![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
There is a relatively unknown market for partial control or corporate influence in Spanish listed firms, where the control transaction size is below the legal threshold that triggers a mandatory tender offer, as this kind of deal looks for exercising some degree of control, but not a full control. The goal of this paper is to go further in its empirical analysis by exploring its distinguishing features, using as the criterion to define its transactions obtaining a seat in the board of directors. We find that these deals are mainly located in the segment of the market of large trades where the rules for private negotiations are easier to implement; the size of the block is relatively large and it is negotiated as a whole block. Besides, the most common buyer has no previous stake in the firm. We find no evidence that the buyers pay, in median, for a seat on the board of directors, but the variability of the premiums for those blocks is higher and shows that buyers that had no control position in the target firm pay more for being among largest shareholders (partial control) and less for not being among them (influence).
Acknowledgements
The authors would like to thank Vicente Salas, Román Mínguez and Javier Martínez for their comments and helpful suggestions, and Beatriz Alonso, Domingo García and Maria Paz Alonso for their support in the process of searching for the data.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 The closest approach to ours is in Bethel, Liebeskind, and Opler (Citation1998) that divide the sample according to different kind of investors and include the activist shareholder, defined as a blockholder who announces its intention of influencing firm policies (611). Barclay and Holderness (Citation1991) use a similar criterion to us but for analyzing private benefits.
2 There are several ways to name partial control blocks in the literature: Park, Selvili, and Song (Citation2008) label this kind of blocks as partial stock acquisitions; OECD (Citation2009) use minority shareholding while Salop and O’Brien (Citation2000) use the term partial ownership.
3 For the OECD there is a passive investment when the rights attached to the shares do not entitled the owner to be represented in the board or to access to sensitive information on the activities of the target.
4 We define market rule as the regulation that guides the large trades that can involve relevant effects on the target firm, as control variations, reorganization of corporations, changes in market prices, …
5 With the exception of three States: Maine, Pennsylvania and South Dakota (Grant, Kirchmaier, and Kirshner Citation2009), that follow a MBR. Nevertheless, for the rest of the country, there are several circumstances, such as ‘looting’ cases, ‘sale of office’ cases and ‘diversion of collective opportunities’ (Schuster Citation2010) that trigger an MBR, although their practical impact is very small.
6 Mikkelson and Partch (Citation1985) analyse a similar market for the USA, the market for secondary distributions which takes place usually after the close of trading; the seller may be required to register the offer with the SEC, providing information about the deal, and the report usually appears in The Wall Street Journal on the day following the filing of the prospectus.
7 Fernández-Blanco and García (Citation2000) use the press announcements of the block trades below 25% of the equity to make up their sample of significant transactions. Crespí-Cladera and Gispert (Citation2002) also use the CNMV as the source of their data focusing on increases of shareholdings by 5% or more.
8 This means that only around 500 transactions of the total consulted registers are related to transactions that concern us. Among them, each transaction implies several relevant facts. An extreme case is Iberdrola; we have computed more than 150 published relevant facts referring to the same deal. This example can explain the reason why our final sample is significantly smaller than the consulted registers related to block transactions.
9 The other weights that the filters represent in relation to the initial sample of block transactions are: 0.9% from transactions that involved non-listed firms; 1.3% from transaction that were not completed; 2.7% from transactions where the buyer had a stake higher or equal to 50% of equity; 1.6% from transactions related to a tender offer in the following year; 0.4% from transactions made for treasury stock; 1.3% from transaction between companies of the same group; 2% from transaction where the target was a financial firm. There are several reasons for excluding financial firms from the sample: Financial firms are subject to especial accounting standards and risk factors, and they are more heavily regulated than non-financial sectors as well. In fact, there exists in Spain, since 2009, a specific law for significant holdings in supervised financial firms (Directive 2007/44/EC of the European Parliament and of the Council of 5 September 2007). The last of these transactions involves six medium-size banks.
10 For instance, Barclay and Holderness (Citation1991) or Mikkelson and Ruback (Citation1985), for the USA, only analyse block trades involving at least 5% of the common stock since increased holdings beyond the 5% level must be disclosed; Crespí-Cladera and Gispert (Citation2002) use the same criterion for the Spanish case; Dyck and Zingales (Citation2004) define as a control transfer any transaction size that goes beyond 20% of the shares.
