Free Cash Flow, Enterprise Value, and Investor Caution
Harlan Platt
College of Business Administration
Northeastern University Sebahattin Demirkan School of Management SUNY Binghamton University
Marjorie Platt
College of Business Administration
Northeastern University
Abstract：By analyzing actual cash flows in comparison with enterprise values (market capitalization plus debt minus cash) we document that the market dramaticall undervalues firms. The findings suggest that the equity market appears to have an extraordinarily high discount rate which negates future earnings in the calculus of firm value. That is, the discount rate is so high that the vast majority of future cash flows are virtually ignored.
Our research finds that stock prices do not reflect future corporate earnings. This finding contrasts with the well known statement in finance textbooks that “the value of a firm equals the present discounted value of future cash flows.” In fact, we find that enterprise values are substantially less than the present discounted value of future cash flows. A one-dollar increase in future cash flows produces only a 75 cent increase in a firm’s enterprise value.
The implication of our work is clear: companies are worth far more than the market believes. This provides strong support to the idea behind the private equity industry. We realize that of late private equity firms have overpaid for acquisitions and may lose their entire investment during the current phase of deleveraging. Yet, if private equity firms acquire companies at reasonable prices using less debt, they are likely to create substantial value as a consequence of the fact that companies are so undervalued by the market relative to their cash flows.
There are no previous research efforts following our methodological design based on actual cash flows. Rather, .prior research studies have focused on the
relationship between forecasted cash flows (by market analysts) and enterprise value. Our approach focuses on a different question – the relationship between discounted future cash flows and the current market value as posited by financial theory.
1
Keywords: Enterprise Value, Actual Cash Flow, Cash Flow, Valuation
1.Introduction
The common explanation provided in finance textbooks for the value of the firm is that it equals the present discounted value of future free cash flows (FCF). Few analysts or market observers disagree with this statement. Despite its universal acceptance, there are few studies of the basic FCF proposition and the theory that underlies the science of valuation. In this paper, we explore the question of whether the value of the firm is related to its future cash flows. Existent literature on this subject includes a few studies conceptually similar to ours and a large body of work on questions peripheral to the basic issue addressed in this paper. Those related works use the FCF valuation theory to address issues of market efficiency. Our work is directed at valuation and not the market efficiency question.
Obviously actual future cash flows are unknown when analysts estimate value. Lacking actual future cash flow data, analysts create careful projections of annual cash flows for several years, usually less than 10, and then estimate cash flows in additional years with a terminal value. Public companies have value forecasts
prepared for them by many unrelated individuals and organizations. Some forecasts are too optimistic while others are too pessimistic. Presumably optimistic forecasters are buyers of securities while pessimistic forecasters are sellers. A security’s market price would then be the share value that clears the market of optimists and pessimists.
The specific projections of all individual forecasters are unavailable. What is known, at a point in time, is the actual market capitalization and enterprise value (EV) that results from the interactions of these many forecasts. Some researchers have tested the relationship between the value of the firm and cash flow forecasts by obtaining a sample of analyst’s forecasts or forecasts from other published sources. We instead substitute actual cash flows for forecasted cash flows. Our null hypothesis assumes that the market-clearing forecast of future free cash flows is correct for every company. In that case, actual cash flows can be substituted for cash flow forecasts. If the market clearing forecast is too optimistic (pessimistic) then the observed EV exceeds (is less than) the present discounted value of actual free cash flows. Our first empirical test examines how closely EV compare with the present discounted value of actual subsequent cash flows. Finding the theory to be less than complete, our second empirical exercise considers additional explanatory factors to explain EV. This portion of the paper tests whether the accepted FCF theory fully explains EVs.
2.LITERATURE REVIEW
The earliest written discussion of the idea that the value of something is related to its future cash flows comes from Johan de Witt (1671); though the basic idea traces back to the early Greeks1. In modern times, the idea that corporate value is related to
1
See Daniel Rubinstein, Great Moments in Financial Economics, Journal of Investment Management (Winter
2003).
