Does Valuation Model Choice Affect Target Price Accuracy?

Abstract

We investigate whether the choice of valuation model affects the forecast accuracy of the target prices that investment analysts issue in their equity research reports, controlling for factors that influence this choice. We examine 490 equity research reports from international investment houses for 94 UK-listed firms published over the period July 2002–June 2004. We use four measures of accuracy: (i) whether the target price is met during the 12-month forecast horizon (met_in); (ii) whether the target price is met on the last day of the 12-month forecast horizon (met_end); (iii) the absolute forecast error (abs_err); and (iv) the forecast error of target prices that are not met at the end of the 12-month forecast horizon (miss_err). Based on met_in and abs_err, price-to-earnings (PE) outperform discounted cash flow (DCF) models, while based on met_end and miss_err the difference in valuation model performance is insignificant. However, after controlling for variables that capture the difficulty of the valuation task, the performance of DCF models improves in all specifications and, based on miss_err, they outperform PE models. These findings are robust to standard controls for selection bias.

Acknowledgements

The authors acknowledge the helpful comments and advice of Martyn Andrews, Richard Barker, Nicholas Collett, Susan Ettner, Theodore Sougiannis, Richard Taffler, seminar participants at Edinburgh University, Reading University and Athens University of Economics and Business, and participants at the European Accounting Association 2007 Conference. The authors are particularly grateful to two anonymous reviewers for their constructive and insightful comments and suggestions, and for the advice of the Editor, Salvador Carmona. Efthimios Demirakos acknowledges a Ph.D. scholarship from the Propondis Foundation during the early stages of this research.

Notes

Gleason et al. (2007, fn. 19) express a similar view.

For a comprehensive literature review see Ramnath et al. (2008).

Green (2006, p. 1) states that ‘institutional investors pay significant amounts to obtain real time access to brokerage firm research through providers, such as First Call’. He finds that ‘early access to stock recommendations provides brokerage firm clients with incremental investment value’. Barker (2001) suggests that analyst reports influence fund manager behavior, while Imam et al. (2008) provide evidence that the need for analysts' research to be credible to fund managers influences their valuation model choices.

Viebig et al. (2008) assemble a collection of chapters written by investment bankers on the equity valuation models they use in practice, including proprietary and sophisticated DCF models.

Comparing the determinants of valuation model choice by sell-side analysts and underwriters requires care due to differences in the motives of the two groups. Underwriters are justifying an offer price to achieve a successful IPO, while sell-side analysts are typically supporting a buy recommendation for a stock. Deloof et al. (2009, p. 132) cite ‘differences in the quality and objectives of IPO prospectuses and equity research papers’ among the factors that explain potential differences in the results of IPO and equity research studies.

Bonini et al. (2008) reach a similar conclusion when analyzing equity research reports for Italian-listed firms published during 2000–2005.

In some cases, the analyst publishes another report, with a target price in GBP on the same day or in the following days, which we include in the sample. The majority of the excluded reports are for pharmaceutical firms (e.g. AstraZeneca, GlaxoSmithKline).

We use the current price to calculate boldness (defined below) and the two measures of forecast error.

Typical statements are ‘DCF-based target price’, ‘DCF-driven target price’, ‘our preferred valuation methodology is …’, ‘we set our target price based on …’

We also include reports that use these terminologies interchangeably.

For example, the average report length for the study of Asquith et al. (2005) is 6.3 pages.

Where a firm makes a capital change (e.g. a stock split) during the 12-month forecast horizon, we adjust the target price by multiplying by the ratio of the firm's adjusted stock price to its unadjusted stock price on the date of the report's publication. We use adjusted stock prices to estimate the 12-month-ahead Highest/Lowest stock price range. Where there is no capital change, we make no adjustments and use unadjusted stock prices to estimate the range. Data on stock prices are from Datastream.

As the current price, we use the stock price available to the analyst before the publication of the report. The first page of the report indicates the current stock price along with the target price. Where a firm makes a capital change (e.g. a stock split) during the 12-month-ahead forecast horizon, we adjust the target and current stock prices accordingly (see previous footnote).

