Quantitative Versus Qualitative Valuations

Page 105

Page 106 Studies indicate that analysts have a bias towards herding (Desai, Liang & Singh, 2000), as their recommendations tend to align with those of other analysts, which in turn affects the performance of the recommendations. If a new 5-star stock is already considered undervalued by the consensus of analysts, the market might already have been aware and bought into the stock (thus raising the price and lowering its subsequent returns).

Data mining and biases in backtests

A systematic approach helps reduce a trader’s own behavioral biases (Pedersen, 2015, p. 57), and since the quantitative valuations are based on realized historical financials - apart from our assumptions to WACC and growth - they are immune to most human biases. They are also ignorant to news flows and any guidance from the management of the firms being valued, which can be both an advantage and a disadvantage. On the one hand, having more information in a timely manner should give qualitative analysts an edge. On the other hand, we could argue that quarterly reporting and guidance may fixate markets on the short term, whereas our models are solely focused on estimating the long-term steady state. If the short-term trends of growth and margin improvements in a full-year guidance is extrapolated by analysts in their 5- or 10-year explicit forecasts, this may considerably affect not just the value in the explicit forecast period but also the terminal value which relies greatly on the fundamentals at the end of the forecast period (McKinsey, 2015, p. 250).

Applying an explicit forecast period (as most analysts do) or to exclude it (as our quantitative models do) can be valuable in some scenarios or prove unnecessary in other. Explicit forecast periods are necessary so the company can reach a steady-state level of growth and profitability before the analyst calculate the terminal value (McKinsey, 2015, p. 502). If a firm is already mature and has stable margins and low growth, we would argue there might not be a purpose for short-term forecasts, as these could likely exaggerate growth in accordance with the results of Stotz (2016). At the same time, we do experience that the undervalued and profitable investments identified by our quantitative models tend to be characterized by slower growth. The models have been less successful at systematically identifying and shorting overvalued stocks that provide negative alpha, and this could be because they simply lack the explicit forecast period necessary to correctly value immature and fast-growing stocks.

Quantitative models are exposed to several biases. Backtests typically look better than an actual trading strategy would in the real world because data mining and look-ahead bias can greatly influence results (Pedersen, 2015, p. 48). Although we have only applied past fundamentals from previous annual reports in the quantitative valuations, our assumptions for the cost of capital (WACC) are based on samples taken in 2018. We also apply an average growth rate of the U.S.

economy measured from 2003 to 2017 to every valuation. The realized growth rate and Morningstar’s WACC estimates of 2018 could not have been known back in 2003. Fortunately, the performance of our models is robust despite changes in our growth and WACC assumptions.

Page 107 As the applied sector costs of capital depend directly on the estimates of Morningstar, Bloomberg or NYU Stern’s Aswath Damodaran, this could present itself as a source of anchoring or herding, as our discount rates imitate those of other analysts. To accommodate this bias, and simultaneously improve the quantitative model’s ability to differentiate WACC between firms in the same sector, one could estimate WACC quantitatively for each individual stock.

Can publicly available recommendations outperform in efficient markets?

A semi-strong form of market efficiency implies that investors should not be able to outperform by trading on publicly available information, such as analyst recommendations (Malkiel and Fama, 1970). The quantitative valuation models do have an advantage because the valuations are not public knowledge - although any investor can put last year’s free cash flow in a Gordon Growth model. Popular multiples such as P/E and B/M, however, have been shown to outperform over time despite being publicly available.

Changes in target prices and recommendations are quickly absorbed and implemented in the stock prices, but quickly adjusting portfolios as soon as new recommendations are announced have been shown to outperform - but only before transaction costs (Barber, Lehavy, McNichols

& Trueman, 2001). In this context, rebalancing portfolios daily to quickly account for any of Morningstar’s rating changes sound like the necessary success criteria, as we only rebalance at the end of each month in the backtest. Apple could receive a 5-star rating on the 2nd of March, and our star portfolios would not know about it or adjust before the end of March. Much could happen with the stock price in the span of a month. Yet, a trading strategy or recommendation that relies on daily rebalancing is both costly and not very investor-friendly. Simultaneously, Morningstar does emphasize their long-term mindset and that market prices should converge on their fair value estimates within generally three years (Morningstar, 2015, p. 10). However, we did test whether daily rebalancing would change the overall performance of Morningstar’s ratings. It did not.

