Price Clustering, Preferences for Round Prices, and Expected Returns

Abstract

Researchers have documented that individuals have a strong penchant for round numbers in a variety of settings, such as trading in financial markets. Beginning as early as the 1930s, empirical research has shown that security prices tend to cluster on round increments. This anomalous finding has persisted over time and in a wide range of different types of securities markets. We examine whether stocks with greater price clustering experience excess demand and consequently, negative return premia. Using a variety of traditional asset pricing tests, we find support for this argument as price clustering is associated with a robust, negative return premium. Our results are robust to transaction-level clustering, cross-sectional regressions, and multi-factor models.

Keywords:

• Price clustering
• Expected returns
• Behavioral economics
• Behavioral finance
• Round numbers

Notes

1 See e.g., Niederhoffer (1965, 1966), Niederhoffer and Osborne (1966), Harris (1991), Christie and Shultz (1994), Christie, Harris, and Schultz (1994), Gwilym, Clare, and Thomas (1998a, 1998b), Sopranzetti and Datar (2002), Gwilym and Alibo (2003), Ni, Pearson, and Poteshman (2005), and Sonnemans (2006), among others.

2 Clustering has been shown to exist in commodity markets (Ball, Torous, and Tschoegl 1985), bond markets (Gwilym, Clare, and Thomas 1998a), money markets (Sopranzetti and Datar 2002), and derivative markets (Gwilym, Clare, Thomas 1998b; Gwilym and Alibo 2003; Ni, Pearson, and Poteshman 2005).

3 Admittedly, other researchers have examined how price clustering might affect trading strategies. For instance, Bhattacharya, Holden, and Jacobsen (2012) acknowledge the presence of price clustering on round prices (or pricing increments of $0.05) and examine a trading strategy where the investor buys at prices on integers one penny below round increments and sell at prices on integers on one penny above round pricing increments. Intraday trading strategies at the 24-hour interval yield mixed results regarding the profitability of these strategies.

4 In unreported Fama and MacBeth (1973) cross-sectional tests, we find that price clustering leads to higher future stock prices. These results support the notion that the demand for round numbers places upward pressure (“pushes up”) on stock prices.

5 For the TAQ sample, we remove stocks with average intraday prices of less than $1 due to inactive trading, which creates inconsistencies in the clustering measures.