2
future dividends was first described by John Williams (1938)2. Durand (1957) observed what later became known as the Gordon growth model, that a dividend growing at a constant rate forever can be capitalized to estimate a firm’s value.
The literature that tests the FCF theory examines a variety of valuation methods. All of these tests rely on forecasts of cash flows or earnings made contemporaneously with the valuation estimate. That is, starting in a given year, they compare actual EV against forecasts, made that year, for the same company. For example, Francis,
Olsson, and Oswald (2000) compared three theoretical valuation models-- discounted dividends (DD), discounted FCF, and discounted abnormal earnings (AE)3 – by analyzing Value Line annual forecasts for the period 1989 – 1993 for a sample of 2,907 firm years that ranges between 554 and 607 firms per year. They found that the AE model had a 27% lower absolute prediction error than the FCF model and a 57% lower absolute prediction error than the DD model.
Sougiannis and Yaekura (2001) also consider three multiperiod accounting based valuation methods: an earnings capitalization model (similar to FCF), residual income (a version of AE) without a terminal value, and residual income with a terminal value4. They put analyst’s earnings forecasts into the three theoretical models and find overall that they provide greater insight than merely relying on current earnings, book values or dividends. Their sample covered 36,532 firm years over the period 1981 – 1998 of which 22,705 consisted of one year forecasts, 9,420 of two year forecasts, 1,279 of three year forecasts, and 3,128 of four year forecasts. They found that the AE model with a terminal value most accurately predicted current equity values in 48% of cases, the FCF model was most accurate in 18% of cases, and the AE without a terminal value was most accurate in 13% of cases. Current income and book values provided the best forecasts for the remaining 21% of the sample.
Liu, Nissim and Thomas (LNT) (2002) in an article similar to Sougiannis and Yaekura (2001) found that multiples based on analyst’s forward earnings projections (made in the same year) explain stock prices within 15% of their actual value while historical earnings, cash flow measures, book value, and sales were not nearly as insightful. LNT argue that multiples value future profits and risk better than present value forecasts. Their multiples are derived based on current earnings and stock prices.
Gentry, Whitford, Sougiannis, and Aoki (2001) took a different theoretical and empirical approach comparing an accounting method which looked at the discounted value of future net income to a finance method that looked at the discounted value of FCFs to equity. Their analysis tested the closeness with which each model predicted capital gains. The sample included both US (1981 – 1998) and Japanese companies (1985– 1998). Each year had between 881 and 1034 US companies and 166 to 365
2
See, Aswath Damodaran, “Valuation Approaches and Metrics: A Survey of the Theory,” Stern School of
Business Working Paper, November 2006. Damodaran notes that Ben Graham saw the connection between value and dividends but not with a discounted valuation model.
3
Abnormal earnings as discussed by Ohlson (1995) assume that the value of equity equals the sum of They also report that a 4% constant growth rate provides the best terminal value, even better than ones
3
book value plus abnormal earnings.
4
based on individual firm growth forecasts.
Japanese companies. They found that the FCFs to equity method were not closely related to capital gains rates of return for either US or Japanese companies. In the US they found a strong relationship between cash flows associated with operations, interest, and financing (the accounting method) to capital gains; no similar relationship was found in Japan.
Finally, Dontoh, Radhakrishnan, and Ronen (2007) compared the association between stock prices and accounting figures. They found that the association between stock prices and accounting numbers has been declining over time. They suggest that this may be due to increased noise in stock prices resulting from higher trading volume driven by non-information based trading.
A further related literature examines the relationship between valuation and
changes in dividends . These studies are concerned with market efficiency. Dividends are a straightforward concept: they are the payments made to equity holders by a company. Dividends may also be thought to include all cash payouts to equity
including share repurchases, share liquidations, and cash dividends. Several studies have examined whether changes in dividends relate to changes in equity values; among these are Shiller (1981), LeRoy and Porter (1981), and Campbell and Shiller (1987). These tests generally find that stock market volatility can not be explained by subsequent changes in dividends. Larrain and Yogo (2008) take a slightly different look at equity volatility. Using a more aggregate sample they find that the majority of the change in asset prices (88%) is explained by cash flow growth while the
remaining 12% is explained by changes in asset returns. They conclude that stock prices are not explained by dividend changes.
