Labor protection, ownership concentration, and cost of equity capital: international evidence

Abstract

In this study, we investigate how labor protection institutions and the presence of controlling shareholders interact to determine a firm’s cost of equity capital in an international setting. Our firm- and country-level analyses reveal that firms in countries with greater labor rights have a higher cost of equity, over and above the differences explained by investor protection environments and firm- and country-level risk factors. We find that operating leverage and information asymmetry serve as channels through which labor protection increases the cost of equity capital. More interestingly, the detrimental effect of labor protection on the cost of equity is mitigated in the presence of controlling shareholders, especially in the case of a family controlling owner, indicating the effectiveness of controlling owners at coping with difficult labor relations. The economic significance of this advantage is non-trivial: a one-unit increase in the labor market regulation index increases the annual cost of equity by 0.776% in widely held firms, whereas it increases by only 0.415% in firms with controlling shareholders. Our study identifies better labor relations as a potential channel through which concentered ownership affects the interaction of labor power and the cost of equity capital.

Appendix: Cost of equity models and estimation procedure

1.1 Cost of equity models

Below are the four cost of equity models used in this study. They are stated in the form of a pricing equation and include the model-specific assumptions.

Notation:

• $P_{t}$ = Price per share at date t

• $bv_{t}$ = Book value of equity per share at date t

• $bv_{t + \tau }$ = Expected book value of equity per share at date t + τ

• $dps_{t + \tau }$ = Expected dividend per share at date t + τ

• $eps_{t + \tau }$ = Expected earnings per share at date t + τ

• $g,g_{st} ,g_{lt}$ = Expected perpetual, short-term, and long-term growth rate, respectively

• $r_{CT} ,r_{GLS} ,r_{OJ} ,r_{PEG}$ = Implied cost of equity capital of Claus and Thomas (2001), Gebhardt et al. (2001), Ohlson and Juettner-Nauroth (2005) and Easton (2004), respectively

• Claus and Thomas (2001)

$$P_{t} = bv_{t} + \sum\limits_{\tau = 1}^{T} {\frac{{(eps_{t + \tau } - r_{CT} \cdot bv_{t + \tau - 1} )}}{{(1 + r_{CT} )^{\tau } }}} + \frac{{(eps_{t + T} - r_{CT} \cdot bv_{t + T - 1} )(1 + g)}}{{(r_{CT} - g)(1 + r_{CT} )^{T} }}$$

This model is an abnormal earnings model. The pricing equation shows that the current stock price equals the current book value of equity plus the present value of future expected abnormal earnings. Future abnormal earnings are proxies for economic profits and are computed by deducting a charge of equity capital from expected earnings. The model assumes an explicit forecast period of 5 years (T = 5), hence uses a stream of five expected earnings per share forecasts. Beyond year 5, all of the future earnings are assumed to grow perpetually at g, which is proxied by the country-specific 1-year ahead realized annual inflation rate.

• Gebhardt et al. (2001)

$$P_{t} = bv_{t} + \sum\limits_{\tau = 1}^{T - 1} {\frac{{(eps_{t + \tau } - r_{GLS} \cdot bv_{t + \tau - 1} )}}{{(1 + r_{GLS} )^{\tau } }} + \frac{{(eps_{t + T} - r_{GLS} \cdot bv_{t + T - 1} )}}{{r_{GLS} (1 + r_{GLS} )^{T - 1} }}}$$

This model is an abnormal earnings model. The pricing equation shows that the current stock price equals the current book value of equity plus the present value of future expected abnormal earnings. It specifies a forecast period of 12 years (T = 12). It first forecasts earnings explicitly for 3 years and then forecasts implicitly by mean reverting the period t + 3 firm return on equity to the industry-specific median return on equity. The mean reversion attempts to capture the long-term erosion of abnormal earnings over time. It further assumes that any growth in earnings past year 12 is value neutral, hence all future abnormal earnings thereafter are assumed to be constant.

• Ohlson and Juettner-Nauroth (2005)

$$P_{t} = \frac{{eps_{t + 1} }}{{r_{OJ} }} \cdot \frac{{(g_{st} + r_{OJ} \cdot dps_{t + 1} /eps_{t + 1} - g_{lt} )}}{{(r_{OJ} - g_{lt} )}}$$

This model relates share price to expected earnings per share 1-year ahead, the expected dividends per share 1-year ahead, the short-term growth and long-term growth in expected earnings, and the cost of equity capital. It assumes that the present value of dividends per share affects share price though the dividend policy is irrelevant. Following the approach of Gode and Mohanram (2003), g_st is proxied by the average of the growth in expected earnings between period t + 1 and t + 2 and the explicit 5-year growth forecast. The model also assumes positive growth in expected earnings so as to generate a solution. g_lt mirrors the growth rate of the overall economy and is proxied by the country-specific 1-year ahead realized annual inflation rate.

