Original paper

SSRN-id3178186.pdf381.2KB

Abstract

In this paper, after controlling for the level of R&D expenditures, I find that profitability of R&D intensive firms is more important for subsequent returns than the R&D intensity (measured with R&D-to-market value or R&D-to-assets) and past performance. In a sample of firms where I am able to compute abnormal returns around quarterly earnings announcements R&D-to-market value variable becomes insignificant even when not controlling for the firm’s profitability. Since quarterly profits are reported by big firms this suggests that R&D anomaly is driven by small stocks. Big firms with high R&D-to-market value earn lower not higher subsequent returns, once I control for the level of R&D expenditures and firm’s profitability. The previously recorded anomaly, namely that R&D intensive firms earn positive abnormal returns, seems to be driven by the level of R&D investment, its profitability and small stocks.

Keywords: cross-section of stock returns, R&D expenditures, profitability, small stocks, five-factor asset pricing model, earnings momentum, Fama MacBeth regression

Trading rules

Target stocks: Focus on shares from NYSE, AMEX, and NASDAQ priced over $5.
Calculate “Ability” measure yearly for each firm

Measure firm’s effectiveness at converting R&D expenditures into sales growth
Use rolling firm-level regressions on sales growth with lagged R&D for 5 different lags
Average the five R&D regression coefficients to obtain “Ability”

Create a long-short equity portfolio

Sort stocks into quintiles based on “Ability” and three tiers based on R&D expenditures (low 30%, middle 40%, high 30%)
Go long on stocks with high “Ability” (top quintile) and large R&D spending (top 30%)
Short stocks with low “Ability” (bottom quintile) and large R&D spending (top 30%)

Ensure even distribution of stocks in the portfolio and adjust its composition annually.

Python code

Backtrader

import backtrader as bt
import pandas as pd
import numpy as np

class AbilityStrategy(bt.Strategy):
    def __init__(self):
        pass

    def next(self):
        pass

    def ability_measure(self, data):
        pass

    def rebalance_portfolio(self):
        pass

class NYSEAboveFive(bt.feeds.PandasData):
    lines = ('ability',)
    params = (
        ('datetime', None),
        ('open', -1),
        ('high', -1),
        ('low', -1),
        ('close', -1),
        ('volume', -1),
        ('openinterest', -1),
        ('ability', -1),
    )

def preprocess_data(data):
    pass

if __name__ == '__main__':
    cerebro = bt.Cerebro()

    data = pd.read_csv('your_data_file.csv')
    data = preprocess_data(data)

    data_feed = NYSEAboveFive(dataname=data)
    cerebro.adddata(data_feed)

    cerebro.addstrategy(AbilityStrategy)
    cerebro.addsizer(bt.sizers.EqualWeight)

    cerebro.broker.set_cash(100000)
    cerebro.broker.setcommission(commission=0.001)

    print('Starting Portfolio Value:', cerebro.broker.getvalue())
    cerebro.run()
    print('Ending Portfolio Value:', cerebro.broker.getvalue())

Please note that this code is just a template to get you started, and you will need to fill in the preprocess_data, ability_measure, and rebalance_portfolio functions with the specific logic to calculate the “Ability” measure, preprocess the data, and rebalance the portfolio according to the trading rules provided. Additionally, you’ll need to have a dataset containing stock data with the required information (prices, R&D expenditures, sales growth) to run this code.

Profitability, R&D Investments and the Cross-Section of Stock Returns

Original paper

Abstract

Trading rules

Python code

Backtrader