Original paper
Abstract
Top management team diversity matters for stock returns. We develop a new text–based measure of team diversity and apply it to a sample of over 40,000 top executives in U.S. firms from 2001 to 2014. Buying firms with diverse teams and selling firms with homogenous teams — a strategy we call “diversity investing” — outperforms leading asset pricing anomalies over our sample period on a value-weighted basis. We examine a range of possible explanations and find strong evidence for the view that analysts and investors have downward-biased return expectations on firms with diverse teams, consistent with a mispricing explanation for diversity returns.
Keywords:Â Behavioral Finance, Top Management Teams, Anomalies, Diversity
Trading rules
- Target stocks: Those registered on NYSE, AMEX, and NASDAQ having a positive book value and a history of returns for at least a year in the CRSP Database.
- Source for company details: Yearly reports from the EDGAR database, with a preference for 10-K, 10KSB, DEF 14A.
- Diversity evaluation: Refer to Equations (1) and (2) on page 7; For companies with only a single top-level executive, set D = 0.
- Buy companies in the highest quintile based on diverse leadership.
- Sell companies in the lowest quintile with (firms with uniform leadership).
- Portfolio adjustments: Carried out yearly.
- Duration for each position: 12 months.
- Portfolio weighting: Value-weighted stocks.
Python code
Backtrader
Here’s the Backtrader Python code for the given trading rules:
import backtrader as bt
class DiversityStrategy(bt.Strategy):
def __init__(self):
self.diversity = self.datas[0].diversity
def next(self):
long_positions = []
short_positions = []
for data in self.datas:
if len(data) < 12 or data.book_value[0] <= 0:
continue
if data.diversity[0] >= data.diversity.get(size=5)[0]:
long_positions.append(data)
elif data.diversity[0] <= data.diversity.get(size=5)[-1]:
short_positions.append(data)
long_weight = 1.0 / len(long_positions)
short_weight = -1.0 / len(short_positions)
for data in long_positions:
self.order_target_percent(data=data, target=long_weight)
for data in short_positions:
self.order_target_percent(data=data, target=short_weight)
if __name__ == '__main__':
cerebro = bt.Cerebro()
cerebro.addstrategy(DiversityStrategy)
# Add data feeds from NYSE, AMEX, NASDAQ with biographical data from EDGAR (not shown)
# Use the provided diversity measure as an additional data field (not shown)
cerebro.run()Please note that this code assumes you’ve already prepared the data feeds with the required information from the NYSE, AMEX, and NASDAQ, and have calculated the diversity measure according to the formulas given in your question.