Page 49
4 – Backtesting Performance
In this section we will assess whether the quantitative terminal value measures described above can successfully carry out one of the most important tasks of equity analysts; provide profitable investment recommendations. To test this, we will compare the models to various benchmarks and control for several common risk factors to find out if the investment strategies are taking on more risk to achieve superior returns or if they provide true alpha (Pedersen, 2015, p. 29).
Page 50 Sector exposure
We were curious to see whether our valuation models favored some sectors over others. This analysis is based on Morningstar sectors (not Global Industry Classification Standard (GICS)), so communication services do not include recent technology additions such as Facebook, Alphabet, and Netflix. Table 4.2 illustrates the sector exposures of the strategies relative to the total sample. This analysis is performed by calculating the total trades in a sector for a specific strategy and dividing it by the strategy’s total trades. This number is compared to the sector exposure of the S&P 500 adjusted. The adjusted benchmark is equal weighted, so the sector weights are different than what you would see in the classic S&P 500 index that has larger exposures to a few large tech-names. The formula applied is illustrated below:
𝐿𝑌 𝐹𝐶𝐹𝐹𝑇𝑒𝑐ℎ 𝑒𝑥𝑝= 𝑇𝑜𝑡𝑎𝑙 𝑡𝑟𝑎𝑑𝑒𝑠 𝑖𝑛 𝑇𝑒𝑐ℎ𝐿𝑌 𝐹𝐶𝐹𝐹
𝑇𝑜𝑡𝑎𝑙 𝑡𝑟𝑎𝑑𝑒𝑠𝐿𝑌 𝐹𝐶𝐹𝐹 − 𝑁𝑜. 𝑜𝑓 𝑇𝑒𝑐ℎ 𝑠𝑡𝑜𝑐𝑘𝑠𝑆&𝑃500 𝑎𝑑𝑗.
𝑇𝑜𝑡𝑎𝑙 𝑛𝑜. 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘𝑠 𝑆&𝑃500 𝑎𝑑𝑗.
= −2%
Tech: technology companies Exp: technology sector exposure
A value of -2% indicates that the portfolio (on average) includes 2% fewer technology stocks than the benchmark in the 15-year period.
Table 4.2: Relative sector exposure of the Gordon Growth strategies
LY FCFF 3Y Norm 5Y Norm 10Y Norm 3Y avg. 5Y avg. 10Y avg.
Technology -2% -2% -2% -2% -1% 0% 2%
Consumer Cyclical 0% 0% -1% -1% 1% 1% 4%
Healthcare 2% 1% 2% 4% 1% 2% 3%
Energy -2% -3% -3% -3% -3% -3% -3%
Communication Services 0% 0% 0% -1% 0% -1% -2%
Consumer Defensive 5% 7% 7% 7% 6% 6% 6%
Industrials 1% 1% 0% 0% 0% -2% -3%
Basic Materials 0% 0% 0% 0% 0% 0% -1%
Utilities -1% -1% -1% -1% -1% -1% -2%
Real Estate -2% -2% -3% -3% -3% -3% -3%
Sector exposure for Gordons Growth strategies. The larger the green bar, the larger the overweight in a sector compared to the S&P500 adj. index and the opposite for the red bars.
Weighting: Equal weighted and monthly rebalancing Growth: 3.95%
WACC: Morningstar sample test
Source: Morningstar direct, Kenneth French database and own estimations.
Market & period: 2003.04-2018.09, S&P 500 excluding financials and duplicates
Page 51 Table 4.2 illustrates that all the Gordon Growth strategies overweight stocks in less cyclical sectors such as consumer defensive and healthcare. The strategies typically overweight consumer defensives with at least 5% and healthcare with at least 1%. If we calculate the total sector weights, the largest exposure is to consumer cyclicals - simply because this sector has taken up most space in the index during the period (Appendix 4).
Table 4.3 illustrates the fundamentals in the different sectors to get an understanding of why the strategies over- or underweight a sector. Each month, we calculate the average EBIT margin, revenue growth, etc. for every stock in the S&P 500 excluding financials based on the numbers of the last fiscal year with a 2-month lag. We do this for each of the 10 sectors. We then take a median of the monthly observations from 2003 to 2018 to eliminate outliers.
Table 4.3: Sector fundamentals
Book/
Market Sales/
Market
FCFF/
EV
EBITDA/
EV ROIC EBITDA-margin
EBIT-margin
NIBD/
equity E/P
1Y revenue growth
Technology 0.32 0.36 0.04 0.07 16% 19% 12% -0.29 0.03 7%
Consumer Cyclical 0.32 0.88 0.04 0.10 13% 14% 10% 0.46 0.05 7%
Healthcare 0.28 0.35 0.02 0.07 17% 22% 17% 0.19 0.04 6%
Energy 0.48 0.59 0.00 0.11 9% 25% 17% 0.38 0.04 7%
Communication Services 0.35 0.73 0.04 0.13 9% 30% 17% 0.75 0.06 6%
Consumer Defensive 0.25 0.61 0.03 0.09 17% 18% 14% 0.52 0.05 6%
Industrials 0.31 0.73 0.03 0.10 14% 16% 12% 0.46 0.05 7%
Basic Materials 0.40 0.81 0.03 0.11 13% 18% 12% 0.47 0.04 7%
Utilities 0.61 0.80 0.01 0.11 6% 28% 18% 1.31 0.05 8%
Real Estate 0.42 0.11 0.00 0.05 5% 63% 30% 1.13 0.03 8%
Different fundamentals on a sector level.
Source: Morningstar Direct and own estimations.
Period: 2003.04-2018.09
Table 4.3 indicates why the portfolios favor consumer defensives, as the sector has large free cash flow yields (FCFF to enterprise value). Combined with a low Morningstar WACC, stocks in this sector are set up for high valuations in the models.
The healthcare sector is overweighted across strategies during the period, but healthcare has not appeared particularly cheap during the period. We must again turn towards the low sector WACC of 7%, which is the second lowest in our Morningstar sample. With a lower WACC, valuations increase, so this can possibly explain our exposure to healthcare. This illustrates one of the benefits of using the single-period valuation model instead of relying on simple price multiples such as earnings/price and book/market that do not adjust for risk (cost of capital). The large healthcare stocks in the S&P 500 might not look cheap on multiples, but many of them have sustainable competitive advantages (moats if you ask Morningstar), and their business models are stable across market cycles - factors that contribute to lower risk and lower cost of capital.
Page 52 Because our valuations apply the lower WACC of healthcare, they have increased their exposure to the best performing sector in the period. Later in this section, we will investigate how stressing our assumptions to WACC will change the sector exposures.
When we base our estimate of steady-state FCFF on a longer historical average of free cash flow, we get more exposure to consumer cyclicals and technology which have been strong performers in the 15-year period. This is interesting because the 3-, 5-, and 10-year average free cash flow portfolios all have the highest Sharpe ratios above 1. The 10Y average strategy has the highest exposure towards consumer cyclicals and the technology sector relative to the adjusted S&P 500.
Since our portfolios have materially different sector exposures compared to the index, they provide abnormal returns from both allocation and stock selection. They gain from allocation by overweighting high-return sectors such as healthcare while underweighting sectors such as energy, real estate, and utilities with lower returns. The portfolios gain from stock selection as they pick the most undervalued stocks within each sector.
The number of undervalued stocks vary over time
Appendix 6 illustrates the average number of stocks in our portfolios in each month from 2003