The residual income method is conceptually more similar to FCF than to
dividends. Residual income at its most basic equals the firm’s net income minus the cost of its capital. In the accounting literature, Ohlson’s (1991, 1995) formulation of a residual income model (RIM) is widely accepted and has been subjected to numerous tests. RIM begins with an accounting identity; namely that the change in book value equals the difference between net income and dividends. Ohlson then defines AE as the difference between net income and lagged book value. It is then a small step to observe that the present discounted value of expected future abnormal earnings plus the book value of equity equals stock price5. Jiang and Lee (2005) test both the RIM and the dividend discount model. Their test of equity volatility finds that RIM provides more and better information than dividends.
3.METHODOLOGY
Unlike previous studies, we rely on actual subsequent cash flows over a period of time rather than forecasts of cash flow made contemporaneously with EV. Previous researchers can be thought of as studying the consistency between contemporaneous EV determined in the market and forecasts of future cash flows. Our study does not have that focus. We instead are interested in the actual accuracy of market determined EVs. We compare EVs at a point in time to subsequent cash flows. The closer these values are the more accurate is the market in valuing companies based on their future
5
See Jiang and Lee (2005), page 1466.
4
cash flows.
In order to estimate corporate value with FCFs, annual costs of capital must be estimated for each company. An alternative is to determine value using the capital cash flow (CCF) method. CCF yields the same present value as FCF6 but only
requires a single cost of capital estimate for each firm. This is the approach we follow.
CCF is determined following Arzac (2005) as follows:
CCF = net income + depreciation - capital expenditures – Δ working capital +
Δ deferred taxes + net interest
Estimated enterprise value (EEV) is calculated with the CCF estimates as follows:
EEV =Σ(CCFi,j ) /(1+ kj )t TVj /(1+ kj )y , (i=1….y)
where k is cost of capital, TV is terminal value, i is year, y is the final year with cash flow data and j represents firm. Terminal value is estimated according to the Gordon growth model. EEV estimates are compared with EV, the firm’s actual value as of the last trading day of the year. EV is calculated following Arzac (2005) as follows;
EV = MarketCap + Debt − Cash
The comparison between EV and EEV is a test of the accuracy of the market’s valuation process. Cases where EV exceeds (is less than) EEV are ones of overly optimistic (pessimistic) market valuation.
4.Data
We begin with all firms with fiscal year end for which there is data for: • cash and short-term investments (data1), • total assets (data6), • current assets (data4), • current liabilities (data5), • short-term debt (data44), • long-term debt (data9),
• notes payable (data206), and • deferred taxes (data74),
• capital expenditures (data128) • sales (data12),
• net income (data172) • depreciation (data14) • interest expense (data15) • interest income (data62)
• common shares outstanding (data25), • year-end stock price (data199).
This results in an initial sample of 131,518 firm-year observations. All firms are classified into their respective industries using historical SIC codes (data324).
For each firm-year in the initial sample, we compute the following variables; EV = Market Cap (data199*data25) + Debt (data9 + data44 + data206)
6
See Arzac (2005) or Platt (2008).
5
- Cash (data1)
WC= Net current assets (data4 - data5) – cash (data1) + notes (data206) D= Long term (data9) + short term (data44 + data206) E= Share price (data199) * Number of shares (data25)
(where EV is enterprise value, WC is working capital, D is debt, and E is equity) In addition we also compute lagged values for WC and deferred taxes (data74). Next, we obtain betas for firm-years from Compustat’s Research Insight. Betas are winsorized at the 1st and 99th percentiles to account for extreme outliers in the data.
Interest rates based on the 10-year constant maturity series (I10YR) are obtain from the Federal Reserve Bank’s website. After merging with the interest rate data and the betas, the sample size reduces to 69,643 firm-year observations. The loss in observations is largely due to missing data on the betas or deletions due to non-availability of lagged firm-year data.