Risk equals the standard deviation of daily stock returns from 1 July 2000 to 30 June 2002 for reports published in the period 1 July 2002 to 30 June 2003, or from 1 July 2001 to 30 June 2003 for reports published in the period 1 July 2003 to 30 June 2004.

However, contrary to expectation they find that target prices are more difficult to meet for firms with higher price volatility. They argue that this result is due to the high correlation between price volatility and TP/CP.

Data on market values (Datatype: MV) are from Datastream.

From another perspective, large successful firms that are ‘national champions’ in their industry sectors might have few comparable UK companies. However, analysts compare these firms with international companies. For example, analysts compare Tesco to Carrefour, GlaxoSmithKline and AstraZeneca to the ‘Global Big-Pharma’ industry, Shell and BP to the ‘Global Big-Oil’ industry, etc. These companies are global players, with similar business and financial models, facing similar risks and opportunities.

In the analyst earnings forecast literature, a typical definition is that a forecast is bold if it is above both the analyst's previous forecast and the consensus forecast, or below both (e.g. Clement and Tse, 2005). Our definition of target price boldness bears a closer relation to measures in the target price literature based on the ratio of target price to current stock price (TP/CP). Bradshaw and Brown (2006) condition much of their analysis on this variable. Bonini et al. (2008) refer to TP/CP−1 as the implicit return.

However, we expect target prices that support neutral recommendations to be more easily achieved at some point during the 12-month forecast horizon, because they are often set at a level close to the current stock price.

To estimate sales growth rates, we use data on sales from Worldscope (Net Sales or Revenues: WC01001).

We also use the number of firms that are constituents both of the FTSE industry specific index and the FTSE All-Share Index in August 2005. The results are qualitatively similar with this alternative measure.

From a theoretical perspective there should exist a sufficiently low discount rate to give an identical DCF valuation to a PE-based valuation based on bullish multiples. The discussion in the text reflects the practical reality of how analysts implement these valuation models. The evidence from the interviews of Glaum and Friedrich (2006) is consistent with the contextual factors that Imam et al. (2008) report as motivating analysts' valuation model choice.

However, there are analysts who exhibit differential ability to make profitable earnings forecasts and stock recommendations. For a literature review see Bruce and Bradshaw (2003).

Possible explanations for the difference between our study and Bradshaw (2002) and Asquith et al. (2005) include differences in the length of the sample equity research reports, differences in the sample period (Glaum and Friedrich, 2006), institutional differences in the equity research output between the City of London and Wall Street (Breton and Taffler, 2001), and the fact that we require attribution of valuation model choice in the reports.

The proportion of loss-making firms is higher among firms with negative growth: 46.22% (13.54%) of the sample reports of negative (positive) growth firms. Hence, this difference in valuation model choice might also be due to profitability differences between the top and bottom growth quartiles.

In Asquith et al. (2005) the proportion of neutral/negative recommendations is 29.2%.

Similarly, in Gleason et al. (2007), TP/CP increases from 1.24 to 1.40 in the period 1997–2000 and declines to 1.26 in 2003.

However, it might also be due to institutional differences between the USA and the UK. As discussed earlier, our sample has a greater number of neutral/negative recommendations, which reduces the mean (median) value of this ratio.

In our sample, 22.24% of the reports have a target price below the current price compared to only 8.3% in Gleason et al. (2007) and 2.7% in Asquith et al. (2005). In particular, our sample contains 109 reports with a target price below the current price (48 reports with neutral recommendations, 60 reports with negative recommendations and one report with a positive recommendation).

This might indicate that DCF models better capture a firm's long-term value, while PE-based target prices focus on the short-term and fail to fully impound information about future performance. Subsequent sections provide a more in-depth analysis of the dynamics of target price setting and performance.

See also note 19.

See Edwards (2004) for an application of this treatment effects framework.

Note that boldness controls for the effect of optimistic target prices that are likely to support positive recommendations. Table 6 shows that boldness and recm have a 0.414 correlation.

The data collection for this study took place in August 2005. Investext changes its collection of equity research reports based on collaborations with investment banks.

We are grateful to an anonymous reviewer for suggesting an analysis of the accuracy and prediction errors of target prices derived from a mixture of valuation models.