For the Analyst: Peer Benchmarking Methods to Improve Earnings Forecasts

Finding suitable peers for financial analysis is a vexing task that requires careful consideration of firms’ underlying economics, accounting choices, and financial statement presentation. But without comparable financial statement information, peer benchmarking may yield less meaningful and even misleading insights that negatively impact earnings forecasts.

In a recent study published in The Accounting Review, we developed a methodology to identify comparable firms for benchmarking and analyzed its implications for analyst outcomes and valuation with multiples. In this post, we will highlight the salient details, some of which may surprise you.

There are different ways to define peer firms, such as industry membership, stock index membership, closeness in market capitalization, and similarity in value drivers (e.g., P/E ratio, return-on-invested capital, and growth).

As an alternative to traditional classifications, researchers have tested new ways to identify peer firms, such as investors’ co-search, intensity of firms’ filings with the SEC’s EDGAR, and stock information on Yahoo! Finance.

These widely utilized methods fail to directly address a crucial aspect of firm benchmarking: the availability of key financial statement information for peer firms. When several financial statement line items are missing for a peer firm, analysts struggle to derive meaningful inferences from the comparative to the focal firm’s financial statements.

Our financial statement benchmarking (FSB) measure aims to fill this gap. The data and code are freely available on our website.

Capturing the Degree of Overlap Between Financial Statement Items

Built on the Jaccard similarity coefficient, pairwise FSB captures the degree of overlap in financial statement items reported by two firms, with scores ranging from 0 (no overlap) to 1 (full overlap). The higher the FSB score, the greater the benchmarking information available to external users.

For instance, if the focal firm has reported 270 items, 200 of which overlap with 220 items reported by the peer firm, the FSB score is 0.69 (200 / (270 + 220 – 200). To put this into context, the average score for analyst-chosen peers in our sample is 0.68.

Assuming that FSB is a helpful metric in capturing the similarity of two firms’ underlying economics and accounting choices, we expect it to be positively correlated with analysts’ choices of peer firms.

Our sample of analyst-chosen peers comes from a Review of Accounting Studies article, “Analysts’ choice of peer companies.” By manually screening more than 2,500 sell-side equity analysts’ reports, the authors extracted data on comparable peer firms selected for the focal firm in each report.

In our study, for each analyst-chosen peer firm, we selected a matching firm in the same industry that was not chosen but which had a similar size and valuation multiple. The results show that analysts tend to choose peer firms that are more comparable to a focal firm from a financial statement benchmarking perspective.

When FSB is higher by one-standard-deviation, the likelihood of being selected as a peer firm by an analyst increases by 13%.

Higher FSBs Increase Accuracy of Earnings Forecasts

Does choosing peers with higher FSBs have positive implications for analyst performance? We find that when the average FSB of the set of analyst-chosen peer firms is one-standard-deviation higher, the accuracy of analysts’ earnings forecasts increases by about 23%.

When selecting peer firms, look for firms that have more similar financial statements to the focal firm, even if that means searching outside the focal firm’s main industry. In fact, only 40% of the analyst-chosen peer firms operate in the same product market as the focal firm.

Which companies do you think would be good peer firms to choose when analyzing Colgate-Palmolive? Morningstar lists Procter & Gamble and Unilever as top peers for the company. Despite being listed on a US stock exchange, Unilever has a modest 0.69 FSB score with Colgate-Palmolive.

This is likely because the company uses International Financial Reporting Standards to prepare its financial statements. Using different accounting standards reduces comparability due to differences in the recognition and presentation rules. In contrast, P&G and Colgate-Palmolive have a higher FSB score of 0.77, suggesting a greater comparability than Unilever and Colgate-Palmolive.

In contrast to Morningstar’s approach, Google Finance creates a list of peer firms based on investors’ co-search activity. Notably, among the peer firms Google Finance lists for Colgate-Palmolive is Coca-Cola. Although this observation may seem unintuitive at first blush, our methodology suggests that, from a financial statement benchmarking perspective, Coca-Cola would be an excellent fit in this case because its FSB score with Colgate-Palmolive is well above the average at 0.82. This may explain why investors extensively co-search the financial information of the two companies.

Validation and Testing

After validating and testing the pairwise FSB metric, we aggregated data across all industry peers of the focal firm to understand how easy it is to benchmark a firm’s financial statements overall. This process yielded a large panel of firm-level FSB data. Also, to enrich our methodology, we decomposed FSB at the financial statement level, generating separate FSB scores for the income statement, balance sheet, and statement of cash flows.

While analysts’ consensus earnings and net debt forecasts are more accurate when firm-level FSB is high (i.e., it is easy to benchmark and understand a firm’s financial statements), income statement and balance sheet benchmarking play different roles in those outcomes.

We find that the Income statement FSB score predicts the accuracy of earnings forecasts but not net debt forecasts. In contrast, balance sheet FSB score predicts the accuracy of net debt forecasts, but not earnings forecasts. In economic terms, a one-standard-deviation increase in income statement (balance sheet) FSB is associated with a 17.3% (12.1%) more accurate consensus earnings (net debt) forecasts. These findings highlight that benchmarking benefits depend on the context of the analysis. 

For the Investor: Industry, Industry-Size, or FSB Peers

Beyond positive analyst outcomes, a key question for investors is whether choosing peer firms based on FSB improves valuation with comparables. To this end, we compared the predictive ability of the valuation multiples formed using FSB-based peers to those of the models employing traditional methods for peer firm selection, such as industry- and size-based peers. Specifically, we regressed the future enterprise value-to-sales ratio (EVS) of the focal firm on the average EVS calculated for three sets of peers: (1) industry peers, (2) industry-size peers, and (3) FSB peers. 

When predicting one-year, two-year, and three-year-ahead EVS, the models using the average valuation multiple of FSB-based peers consistently outperform those employing only industry- and industry-size peers. For instance, the R-squared of the model predicting one-year-ahead EVS increases from 24.8% to 31.8% when the average EVS of the four highest FSB peers is included in the model.

In conclusion, we note that FSB is a simple, straightforward measure summarizing the overlap in peer firms’ underlying economics and accounting choices, which are the key factors that shape financial statements. FSB is available at the pairwise and firm levels, allowing external users to find suitable peer firms for various purposes, including relative performance, compensation, and valuation benchmarking.