Table 1.
Stability of Risk Preferences
| Paper | Population | Time | Corr | Sig | Inc |
|---|---|---|---|---|---|
| Menkhoff and Sakha (2014) | 384 rural Thai | 5 years | ? | yes | inc |
| Levin et al. (2007) | 124 US children/parents | 3 years | 0.20 – 0.38 | yes | inc |
| Guiso et al. (2011) | 666 Italian investors | 2 years | 0.131 | yes | hyp |
| Kimball et al. (2008) | 700 older Americans | 2 years | 0.27 | ? | hyp |
| Love and Robison (1984) | 23 US farmers | 2 years | −0.38 – 0.231 | no2 | hyp |
| Sahm (2012) | 12000 older Americans | multiple years | 0.18 | yes | hyp |
| Beauchamp et al. (2012) | 489 Swedish twins | 1 year | 0.483 | yes | hyp |
| Goldstein et al. (2008) | 75 Americans | 1 year | 0.43 | yes | hyp |
| Lönnqvist et al. (2014) | 43 German students | 1 year | 0.21 | no | inc |
| Smidts (1997) | 205 Dutch farmers | 1 year | 0.44 | yes | hyp |
| Wehrung et al. (1984) | 84 N. American businessmen | 1 year | 0.36 | yes | hyp |
| Andersen et al. (2008) | 97 Danes | 3– 17 months | ? | yes | inc |
| Harrison et al. (2005) | 31 US students | 6 months | ? | yes | inc |
| Vlaev et al. (2009) | 69 British students/adults | 3 months | 0.20–0.634 | yes | hyp |
| Horowitz (1992) | 66 US students & 23 PTA | 2 months | ? | no | inc |
| Wölbert and Riedl (2013) | 53 Dutch students | 5–10 weeks | 0.36–0.68 | yes | inc |
| Schoemaker and Hershey (1992) | 109 US MBA students | 3 weeks | 0.55 | yes | hyp |
| Hey (2001) | 53 British students | a few days | ? | yes | inc |
Corr - the correlation of risk preferences over time. Sig - whether risk preferences are significantly related over time. Inc - whether the experiment was incentivized rather than hypothetical.
1
Our own calculation, from the raw data reported in the original paper.
2
In fact, the negative correlation of −0.38 is significant at the 10% level.
3
This is a polychoric correlation, which may be larger because it suffers from less attenuation bias.
4
Interestingly, the one correlation which was not statistically significantly different than 0 (value of 0.20)was for the risk question which used the Multiple Price List (MPL) mechanism over gains. This is arguably the most common method for collecting risk preferences.