11 In Spain the ‘upper bound’ has been modified because of the change of tender offers law in 2007 that, in general terms, varies the threshold from 25% to 30% of the target’s equity. So the use of a qualitative criterion to analyse the transactions, instead of a quantitative one (% of equity), avoids a change in the criterion (size of the block) in order to be included in the sample.
12 This criterion means 0.5% of the block transactions. Summing up, the final sample is made up of about 25% of the block transactions identified in the CNMV registers. Baixauli (Citation2004), using the same source for constructing his sample of significant stakes in the Spanish Stock Market, has to rule out around 50% of the transactions having to use less variables than in our case (ownership structure, accounting variables, …). Fernández and Baixauli (Citation2003), basing also on the reports of block transfers made to the CNMV, have to rule out 80% of the initial sample. So our percentages are more or less in line with other papers based on the same source of information to construct their samples.
13 These figures mean 66 different target firms and 70 different acquirers over 22 years. Target firms represent 40% of the average of listed firms between 1990 and 2012 and 60% of the firms included in the IBEX-35, the main index of the Spanish Stock Market. Of those 66 target firms, 36 have been affected by only one deal.
14 Mikkelson and Partch (Citation1985) also find that on average the number of shares offered in registered distributions represents a higher figure than in the non-registered operations (8.1% and 2.7%, respectively).
15 In the case of the partial control deals under Applications and Tomas de Razón, the block size is slightly lower, around 10%, whereas this value is around 5% for partial control deals carried out in the Block Market.
16 We consider that the ownership structure is concentrated when the ratio of the sum of the stakes of the three largest shareholders to the total equity is over 50%.
17 As Agresti (Citation1990) indicates, GLM are defined by three components: a random component, a systematic component and a link which describe, respectively, the probability distribution of the response variable, the linear function of explanatory variables that is used as a predictor, and the relationship between the systematic component and the expected value of the random component. For an overview of the theory and applications of GLM see also Hardin and Hilbe (Citation2007).
18 For instance, if a distribution theoretically should have a dispersion parameter equal to 1 but the observed data violate this requirement, the QML framework allows to estimate the model with a free dispersion parameter. If the dispersion parameter exceeds 1, this situation is called ‘overdispersion’ (the opposite is known as ‘underdispersion’). Accounting for these situations is crucial to guarantee valid inference. QML estimators also permit to estimate GLM-like models with mean-variance specifications that do not correspond to exponential family distributions. In fact, Wedderburn (Citation1974) uses the QML setting to estimate a logit with a squared binominal variance function.
19 Results are very similar to those obtained when we estimate the logit in the ‘more common’ way, that is, not in a QML framework but like a logistic regression. The same variables are statistically significant and the tests applied to assess the goodness of fit (Homer–Lemeshow and Pearson ) indicate a well-fitting model. Furthermore, the pseudo R2 has a value of 0.542 and the LR statistic indicates that we can reject the null hypothesis that all slope coefficients except the constant are zero.
20 We have also calculated (t − 10) control premia to test for the robustness of the results as we will see later. In this case, we compare the block price with the market price 10 days before the date t. Beware with the meaning of t, as for t − 1 t means the day in which the Stock Market Bulletin publishes that the transaction took place the day before. For t − 15 t means the earliest date among the possible sources of information about the deal: the CNMV official registry of Relevant Facts, the business press or the Stock Market Bulletin.
21 We consider outliers those premiums below the first quartile minus three times the interquartile range and above the third quartile plus three times the interquartile range.
22 Other tests for the equality of variances have also been considered and the results do not change in relation to those shown here.
23 This result does not change when we consider the three different market segments, that is, when comparing the block premiums of each segment at t − 15 and t − 1. All the results of the different tests are available from authors upon request.
24 The only exception is the case of the Block Market for t − 1 premiums of those transactions which are not related on the obtaining of a seat in the board that shows a significant discount on median (Wilcoxon range test).
25 For testing the significance of coefficients of interacted variables, the Wald test is used. The coefficient of the variable is the result of adding the coefficient of the single variable and the coefficient of the interacted variable.
26 For t − 15, the only difference found is that the coefficient of ‘ownership structure’ is no longer significantly different from zero when clustered errors are used. For t − 10 and t − 1, we also obtained results very similar to those from Huber–White standard errors. All these results are available from authors upon request.