective funds are exactly that: collectivities. There are no personal accounts. In a sense, everybody shares alike in the (relatively high) returns of investments of the funds. This is possible as the funds invest as a collectivity with a very long time horizon. As there are no personal accounts, just pension rights, there is no need (or possibility) for individual members of the funds to change from high yielding risky assets to low yielding ‘save’ assets when they get older, meaning that the investment portfolio can consist of a relatively large share of risky assets, even when a fund might contemplate this when the average age of its members increases. This is difficult to understand for the neoliberal reformers, whose idea of society is based on the idea of individual contracts and purchases. Based upon this world view – but contrary to the nature of the funds – they use rhetorics which consistently describe the collectivities as a set of abstract personal accounts, some ‘abstractly owned’ by young people and some by old people. As young people (read: people still paying premiums) are supposed to have more years left and hence to have a higher risk appetite and hence fancy more risky assets and hence obtain a higher average return on investment than old people (read: people receiving a pension) while, in the present system, old and young share alike in the return of the funds, neoclassical economists state that the young are in effect subsidizing the old. According to them, this has to stop (even when the Dutch pension reality is that the old are basically subsidizing the young – I’ll return to this in another post). Pension funds should according to them channel more of their returns to the young! To be able to do this, the collectivity has to make way for a myriad of personal accounts (about ten million) while, to be able to re-distribute returns from the old to the young, it’s also necessary to know the difference in neoclassical defined risk aversion between the young and the old. In reality differences in social class, i.e. between rich and highly educated people (who live long) and poor and lowly educated people (who die early) is a more glaring inequity of the system than the difference between young and old. In my country, the young get old. But the poor most of the time do not get rich… But that’s left out of this discussion. The idea that the young are subsidizing the old has gotten the upper hand in the political discussion, funds have to introduce millions of personal accounts while also measuring generational risk aversion as the risky and save collective assets have, at least arithmetically, distributed among these accounts.

But: can the funds measure ‘risk aversion’ as defined by neoclassical economists. Nope. To be able to do this and as they use neoclassical models they need to measure relative risk aversion of generations. This is even one of the legal cornerstones of the proposed new system – it might soon be law law. But (neoclassical) economists are not even able to properly measure risk aversion of individuals, let alone risk aversion of entire generations. There is no proven way to do this. The new system will have a cornerstone which not yet exists. The redesign and the proposed law are squarely based on the assumption that (neoclassical) economists are able to measure risk aversion in a precise way. They aren’t.

I’ll give an example of this ineptitude. One institution playing an important role is the Autoriteit Financiële Markten (AFM), a financial watchdog for the entire Dutch financial sector. They advise (in fact: will force) pension funds to use neoclassical inspired surveys to measure ‘relative generational risk aversion’, using the FRAME-system, This, however, does not help. ‘FRAME’stands for:

  • Feasible
  • Rationalising
  • Appropriate
  • Measureable
  • Errable

What do they mean with this?

About Feasible. Of course, any survey has to be feasible: simple, short, unambiguous questions etcetera. And one can –and has – use some model of pensioners and pensioners-to-be to derive questions for such a simple, short, unambiguous survey and, after appropriate qualitative and quantitative analysis of the primary results, to have something which can be used to make sense of results like averages, medians, Cronbach-alfa’s, word clouds and whatever. But: can one use the neoclassical consumer model to derive questions about risk preference? The AFM itself already states that people (which might not always have a degree in economics and sometimes functionally illiterate) often get confused by the neoclassical concept of ‘risk preference’. Heck, many economists get confused by the concept, not least as experts are still arguing about the exact shape of the utility curve as well as ways to measure neoclassically defined risk preferences (look here, here and here, mind the extreme differences in measured preferences)! It is as yet not possible to derive an unambigous answer to the question which questions have to be posed (even assuming that risk aversion can be measured using surveys).

