Sensation Seeking and Overconfidence Drive Investor Trading Frequency

Sensation seeking is the tendency to seek novel, varied, and intense experiences, including activities with financial or physical risk. In "Sensation Seeking, Overconfidence, and Trading Activity," Grinblatt and Keloharju (2009) linked five Finnish administrative databases for 95,804 investors. From July 1997 to November 2002, each additional speeding ticket raised trade count by 9.8% and trading probability by 4.7%. Overconfidence — derived from psychological exams of 12,379 males at approximately age 19 or 20 — also predicted trading frequency.

What the Study Found

Each additional speeding ticket increased annualized portfolio turnover by approximately 3.6%, from a 40% annual baseline for active traders. Income-related fines and flat fines had nearly identical effects: each raised trade count by 10.1% and 9.7%, respectively. Buy-trade and sell-trade probit coefficients were 0.045 (t = 5.75) and 0.053 (t = 6.78). Within the male subsample (N = 11,521 for the trade decision regression), overconfidence was significant at the 5% level. Overconfidence was a marginally insignificant predictor of turnover in the 7,359-investor subsample, while sensation seeking remained highly significant.

Methodology

The study linked five Finnish administrative datasets covering stock trading records, vehicle violations, military psychological profiles, and tax data. The main sample included 95,804 individual investors in the provinces of Uusimaa and East Uusimaa. Trading was measured from July 1, 1997 to November 29, 2002; speeding tickets were recorded from July 1, 1997 to December 31, 2001. Controls included income decile dummies, financial wealth decile dummies, age and birth year dummies, gender, marital status, employment status, homeownership, and number of stocks.

Key Statistics

Metric	Finding	Context
Effect per speeding ticket on probability of trading	+4.7%	Full sample, probit regression
Effect per speeding ticket on number of trades	+9.8%	Full sample, Heckman two-stage
Effect per speeding ticket on annualized turnover	+3.6% per year	Full sample, OLS regression
Effect per income-related fine on number of trades	+10.1%	Full sample, Heckman two-stage
Effect per flat fine on number of trades	+9.7%	Full sample, Heckman two-stage
Fraction of sample that traded stocks	55.7%	Full sample, July 1997–November 2002
Mean monthly portfolio turnover	0.019	Full sample, July 1997–November 2002

Why This Matters

Prior tests of the overconfidence-trading link used gender as an indirect proxy, making it difficult to separate overconfidence from other gender-related traits. Measuring overconfidence through a military psychological exam taken years before trading began suggests the trait has lasting effects on financial behavior. Income-linked fines in Finland create higher financial stakes for speeding than flat fines, strengthening the causal case for sensation seeking as a trading driver. Gender differences in trading frequency, widely attributed to male overconfidence, may owe more to the stronger sensation-seeking tendencies of males.

Frequently Asked Questions

Each additional speeding ticket raised the number of trades by 9.8% and the probability of making any trade at all by 4.7%. The effect held equally for buying (coefficient 0.045) and selling (coefficient 0.053). Grinblatt and Keloharju argue this buy-sell symmetry dispels the notion that the results are driven by asset sales to finance speeding fines.

Within the male subsample of 7,359 investors, sensation seeking was the stronger predictor of portfolio turnover. The sensation seeking coefficient was highly significant, while the overconfidence coefficient was marginally insignificant in the turnover regression. Grinblatt and Keloharju write that the male-female turnover gap documented by Barber and Odean (2001) may reflect sensation seeking differences as much as overconfidence differences.

Overconfidence was derived from a self-confidence scale on a compulsory Finnish military psychological test, administered at approximately age 19 or 20. The authors regressed self-confidence on intellectual ability, income, wealth, marital status, and other controls; the residual was the overconfidence measure. The test predated the trading sample period by many years, covering a sample of 12,379 male investors.

95,804 individual investors in the Finnish provinces of Uusimaa and East Uusimaa form the core sample. Records covered July 1, 1997 through November 29, 2002 for trading and July 1, 1997 through December 31, 2001 for speeding violations. Five administrative databases were linked: Finnish brokerage records, stock prices, driving records, military psychological profiles, and tax data.