Original paper
Abstract
We document that several well known asset-pricing implications of accruals differ for investment and non-investment-related components. Exposure to an investment-accruals factor explains the cross-section of returns better than the accruals themselves, and this factor’s returns are negatively predicted by sentiment. The opposite results hold for non-investment accruals. Further tests show cash profitability only subsumes long-term non-investment accruals in the cross-section of returns and economy-wide investment accruals negatively predict stock-market returns while other accruals do not. These results challenge existing accruals-anomaly theories and help resolve mixed evidence by showing that the anomaly is two separate phenomena: a risk-based investment accruals premium and a mispricing of non-investment accruals.
Keywords:Â Accruals Anomaly, Profitability, Real Investment, Cross-section of Stock Returns
Trading rules
- Investment Universe: NYSE, AMEX, and NASDAQ stocks
- Calculate Balance Sheet Accruals (BS_ACC): BS_ACC = (∆CA - ∆Cash) - (∆CL - ∆STD - ∆ITP) - Dep
- ∆CA: Annual change in current assets
- ∆Cash: Change in cash and cash equivalents
- ∆CL: Change in current liabilities
- ∆STD: Change in debt within current liabilities
- ∆ITP: Change in income taxes payable
- Dep: Annual depreciation and amortization expense
- Sort stocks into ten groups using accruals as the criterion.
- Take long positions in stocks with the minimal accruals and short those with the maximal accruals.
- Adjust the portfolio each year in May, after the announcement of earnings.
Python code
Backtrader
import backtrader as bt
class AccrualAnomaly(bt.Strategy):
def __init__(self):
self.accruals = {}
def next(self):
if self.data.datetime.date().month != 5: # Rebalance only in May
return
self.accruals.clear()
for data in self.datas:
ca = data.current_assets
cash = data.cash_equivalents
cl = data.current_liabilities
std = data.short_term_debt
itp = data.income_taxes_payable
dep = data.depreciation_amortization_expense
delta_ca = ca - ca(-1)
delta_cash = cash - cash(-1)
delta_cl = cl - cl(-1)
delta_std = std - std(-1)
delta_itp = itp - itp(-1)
bs_acc = (delta_ca - delta_cash) - (delta_cl - delta_std - delta_itp) - dep
self.accruals[data] = bs_acc
sorted_stocks = sorted(self.datas, key=lambda d: self.accruals[d])
low_decile = int(len(sorted_stocks) * 0.1)
high_decile = int(len(sorted_stocks) * 0.9)
long_stocks = sorted_stocks[:low_decile]
short_stocks = sorted_stocks[high_decile:]
self.sell_stocks(short_stocks)
self.buy_stocks(long_stocks)
def sell_stocks(self, short_stocks):
for stock in short_stocks:
self.sell(data=stock)
def buy_stocks(self, long_stocks):
for stock in long_stocks:
self.buy(data=stock)
if __name__ == '__main__':
cerebro = bt.Cerebro()
cerebro.addstrategy(AccruralAnomaly)
# Add your data feeds for NYSE, AMEX, and NASDAQ stocks here
cerebro.run()This code snippet provides a basic implementation of the Accrual Anomaly trading strategy in Backtrader. You’ll need to add your data feeds for NYSE, AMEX, and NASDAQ stocks to complete the implementation.