About Rationalising. It won’t come as a surprise but all kind of criticisms of and additions to the neoclassical homo economicus, for instance by Kahneman and Tversky, are not only disregarded but explicitly excluded from the proposed survey of the AFM. All kind of proven ‘irrationalities’ which defiate from the homo economicus should according to the AFM explicitly be excluded from the model and the data. And ideas like those of the data driven brain scientist Dick Swaab and the equally data driven neurologist Viktor Lamme are totally disregarded. Both scientists state that the brain is a network with multiple nodal points. Depending on the ‘internal’ biological situation (think: hormones, aging and the like) and the external situation, some nodes will be more active or less active, resulting in behavior which changes from situation to situation. Meaning that answers to a survey about preferences might (as we all know) not correspond to actual behavior and hence be useless as measures of generational relative risk aversion. There just might be no ‘utility curve’ at all, according to work of these scientists. The neoclassical approach to consumer behavior is just one approach – and not the one most favored by for instance marketeers… Disregarding everything which contradicts the homo economicus does not seem to be a wise strategy. Some measure of risk aversion will be found, but what does it tell us when its foundations are flawed and incoherent?

About Appropriate. The questions should be appropriate, meaning that they have to be about pensions and risk. This has to be made clear in the survey. According to the AFM source linked above (my translation): ‘’sketching the right pension context can be done by presenting choices in terms of total net income after pensioning, including AOW [basic income in the Netherlands for people above 67, M.K.] and corrected for inflation”. This presupposes knowledge about future taxes and the future of AOW (we’re talking about decades here!) and presumably we also do know the future rate of inflation and (but that’s somewhat nerdy) hence also the future composition of the ‘basket’ of goods and services used to calculate inflation and hence also future consumption habits. Again, this is not about next year but about 2050 and 2070. It’s bonkers. We do not know. We can’t sketch such a future. We just can’t ask a question which, in an implicit or explicit way, asks from people to make choices based on information we do not have.

About Measurable. The AFM states that measured ‘relative risk aversion’ per generation has to be the basis for the redistribution of yields on total investments of pension funds to the young. It states that this is a well known variable (questionable) which is well conceptualized (wrong) and which can, supposedly, easily be measured by using a survey (wrong). The concept is not clear, the basic model (the shape of the utility function) is not clear (let’s be honest: utility functions do not exist, even when our ‘left brain interpreter’ continuously tells us that our clearly inconsistent and incoherent choices and behaviors are highly rational and coherent). Attempts at measuring risk aversion are still in their infancy and and yield wildly different outcomes. Aside of this, there are the usual problems with surveys regarding representativity (what to do with the illiterate etc. etc.) and validity (even small changes in questions can lead to quite different results). Simple and quite large surveys like the unemployment survey or the consumer confidence survey enable us to track changes in time. But even for unemployment, levels are highly dependent on the exact phrasing of questions (look here for a recent Dutch example). Even on its own terms, neoclassical economics is not ready to measure generational risk aversion. Also: what’s a generation? I can come up with multiple definitions but which is the best one?

About Errable. If answers are, according to the questions posed and the model behind these questions, inconsistent they have to be set aside according to the AFM, for additional analysis. Inconsistencies may be due to the respondents not ‘getting it’. They may also be due to ambiguities in the questions. The even may be due to a model of pensioners which is clearly inconsistent with how people really are… Should such an inconsistent model really be a cornerstone of the pension system? Might the errors in fact be ‘true behavior’?

Summarizing: the present Dutch pension system has problems. However – one can doubt if differences in badly conceptualized risk aversion between badly defined generations which (according to the rhetorics of neoclassical economists) necessitate transfer incomes between the old and the young are a real problem. Other, social class based, problems like the difference between the well educated and rich on one side and the poorly educated and poor on the other seem to be more pressing. Economists simply can’t measure risk aversion. Neither the concept nor the model nor the measurement methods used enable apt measurement. As measuring generational risk aversion is a cornerstone of the proposed new system the whole new system risks becoming an unmitigated disaster.

This is not the only problem. Another one is that, if the entire world would have funded pensions, there simply would not be enough assets to fund them… Basic incomes and free medical care seem to be better solutions. But let’s leave that for another time.

  1. Romar Correa
    August 1, 2022 at 5:56 am

    Grateful for your scholarly diatribe, merijntknibbe. Equivalently, we have the hoary problem of going from individual discount rates to social discount rates. The debate revolving around climate change has stretched neoclassical economics to the point of bathos. The severe strictures against surveys purveying wisdom were inspired! Yesterday, we had low inflation. Which economist, if asked then, would have forecast high inflation? Do we sweep everything under the carpet by recourse to supply shocks, a pandemic and a war in this case? Are these shocks temporary or permanent? I would welcome a non-anonymous survey of economists about whether the current inflation will persist over the rest of the year and 2023.

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: