Denmark paid teenagers real money to reveal their patience, then watched it pick their degree

The coefficient from Claus Thustrup Kreiner is the one to sit with: applicants ranked 1 to 100 for patience, with 10 ranks of movement associated with almost 1.5 more years of education ^[11]. Read straight across the whole scale, that implies about 15 years between the least and most patient applicant ^[1], which no realistic menu of programs can supply. The two programs the write-up uses to illustrate its point, five-year economics and three-year physiotherapy, sit two years apart ^[2]. So the relationship is either local to part of the distribution, or the linear phrasing is flattering an estimate that bends. The account does not say which, and that matters to anyone who wants to use the number rather than admire it.

The earnings side is where sorting compounds. Economics applicants in the sample go on to average annual earnings growth of 11.5% between ages 20 and 45 ^[13]; physiotherapy applicants, 6.3% ^[14]. Run both across the 25 years the window covers and the first path multiplies starting earnings roughly 15-fold against roughly 4.6-fold, a ratio of about 3.3 ^[3]. Growth rates are not levels, so this says nothing about salaries at 45. It says how much distance one step across the program menu accumulates when the step is held for a career.

Risk appetite sorts on a different axis. More risk-averse participants were likelier to apply to programs with more predictable graduate earnings, meaning both a lower chance of ending up with low earnings and a narrower gap between the top and bottom earners from that program ^[12]. Physiotherapy carries a 7.1% low-earnings risk against economics at 3.4% ^[13][14], about twice as high ^[4]. The two measured traits are therefore not one ambition factor under two names: patience tracks duration and slope, risk aversion tracks spread.

Then there is the size of the lever. Kristoffer B. Hvidberg's framing is that theory has long predicted preferences shape educational investment, and the novelty is measuring it directly against the programs people actually applied for ^[2][3]. The experiment paid one randomly selected decision per person, averaging around DKK 250 by MobilePay ^[6], so choices over a few hundred kroner are being asked to predict a choice worth decades. Soren Leth-Petersen's defence is that the incentive is exactly what separates a genuine read on patience from a survey correlation ^[7]. The harder evidence is the control set: with school grades and parental education and income included, the authors report surprise at how much explanatory power the two preference measures retained ^[15]. That residual is the finding. The rest is infrastructure built to make it believable.