Measuring Herding Behavior: The Cross-Sectional Standard Deviation of Returns

Detecting and measuring herding behavior is a significant challenge for financial market participants. One of the most widely used methods is the Cross-Sectional Standard Deviation (CSSD) of returns, developed by Christie and Huang (1995). This article provides a comprehensive overview of the CSSD methodology, including its theoretical basis, calculation, and interpretation.

Theoretical Foundation

The CSSD method is based on the idea that during periods of market stress, individual asset returns will tend to cluster around the overall market return, as investors follow the herd. This will lead to a decrease in the cross-sectional dispersion of returns.

Conversely, in normal market conditions, individual asset returns are more likely to be driven by their own idiosyncratic factors, leading to a higher cross-sectional dispersion of returns.

The CSSD Formula:

The CSSD is calculated as follows:

CSSD_t = sqrt(sum_{i=1 to N} (R_{i,t} - R_{m,t})^2 / (N-1))_

Where:

• CSSD_t is the Cross-Sectional Standard Deviation of returns at time t.
• R_{i,t} is the return of asset i at time t.
• R_{m,t} is the average return of all assets in the sample at time t.
• N is the number of assets in the sample.

Step-by-Step Calculation Example

Let's calculate the CSSD for a portfolio of three stocks on a particular day.

Step 1: Calculate the average return (R_{m,t})_

R_{m,t} = (2.5% + 1.0% + 3.0%) / 3 = 2.17%_

Step 2: Calculate the squared deviations from the average return

• Stock A: (2.5% - 2.17%)^2 = 0.001089
• Stock B: (1.0% - 2.17%)^2 = 0.013689
• Stock C: (3.0% - 2.17%)^2 = 0.006889

Step 3: Sum the squared deviations

Sum of squared deviations = 0.001089 + 0.013689 + 0.006889 = 0.021667

Step 4: Calculate the CSSD

CSSD_t = sqrt(0.021667 / (3-1)) = sqrt(0.0108335) = 0.1041 or 10.41%

Interpreting the Results

A lower CSSD value suggests a higher degree of herding, as individual asset returns are more tightly clustered around the market average. Conversely, a higher CSSD value suggests a lower degree of herding.

To formally test for the presence of herding, Christie and Huang (1995) propose a regression-based approach. They regress the CSSD on dummy variables representing extreme market movements.

The Regression Model:

CSSD_t = a + b_1 * D_{up,t} + b_2 * D_{down,t} + e_t

Where:

• D_{up,t} is a dummy variable that equals 1 if the market return is in the top x% of the distribution on day t, and 0 otherwise.
• D_{down,t} is a dummy variable that equals 1 if the market return is in the bottom x% of the distribution on day t, and 0 otherwise.

If the coefficients b_1 and b_2 are negative and statistically significant, it provides evidence of herding during periods of extreme market movements.

Actionable Example for Portfolio Managers

A portfolio manager can use the CSSD to monitor the level of herding in their portfolio and in the broader market. A sudden decrease in the CSSD could be a warning sign of increased systemic risk.

Practical Application:

1. Calculate the daily CSSD for the portfolio.
2. Establish a baseline or historical average for the CSSD.
3. Set up alerts for when the CSSD deviates significantly from the baseline.
4. When an alert is triggered, conduct a deeper analysis to understand the drivers of the increased herding.

By incorporating the CSSD into their risk management framework, portfolio managers can gain valuable insights into market dynamics and make more informed investment decisions.