Price and Earnings Momentum, Transaction Costs, and an Innovative Practitioner Technique

• Date: 01 December 2015
• DOI: http://doi.org/10.1111/irfi.12065
• Author: Helen Roberts,Reza Tajaddini,Timothy Falcon Crack
• Published date: 01 December 2015

Price and Earnings Momentum,

Transaction Costs, and an

Innovative Practitioner Technique*

REZA TAJADDINI†,TIMOTHY FALCON CRACK‡AND HELEN ROBERTS‡

†Accounting, Economics and Finance, Faculty of Business and Law, Swinburne

University of Technology, Melbourne, Vic., Australia and

‡Accountancy and Finance, University of Otago, Dunedin, New Zealand

ABSTRACT

We use an innovative practitioner technique to investigate the interplay

between the ex post performance of momentum strategies and transaction

costs, rebalancing frequency, turnover constraints, and fund size. We have

three interrelated main results: first, the level of and correlation between

active returns to price momentum and earnings momentum strategies vary

dramatically with these factors; second, strategies that are fearful of ex ante

transaction costs generate returns net of transaction costs that are far superior

to the net returns of naive strategies; and third, obtaining better traction with

the unique elements of each strategy yields a more profitable combined

strategy.

JEL classification: G11; G14; G15

I. INTRODUCTION

This paper is part of a recent and growing academic literature that emphasizes

the importance of weighing returns and transaction costs as part of the portfolio

formation process, especially in active trading strategies (Trethewey and Crack

2010; Frazzini et al. 2012; Garleanu and Pedersen 2013). Two of the most

popular active strategies in equities are price momentum (hereafter MOM) and

earnings momentum (i.e., post-earnings announcement drift, hereafter PEAD)

(Jegadeesh and Titman 1993, 2011; Fama and French 2012). We use a quanti-

tative practitioner technique (Grinold and Kahn 2000) to analyze the perfor-

mance of MOM and PEAD equity trading strategies in New Zealand from 1999

* We thank the New Zealand Stock Exchange (NZX) for supplying analyst forecast data, NZSE 40

and NZX 50 index weights and advice about NZX data issues. We thank Scott Chaput and Mark

Tippett for helpful conversations. We also thank seminar participants at Otago University; partici-

pants at the 25th Australasian Finance and Banking Conference, Sydney, Australia, December 2012;

participants at the 17th annual New Zealand Finance Colloquium, Dunedin, New Zealand, February

2013; and anonymous referees. Any errors are our own.

bs_bs_banner

International Review of Finance, 15:4, 2015: pp. 555–597

DOI: 10.1111/irfi.12065

© 2015 International Review of Finance Ltd. 2015

to 2011.1Although our technique pays particular attention to transaction costs,

its practitioner roots mean it has appeared only once previously in the academic

literature (Trethewey and Crack 2010). We apply this technique to momentum

strategies, but it can be used for any active strategy that tilts a portfolio toward

stock characteristics associated with outperformance.

Classic finance theories (e.g., the efficient market hypothesis and modern

portfolio theory) usually ignore the impact of actual transaction costs on the

behavior and performance of trading strategies. Conventional academic tech-

niques, therefore, often blindly chase returns and account for transaction costs

and risk only after the fact. Our practitioner technique, however, is superior to

such conventional techniques because it simultaneously weighs returns, risk,

and transaction costs at each portfolio rebalance.

The superiority of our technique is simple to demonstrate. For example, we

find that a simple combined MOM–PEAD strategy beats the benchmark by an

impressive 18% per annum before transaction costs over our sample period

(consistent with previous research in New Zealand). If, however, we take the

trade list generated by that strategy and execute it in a realistically sized

portfolio, the strategy underperforms the benchmark by more than 10% per

annum after we account for bid-ask spreads and price impact. A more intelligent

strategy that is fearful2of an ex ante price impact, however, produces a conser-

vatively estimated outperformance of 92 basis points (bps) per annum after

transaction costs. The key is that, like a practitioner, our intelligent strategy can,

for example, avoid overweighting a small stock with attractive momentum

characteristics if it has unattractive ex ante transaction costs.