Easton (2004)

$$P_{t} = \frac{{(eps_{t + 2} + r_{PEG} \cdot dps_{t + 1} - eps_{t + 1} )}}{{r_{PEG}^{2} }}$$

This model emphasizes the role of short-term earnings forecasts in valuation. It assumes that price is determined by the abnormal earnings, which is the expected earnings per share 2-year ahead plus earnings from re-invested dividends per share 1-year ahead minus the expected earning per share 1-year ahead. In addition, the model assumes that abnormal earnings are constant over time and constrain positive growth in expected earnings, so as to obtain a solution.

1.2 Estimation procedures

The four models collectively require earnings forecasts of 3 years ahead $(eps_{t + 1} , { }eps_{t + 2} \;{\text{and}}\;eps_{t + 3} )$ and the expected 5-year earnings growth rate $(g_{5} )$. All of the estimates are mean analyst consensus forecasts. Together with stock price $(P_{t} )$, they are gathered from I/B/E/S using the native currency. To be included in the sample, an observation must have $P_{t}$, $eps_{t + 1}$, and $eps_{t + 2}$ data and either $eps_{t + 3}$ or $g_{5}$. If $eps_{t + 3}$ is missing, it is assumed to be the 2-year ahead earnings forecast growing at the 5-year earnings growth rate, that is, $eps_{t + 3} = eps_{t + 2} \times (1 + g_{5} )$. Any earnings forecasts beyond year 3 are generalized in the same way. However, $g_{5}$ can be proxied by computing a growth rate between the 2-year ahead and 3-year ahead earnings forecasts, that is, $g_{5} = (eps_{t + 3} - eps_{t + 2} )/eps_{t + 2}$. All of the negative earnings forecasts and growth rates are eliminated.

The financial data in the models are matched with the price and earnings forecasts taken +10 months after the fiscal year end, following Hail and Leuz (2006), to ensure that all of the financial information are available to investors and can be impounded into the model at the time of cost of equity estimation.²⁴ The book value of equity per share $(bv_{t} )$ is computed as the total common equity divided by the number of common shares outstanding as of the fiscal year end, both of which are taken from Worldscope. Unlike I/B/E/S, Worldscope data applies the Euro retroactively; hence, the total common equity of the eight Euro-in countries can be extracted using the legacy currency throughout the sample period. Currency consistency is required in the cost of equity estimation as all of the price and analyst forecasts are either in the native currency or the legacy currency after the currency conversion.²⁵ For the expected book value of equity per share $(bv_{t + \tau } )$, clean surplus accounting is assumed, that is, $bv_{t + \tau } = bv_{t + \tau - 1} + eps_{t + \tau } - dps_{t + \tau }$. The expected dividend per share $(dps_{t + \tau } )$ is computed as a constant percentage of the expected earnings per share, that is, $dps_{t + \tau } = eps_{t + \tau } \times k_{t}$. Dividend payout ratio is obtained directly from Worldscope and k_t is defined as the historic 3-year mean dividend payout ratio. If k_t is missing or less than zero, it is replaced by the country-year median payout ratio.

Both g and g_lt represent the perpetual or long-term growth rate in expected earnings. They are proxied by the country-specific 1-year ahead realized annual inflation rate, which are gathered from Datastream or the Statistical Yearbook.²⁶ Any negative inflation rates are substituted by the country-median inflation rates over the entire sample period. The rationale is that deflation is not expected to persist in the long-run, consequently a substitution is necessary.

Except for the Ohlson and Juettner-Nauroth (2005) model, there is no closed form solution provided for the equity valuation models. Each cost of equity proxy is therefore determined by an iterative numerical approximation that identifies an annual firm-specific discount rate that equates P_t to the right side of the pricing equation. The iteration starts at zero and increments by 0.0001 and will discontinue either when the cost of equity proxy is accurate at two decimal places or when the proxy reaches a value of one. A maximum of one is set in the iteration procedure because any cost of equity estimate that is greater than 100% seems irrational. To serve as a check over the iteration process, an inputted price is computed by placing the proxy back to the pricing equation and any proxy that results in greater than 1% difference between P_t and the inputted price is eliminated.