It is just a coincidence, but the initials NNT are best known to me as Numbers Needed to Treat. This is a measure of the numbers of patients you need to give a drug to in order to get one cure. For example, an NNT of 5 means that you have to treat five patients in order to get one patient cured. Even very useful drugs do not help everybody, so several people need to take the drug before one of them benefits.

I do not know how many posts I need to write to convince one person that Nassim Nicholas Taleb is an unreliable guide to intelligence research. It might take 30 posts, but I still think it would be worthwhile.

One of NNT’s complaints was that intelligence test scores did not predict important real-world achievements, such as being a good investor. A very quick search immediately came up with a paper which showed that brighter people had more successful investment histories. Those authors found that by making better stock selections and achieving lower transaction costs high IQ subjects did 4.9% per year better than low IQ subjects. Given that real returns average 7%, this is a massive difference which will accumulate over time and result in far high net personal worth for brighter investors. By the way, intelligence was measured at conscription age, long before there is much investment history, so it is more likely to be causal.

https://www.sciencedirect.com/science/article/pii/S0304405X1100211X

In this post I will describe another paper which shows that an intelligence test predicts savings 40 years later. This is real-world skin-in-the-game of which NNT should approve.

Furnham, A., & Cheng, H. (2019). Factors influencing adult savings and investment: Findings from a nationally representative sample. Personality and Individual Differences, 151, 109510. doi:10.1016/j.paid.2019.109510

https://drive.google.com/file/d/10TFhHwSEUhrSIOCBNdZnNUhQxNstPb6-/view?usp=sharing

The authors review the cognitive and personality measures which affect savings rates: neuroticism and conscientiousness affect outcomes. Both these measures involve paper and pencil questionnaires designed by psychologists, yet are associated with real life outcomes.

The National Child Development Study 1958 is a large-scale longitudinal study of the 17,415 individuals who were born in Great Britain in a week in March 1958 (Ferri, Bynner, & Wadsworth, 2003). They were a representative sample of the country at the time. 14,134 children at age 11 completed tests of cognitive ability (response =87%).

At 33 years, 11,141 participants provided information on their educational qualifications obtained (response =72%). At age 50 years, 8210 participants provided information on their current occupational levels (response = 67%); 9790 participants completed a questionnaire on personality (response = 79%); 9762 participants provided information on their self-assessed financial situation (response= 79%), 9729 participants provided information on their savings and investment (response =57%). The analytic sample comprises 5766 cohort members (50% females) for whom complete data were collected at birth, at ages 11years, and the outcome measure at age 50years. Bias due to attrition of the sample during childhood has been shown to be minimal (Davie, Butler, & Goldstein; 1972).

3.2. Measures

1. Family Social Background at Birth Family social background includes information on parental social class and parental education. Parental social class at birth was measured by the Registrar General’s measure of social class (RGSC). Parental education is measured by the age parents had left their full-time education.

2. Childhood Intelligence Childhood intelligence was assessed at age 11 in school using a general ability test (Douglas, 1964) consisting of 40 verbal and 40 non-verbal items. For the verbal items, children were presented with an example set of four words that were linked either logically, semantically, or phonologically. For the non-verbal tasks, shapes or symbols were used. The children were then given another set of three words or shapes or symbols with a blank. Participants were required to select the missing item from a list of five alternatives. Scores from these two set of tests correlate strongly with scores on an IQ-type test used for secondary school selection (r =0.93, Douglas, 1964) suggesting a high degree of validity.

3. Educational Qualifications At age 33, participants were asked about their highest academic or vocational qualifications. Responses are coded to the six-point scale of National Vocational Qualifications levels (NVQ) ranging from ‘none’ to ‘higher degree level’: 0= no qualifications; 1=some qualifications [Certificate of Secondary Education Grades 2 to 5]; 2= O level [equivalent to qualifications taken at the end of compulsory schooling]; 3= A level [equivalent to university entrance level qualifications]; 4=postsecondary degree/diploma and equivalent; and 5=higher post-graduate degrees and equivalent.

