Real options as a component of the market value of stocks: evidence from the Spanish Stock Market

Abstract

This paper aims to examine how investors’ expectations about the value of a firm's real options are reflected in the price of its stocks. If the real-option approach is correct, then the efficient-market hypothesis predicts that stock prices will reflect the available information relative to the real options held by firms and their ability to identify, acquire, maintain and exercise them. The role of investment irreversibility, operating and financial flexibility, business and geographical diversification, and size are examined as indicators of a firm's real option strategy. The empirical analysis of a panel of 101 companies listed on the Spanish Stock Exchange during the period 1991–1997 provides evidence consistent with predictions. The market value of the real option portfolio is significantly and positively related to business diversification, asset irreversibility and operating leverage, and negatively related to size. In addition, financial leverage and geographical diversification are not significantly related to our proxies for the market value of real options. These results are robust even after controlling for industry, and alternative measures of investment flexibility and business diversification.

Acknowledgements

The authors benefited from the useful comments of Michel Dubois, Joan Montllor, Pedro Fortuny, and participants at the XI Congreso de ACEDE. Financial support from AECA, and Junta de Castilla y León (grant: VA05204) is also acknowledged. Any errors are our own.

Notes

A survey of the real options approach can be found in Dixit and Pindyck (1994) and Trigeorgis (1996). See also Amram and Kulatilaka (1999) and Copeland and Antikarov (2001) for a practical perspective.

Empirical evidence has basically concentrated on the analysis of specific investment projects with the aim of verifying the relevance of their embedded options and, thus, of demonstrating the suitability of the options pricing models for their valuation. For an interesting collection of similar case studies, see Trigeorgis (1999).

In addition to these studies, mention must be made of the statistical analysis of ‘option reasoning’ propositions in explaining investment decisions adopted in practice by managers. Some good examples include Kogut (1991) and Hurry et al. (1992).

On the analysis of a firm's strategy as a sequence of options see, for example, Sharp (1991) and Bowman and Hurry (1993).

Besides, investments such as those dedicated to new-products R&D are usually of a highly irreversible nature while constituting at the same time one of the principal mechanisms for the identification and acquisition of new growth options. Moreover, the value of these growth options generally depends positively on their exclusivity, which in turn depends on the specificity of the assets supporting them.

See Chung and Charoenwong (1991) and Chung and Kim (1997).

Using the average beta of the industry instead of a company's beta permits to reduce the problems derived from the influence of real options on stock beta. The industry average beta selected in each exercise to estimate the first proxy of the discount rate is that published annually by the Revista de la Bolsa de Madrid, and is obtained on the basis of the last 36 monthly returns of each industry index. Annual risk-free rate of returns are approximated on the basis of the returns recorded at the end of each year for long-term government bonds (Main Economic Indicators, OECD), and the risk premium is considered constant during the 1991–1997 period and approximated as the average spread of the General Index of the Madrid Stock Exchange on the risk-free rate of return during this period (8.35%).

Logically,

and, hence,

In this respect, the real options ratio that is approximated on the basis of the risk-free interest rate is more exacting than that obtained through the industry average beta.

Extreme values are those that exceed the mean of the full sample plus twice its standard deviation. Of the total of 707 observations that make up the data panel, 90 observations were found with negative values and six extreme values for ROR1; and 90 observations with negative values and five with extreme values for ROR2.

The source of information used to estimate Herfindahl's Index (HERF) and Rumelt's ratio (RUM) is the Registro de Actividades de la CNMV. The analysis would certainly have benefited from considering the exact type of diversification followed by the companies, but the lack of sufficient and reliable information regarding their business lines made this impossible. Data in the Registro de Actividades de las Empresas de la CNMV only refers to the numbers of business lines and their sales.

The inconvenience of this variable lies in the fact that it is only possible to obtain an observation per company for the period as a whole. Hence this cannot be introduced into the analysis of panel data, but can still be used as a discriminating variable for comparing the average real options values. In the sample, 25% of the companies have a correlation coefficient of more than 0.70, and one third present values between 0.55 and 1.

As in the case of the correlation coefficient between growth in sales and investments, the multi-periodic nature of this variable prevents its inclusion in the panel data analysis. Moreover, the limited discriminating power exhibited by the values obtained by this measure in the sample (see Table 4) render it of limited use in the formation of subsamples and the consequent comparison of their influence on the market value of the real option portfolio. In the majority of observations, the correlation between sales and costs reveals values close to 1, and its mean is 0.95.

The restricted panel is obtained by removing those observations with negative values for its assets-in-place or which record extreme values in the ratios ROR1 or ROR2.

One variable of special importance in arriving at the value of real options is Research and Development expenditures. Unfortunately, the lack of information regarding R&D expenditures of the companies in the sample prevents its inclusion in the empirical analysis.

Both the fixed effects models (within) and the random effects model (random) are presented together with the Hausman test, which evaluates the null hypothesis relative to the absence of correlation between fixed effects and the remaining regressors.

However, low statistical significance for this variable does not help us to confirm this prediction. The negative relation between financial leverage and real option value seems to agree with empirical findings in previous studies. See Smith and Watts (1992) and McConnell and Servaes (1995). Regarding the Spanish case, Andrés et al. (2000) observe a negative relation between a firm's market value and financial leverage for those companies showing a larger relevance of growth opportunities.

Note that this same significance of the size coefficient was found in the result regressions of Models 1 and 2.