With the merged dataset, we compute the following variables, where LWC represents the lagged value of WC and Ldata74 is the lagged value for data74:
CCF = net income (data172) + depreciation (data14) - capital expenditures
(data128) + WC - LWC + deferred taxes (data74) - Ldata74 + interest paid (data15) –interest received (data62);
βA = (1 / (1 + D/E))*β
KU1 = I10YR + βA *ERP1 KU2 = I10YR + βA *ERP2 KU3 = I10YR + βA *ERP3;
(where CCF is capital cash flow, βA is the asset beta, ERP is the equity risk premium, and KU1, KU2 and KU3 are estimates of the unlevered cost of capital for three different ERPs (ERP1 = 0.03;ERP2 = 0.05;ERP3 = 0.07)
Results were essential identical regardless of the choice of ERP and so we report on those for ERP3. We then drop all observations with fiscal year greater than 2000 to allow a sufficient numbers of years of actual cash flow data to be in the dataset.
From the summary files of the Institutional Brokers Estimate System (IBES)
database, we extract median values of long-term growth in sales forecasts for all firms. The median value is based on all analyst estimates of long-term (5 to 10 years) growth forecasts made for each firm. Prior studies use this as a measure of the estimated growth rate for a firm’s cash flow. Many of the growth rate forecasts were
extraordinarily large, and so we followed Sougiannis and Yaekura (2001) by using the growth rate in GDP instead of the IBES values.
The final dataset consists of 27,027 firm-year observations with complete data on all variables of interest. Of this 2,820 firm-years are data for companies with five or more years of information. Firm’s whose last year of data had negative FCF were dropped from the sample since terminal value could not be calculated for them. This left us with 1,821 firms.
Some companies in our sample have only five years of actual cash flow data; others have as many as 12 years of data. Recently it has been argued that the terminal value estimate dominates estimates of present value, see Platt and Demirkan (2008).
6
To insure that EEV estimates are not unduly influenced by estimates of terminal value, EEV is calculated repeatedly for each company starting with using five years of data and then using more years,12 years, depending on how much data the company has available.
5.CONCLUSION
We began this paper saying that the most common explanation in finance
textbooks for the value of the firm was that it equaled the present discounted value of future cash flows. Our results suggest that a better description for textbooks is that the value of the firm is related to but unequal to the present discounted value of future cash flows. In conjunction with Platt and Demirkan (2008) which finds that the TV is the principle part of EEV (i.e., approximately 92.3%) it would seem that the market values firms based on their near term (perhaps five years or fewer) subsequent cash flows. In fact, one dollar increase in future cash flows produces far less of an increase in a firm’s EV. Theoretically this conforms to a version of the Gordon (1962)
two-stage growth model with a WACC based discount rate during the early period and a very high discount rate during the future period).
Supporting evidence to our surprising finding appear in everyday stock market tables. For example, the following quote from Bloomberg.com of December 8, 2008 speaks precisely to our findings.
“Cheapest Stocks Since 1995 Show Cash Exceeds Market (By Michael Tsang and Alexis Xydias)
Dec. 8 (Bloomberg) – “Stocks have fallen so far that 2,267companies around the globe are offering profits to investors for free. That’s eight times as many as at the end of the last bear market, when the shares rose 115 percent over the next year.
Bank of New York Mellon Corp. in New York, Danieli SpA in Buttrio, Italy and Seoul-based Namyang Dairy Products Co. Hold more cash than the value of their stock and debt as the slowing world economy wiped out $32 trillion in capitalization this year.”
The Bank of New York Mellon, for example, on that day had a market
capitalization of $31.71 billion, debt of $35.83 billion, and cash of $75.50 billion. In this case, the market has an infinite discount rate on any and all cash flows.
A possible explanation for our higher EEV estimate than actual EV is that our unlevered cost of capital (KU) estimate is too low and therefore associated with a too high TV estimate. However, we calculated three KU estimates, based on generally accepted equity risk premium (ERP) levels and then used the highest KU. It is true however, that there is a KU which equilibrates EV with our EEV.
Another possible explanation is that forecasts relied upon the valuation process are inaccurate and that future cash flows far exceed what analysts had expected. We find this to be the least satisfactory explanation.