4. Personality Traits Personality traits were assessed at age 50, by the 50 questions from the International Personality Item Pool (IPIP)(Goldberg, 1999). Responses (5-point, from “Strongly Agree” to “Strongly Disagree”) are summed to provide scores on the so called ‘Big-5’ personality traits: Extraversion, Emotionality/neuroticism, Conscientiousness, Agreeableness and Openness. Scores on each trait range between 5 and 50 with higher scores equating to higher levels of each trait, of which 10 items for each trait. A preliminary test showed that the associations between traits Extraversion and Agreeableness were not significantly associated with adult savings and investment, thus these two traits were excluded from the following analyses. Preliminary analysis which included these two traits in the SEM model demonstrated that they were not moderator variables. Alpha was 0.88 for emotionality/neuroticism, 0.77 for conscientiousness, and 0.79 for intellect/openness.

5. Occupational Prestige Data on current or last occupation held by cohort members at age 50 are coded according to the RGSC described above, using a 6-point classification.

6. Financial Assessment was assessed at age 50. Participants were asked to assess their personal financial situation on a 5-point measure (1 =Finding it very difficult, 2= Finding it quite difficult, 3=Just about getting by, 4=Doing all right, 5= Living comfortably).

7. Adult Savings and Investment. At age 50, participants provided information on the amount of savings and investment they had, which were logged in the following analyses. In addition, participants also mentioned the specific types of their savings and investment, of which bank or building society =70.2%, ISA=51.8%, premium bonds= 35.0%, stocks and/or other shares =32.9%.

All this is fine, though self-report on wealth could be a problem. Asking people about their wealth can be a tricky business in the UK. If anything, people might be tempted to downplay it to avoid any tax enquiries. One simple control would have been to study the postcodes of participants, from which it is easy to get wealth estimates. Worth doing as a further measure of the accuracy of self-reports.

The verbal and non-verbal scores could have been added for an overall ability estimate, which might have had more predictive power.

The strongest association was between personal financial assessment and adult savings and investment, followed by education and occupation. This is a well-established finding. However, what was particularly interesting was the correlation between IQ measured as age 11 and savings measured 39 years later.

To my eye the correlations are very low, so the effect sizes are small. In partial defence of this observation, they are a bit larger than the effects of paternal social class at birth, and even of parental educational levels at birth, both of which have long been touted as the major determinants in life. Not so. Particularly in the structured equation modelling, the effects of intelligence measured in childhood are apparent.

Although wealth is a good measure of real-life success, in the United Kingdom there is the considerable blunting effect of income tax, which reduces the ability to save. For example, using Office of National Statistics figures the Institute of Financial Studies has just shown that:

A unique IFS analysis of HMRC tax records reveals 43 per cent of adults do not pay income tax, up from 38 per cent in 2010. By contrast, the top one per cent of earners are now paying 27 per cent of the nation’s income tax.

As Thomas Sowell has always pointed out, and the Institute of Fiscal Studies figures now show, membership of the top 0ne Percent is pretty fluid:

Only around three-quarters of people in the top one per cent in one year will be there in the next year, while only half will still be in the top one % in five years.

As a result, someone has a much higher chance of being in the top 1% at some point in their lives than they do in any given year. 3.4% of all people (and 5.5% of men) born in 1963 were in the top 1% of income tax payers at some point between 2000–01 and 2015–16.

In fact, among the high earning top 1% only 6% of income comes from non-work-related earnings, showing that accumulated wealth is a minor addition to work age incomes.

The authors, having shown the correlations and done some structured equation modelling say:

It demonstrated that childhood intelligence, much more than adult personality is a predictor of saving forty years later. Although related to Conscientiousness, it was how intelligence effected education and also occupation (social class) that explains its importance in adult financial success.

Intelligence leads to higher scholastic attainments which lead to a higher propensity to save. The effect is small, but it is larger than social class of origin. The effect might possibly be reduced by sharply redistributive taxation. Overall, intelligent kids go on to being good savers.