At first glance, 92 bps of outperformance may appear small compared with

the 18% per annum number we quoted, but the difference between the two is

that the 18% number is unattainable (if you chase it, you underperform the

benchmark by more than 10% per annum), whereas the 92 bps per annum is a

conservative estimate of genuine outperformance by a practitioner strategy net

of transaction costs. This outperformance is enough to add value.3In fiercely

1 Quantitative techniques are popular. The TABB Group estimated that USD 3.47 trillion

(roughly 34%) of US equity assets, including both passive and active strategies, were quanti-

tatively managed in 2009 (Hua 2009).

2 When we say that a strategy is ‘fearful’ of transaction costs, we mean that we have created a

trading strategy that mimics the behavior of a portfolio manager whose clients are fearful of

the biting impact of transaction costs. So, the strategy/manager weighs returns, risk, and

transaction costs at each portfolio rebalance to avoid chasing returns without regard to active

risk or ex ante transaction costs.

3 Note that an extra 92 bps per annum attained in a tax-deferred retirement savings scheme

improves the ending value of a lump sum deposit by about half as much again over a 45-year

horizon. Many conservative active quantitative funds (sometimes called ‘enhanced index

funds’ or ‘alpha tilt index funds’) are looking for no more than a couple of hundred bps of

outperformance on a benchmark per annum from a combination of perhaps 5–10 strategies

(Grinold and Kahn 2000, chapter 5; Loftus 2000, exhibit 2; Chincarini and Kim 2006, chapter

9; Lejeune and Samatli-Paç 2012). So, getting in excess of 90 bps from a paired strategy is very

attractive.

International Review of Finance

556 © 2015 International Review of Finance Ltd. 2015

competitive capital markets, it would be naïve to expect much more than this

(Berk and Green 2004; Berk 2005; Jones and Wermers 2011); the amazing

returns we find by switching off our strategy’s fear of an ex ante price impact

cannot be attained in competitive markets.

We think that our strategy probably performs best in slightly less liquid

markets where transaction costs are not already very low and potential anoma-

lies may still exist. To demonstrate this, we selected the New Zealand stock

market, which is an open market with freely flowing information and capital

flows, and relatively efficient, but still sufficiently overlooked on the interna-

tional scene that anomalies exist. A direction for future research is to apply our

strategy to small stocks in the United States. For example, Edelen et al. (2013)

emphasize the detrimental effect of transaction costs, especially price impact,

on mutual fund performance in US small-capitalization stock funds. If a

genuine momentum anomaly exists, then, through intelligent trading and

transaction cost avoidance techniques, we hope to be able to extract improved

profits.

Our technique is sophisticated enough that it allows us to simultaneously

account for or control many parameters in our strategies: bid-ask spreads and

price impact, index adds and drops, different fund sizes, different portfolio

rebalancing frequencies, the beta of the strategy, different levels of constraint on

turnover, etc. As a direct consequence, we have been able to obtain brand new

insights into the interplay between these parameters and the inter-relationship

between the active returns to the MOM and PEAD strategies, and the ultimate

performance of the strategies. For example, with the same ex ante alphas, we are

able to produce correlations between active daily returns to MOM and PEAD

strategies that vary from as much as 93% to as little as 13%, depending on the

parameters of the implementation (and the lower the correlation, the more

there is to be gained by combining MOM and PEAD strategies). This shows that

a standard static implementation that reports the correlation between returns to

MOM and PEAD strategies, and makes assertions about the overlap between the

two, can be misleading.

We find that the alphas to a combined MOM–PEAD strategy get a boost

during both the 2001 and 2009 market crisis periods. When examined indi-

vidually, however, we find a good boost to MOM alphas (whereas PEAD alphas

are boosted only slightly or hurt) during the 2001 market crisis period, and a

good boost to PEAD alphas (whereas MOM alphas are hurt) during the 2007–

2009 market crisis period.

We make a number of other contributions. For example, Trethewey and

Crack (2010) implement the Grinold and Kahn (2000) practitioner technique in

a simple fashion, but we push well beyond their implementation using daily

data instead of monthly data, MOM and PEAD instead of just MOM, and a

corrected measure of transaction costs.4In addition, we appear to be the first to

4 Using daily data, rather than monthly data, has many advantages: We can account for index

adds and drops in real time; we have the ability to rebalance our portfolios at the daily

Momentum and an Innovative Practitioner Technique

557© 2015 International Review of Finance Ltd. 2015