REFERENCES
Arzac, Enrique, 2005, Valuation For Mergers, Buyouts, and Restructuring, John Wiley & Sons.
7
Campbell, J., and Shiller, R., 1987, Cointegration and Tests of Present Value Models, Journal of Political Economy, 95(5):1062-88.
Daines, R, 2001, Does Delaware law improve firm value?, Journal of Financial Economics, 62: 525-558
Damodaran, A., 2006, Valuation Approaches and Metrics: A Survey of the Theory, Working Paper, Stern School of Business..
Dontoh, A., Radhakrishnan, S., and Ronen, J., 2007, Is Stock Price a Good Measure for Assessing Value-Relevance of Earnings? An Empirical Test, Review of Managerial Science, 1(1):3-45.
Fama Eugene F. and Kenneth R. French, 1992, The Cross- Section of Expected Stock Returns, The Journal of Finance,47: 427-465.
Fama Eugene F. and Kenneth R. French., \"Size and Book-to-Market Factors in Earnings and Returns\Francis, J., Olsson, P., and Oswald, D, 2000, Comparing the Accuracy and
Explainability of Dividend, Free Cash Flow, and Abnormal Earnings Equity Value Estimates, Journal of Accounting Research, 38(1).
Gentry, J. Whitford, D., Sougiannis, T., and Aoki S., 2001, Do Accounting Earnings or Free Cash Flows Provide a Better Estimate of Capital Gain Rates of Return on Stocks?, Security Analysts Journal, 39(5):66-78.
Hovakimian, A., T. Opler, and S. Titman, 2001, The debt-equity choice, Journal of Financial and Quantitative Analysis, 36(1):1–24.
Larrain, B. and Yogo, M., 2008, Does firm value move too much to be justified by subsequent changes in cash flow, Journal of Financial Economics, 87(1):200-26. LeRoy, S. F. and Porter, R. D., 1981, The Present-Value Relation: Tests Based on Implied Variance Bounds, Econometrica, 49(3):555-574.20
Jiang, X. and Lee B., 2005, An Empirical Test of the Accounting-Based Residual Income Model and the Traditional Dividend Discount Model, Journal of Business, 78(4):1465–1504.
Larrain, B., Yogo, M., 2008. Does Firm Value Move Too Much to be Justified by Subsequent Changes in Cash Flow?, Journal of Financial Economics, 87 (1), 200–226.
Liu, J., Nissim, D., and Thomas, J., 2002, Equity Valuation using Multiples, Journal of Accounting Research, 40(1): 135- 172.
Myers, S. C., 1977, Determinants of corporate borrowing, Journal of Financial Economics, 5:147-175.
Ohlson, J., 1991, The theory of value and earnings, an introduction to the Ball-Brown analysis, Contemporary Accounting Research, 8:1-19.
Ohlson, J., 1995, Earnings, book values, and dividends in security valuation, Contemporary Accounting Research, 11: 661-87.
Polk, C., Thompson, S. and Vuolteenaho, T., 2006, Cross-sectional forecasts of the equity premium, Journal of Financial Economics, 1:101-141.
Platt, H, 2008, Cash Flow Contradistinctions, Commercial Lending Review, 23 (2):19-24
Platt, H. and Demirkan, S., 2008, Perilous Forecasts: Implications of Reliance on
8
Terminal Value, Working Paper, Northeastern University.
Shiller, R.J., 1981, Do stock prices move too much to be justified by subsequent movements in dividends?, American Economic Review,71 (3): 421-36.
Sougiannis, T., and Yaekura, T.,2001, The Accuracy and Bias of Equity Values
Inferred from Analysts Earnings Forecasts, The Journal of Accounting, Auditing, and Finance, 16(4):331–362.
Rubinstein D.,2003, Great Moments in Financial Economics: II. Modigliani-Miller Theorem, Journal of Investment Management. 1(2).
Tsang M. and Xydias A., 2008 Cheapest Stocks Since 1995 Show Cash Exceeds Market, Bloomberg.com, December 8, 2008.
9
因篇幅问题不能全部显示，请点此查看更多更全内容