Previous research by Morningstar suggests that the recommendations need more than a month to work their magic. When extending the holding period to 3 years, Morningstar finds that their ratings outperform on a wider sample of stocks (Collins & Gross, 2018). This implies forming monthly portfolios based on the star ratings and measuring the returns in up to 3 years after the portfolio formation. In other words, the 5-star rated stocks are bought and held for up to 3 years regardless of how their ratings change in these 3 years. We have performed the same exercise in our sample with both 1- and 3-year holding periods, but it was difficult to replicate the results of Collins & Gross.

Page 108 Are valuations and fair value estimates the alpha and omega of equity research?

Although the quantitative valuations clearly outperform Morningstar’s qualitative valuations in terms of risk and returns, equity research serves other purposes than just target prices and investment recommendations. Independent analysts can help investors, clients and managers to get to know the companies they invest in more intimately. Analysts often specialize in a group of firms in a certain industry or sector - thus acquiring in-depth knowledge of their markets.

Simultaneously, they maintain close relationships to the management of the firms covered which puts them in a good position to receive first-hand information. Sophisticated or time-consuming research such as asking a vast number of third-party distributors how their sales have been of Apple’s new iPhone can provide details about revenue expectations before the information is released to the public in a quarterly statement, which is valuable for any fundamentally driven investment strategy.

The quantitative valuations have a distinct weakness in the fact that they may result in negative fair values which are hard to interpret and difficult to trade on. If the stock does not have 10 years of accounting history, the valuation models based on 10-year fundamentals simply cannot provide an estimate. Simultaneously, the valuations are volatile and change considerably whenever they receive fresh data from the new annual reports. If the estimates and target prices of an equity analyst fluctuated wildly once a year, investors and clients would probably be skeptical.

Comparing quantitative terminal value and factor models

The quantitative terminal value models are not as efficient as other traditional factor models at splitting a universe of stocks into value versus growth, quality versus junk, or small versus big.

In the eyes of the valuation-focused models, fast-growing technology stocks can be undervalued while stable and high-returning quality names can be expensive. As stated by Graham and Dodd (1934); “investments must always consider the price as well as the quality of the security.” A fundamental valuation depends on a mix of parameters, that are difficult to capture in a single risk factor model; operating profit, growth, risk, returns on capital, debt, free cash flows, and more. But no matter how high the valuation, the price can be even higher. A stock is only attractive if it can be bought at a discount to its intrinsic value (Graham & Dodd, 1934).

Factor models based on simple metrics such as book/market, EV/EBITDA or free cash flow yield can be replicated faster and more easily, but the assumptions that an asset manager can make to differentiate his factor strategy are limited. The application of the quantitative terminal value models includes a variety of assumptions to WACC, steady-state growth, and the inputs based on the individual firm’s fundamentals. By tweaking these inputs, the model can be tilted towards deep value, growth, quality, more diversification and lower risk, or certain sectors. This way, asset managers can differentiate their application of the terminal value models considerably to provide incremental value for their investors.

Page 109 Much like the mispricing argument for the profitability and investment factors, often combined as the quality factor (Arnott, Harvey, Kalesnik & Linnainmaa, 2019, p. 30), we would argue that there is a similar explanation for the premium to undervalued stocks in the value driver model.

The premium is a result of the undervalued stocks being conservative businesses with sustainable competitive advantages that maintain higher margins and profitability without high growth and large investments, which can hurt the balance sheet and drive up debt. The undervalued stocks stay out of the headlines and do not attract investors with a preference for “lottery-like payoffs”

(Nguyen et al., 2014, p. 2). Such investors, and those who do not pay much attention to fundamentals, underpay and thus provide a premium for the undervalued stocks identified in the value driver models.