Access to finance, foreign ownership and foreign takeovers in Germany

Abstract

With this article we present the first microeconometric analysis of the impact of a foreign acquisition on the target firm’s access to finance. By using a large database of German firms, we furthermore investigate for the first time the link between foreign ownership and access to finance in Germany, one of the world's leading target countries for FDI. We use newly available comprehensive panel data that we constructed from information collected by the German statistical offices and from credit rating scores supplied by the leading German credit rating agency. We find foreign-owned firms in German manufacturing on average to show slightly more financing restrictions than domestically owned enterprises, but this very small difference diminishes once unobserved heterogeneity is taken into account. We further demonstrate that one reason for this finding is the preference of foreign investors for targets with relatively low credit-worthiness. Although the likelihood of a foreign acquisition appears to be correlated with credit rating, there is no impact of foreign takeovers on the credit constraints of the target firms ex post and therefore no support for the hypothesis that foreign takeovers ease financial frictions.

Keywords:

• credit constraints
• foreign ownership
• acquisitions
• Germany

JEL Classification:

• F21
• F23
• G34

Notes

¹ This effect cannot be observed for subsidiaries as they are generally less constrained.

² This data is confidential in the sense that all computations had to be performed via remote access within the Research Data Center of the statistical office Berlin-Brandeburg and single cases could not be identified. The accessibility of this data is described in detail by Zühlke et al. (2004).

³ Some authors refer to credit constraints as the result of frictions on financial markets and therefore as a result of information asymmetries. This is not the case in this analysis as we assume the credit rating score to reflect the true default risk and the term constrained is to be understood relative to other firms' ratings and not relative to their true credit-worthiness.

⁴ For a discussion of the usefulness of investment-cash flow sensitivities in particular, see Kaplan and Zingales (1997).

⁵ The terms foreign-controlled, foreign owned and foreign are used interchangeably in this text.

⁶ Indirect control refers to the fact that enterprise A is controlled by enterprise B and both are domestic companies but enterprise B is, in turn, controlled from an entity abroad. Then, enterprise A will also be foreign controlled. Effective minority control is stated when several minority owners with shares of more than 50% in sum act in concert.

⁷ In the initial sample, the average number of employees per firm in 2010 is 93 (with a SD of 826). The average number of employees for the matched final sample with only firms that are rated in at least one year is 470 for 2010 (with a SD of 2668). These numbers do not change much across the other year cross sections. This difference demonstrates the considerable size bias in the selection of our sample which needs to be considered in the interpretation of our results; another issue is a potential selection bias due to the restriction of our sample to only those firms that report basic figures, such as the number of employees, in every year. The reason is that we are interested in structural differences between ‘normally’ operating firms and therefore welcome the fact that firms that have just entered the market or are about to exit the market are not part of our analysis.

⁸ Therefore, we can restrict the comparison to only dependent affiliates and take into account a general network effect. Unfortunately, we do not have data on the multinational status of the domestically owned firms and cannot distinguish national from international networks.

⁹ Descriptive statistics for these variables can be found in Table A.1.

¹⁰ Although the global financial and economic crisis started at the end of 2008, its major impact unfolded in the year 2009. We therefore consider 2009 as the first year of the crisis.

¹¹ Remember that an increasing rating means a worsening of the credit-worthiness.

¹² We also ran all regressions with more firm-level control variables for robustness reasons but the results only change marginally. The results are reported in Table A.2. Our results are also robust if we replace the number of employees by total turnover as an alternative measure of firm size. Results are reported in Table A.3.

¹³ A reason could again be the disproportionate share of large firms in our data.

¹⁴ The breakdown point gives the level to which an estimator is resistant to outliers. For the MM-estimator, the breakdown point is at 50% (for OLS it is 0%). Since the bias increases with efficiency, the authors recommend an efficiency parameter of 0.7 to achieve an optimized combination of both low bias and high efficiency. The parameter can be set in the Stata ado file ‘mmregress’ that is provided by Verardi and Croux (2009) and has been used for our analysis.

¹⁵ Verardi and Wagner (2011) also provide the Stata ado file ‘xtregrob’ to apply the robust FE estimator.

¹⁶ It needs to be stressed here that although our FE regressions also use ownership changes as the identification strategy, considering only takeovers is a different analysis mainly because we do not consider divestments anymore (changes from foreign to domestic ownership) and we explicitly exclude potentially misidentified takeovers (cf. Section II).

¹⁷ Descriptive statistics, the definition and the identification of foreign takeovers are provided in Section II. Note that in order to keep a sufficient number of observations we no longer distinguish between eastern and western Germany in the analysis of takeovers.

¹⁸ An alternative would be to calculate the marginal effects at the sample mean (MEM), but since MEMs are calculated only for one specific hypothetical case, which is the sample mean, they do not seem to be appropriate in our case as we explicitly assume heterogeneous effects along the credit score distribution. For a detailed discussion of alternative ways of calculating marginal effects and application in Stata, see Buch et al. (2010) and Cameron and Trivedi (2010, 343ff).

¹⁹ These numbers were calculated from Table 3 in combination with Table 2.

²⁰ Note that we are not taking into account a general takeover effect which may occur independently of the buyer’s nationality as we compare foreign takeover targets to nontakeover targets. To consider a general effect of acquisitions (e.g. through a disciplining of the management), we would have to compare foreign takeovers to domestic takeovers but this is not possible with the data at hand.

²¹ For the estimation of this model, we use the Stata command provided by Villa (2012).

²² For a more detailed discussion of causal effects in a treatment analysis context with nonexperimental data, see Imbens and Wooldridge (2009) as well as Angrist and Pischke (2009).

²³ A major drawback here is that we cannot claim to consider all confounding factors to reach a conditional independence of the outcome and the assignment, mainly for two reasons: firstly, there is no theory predicting the exact determinants of a company acquisition, and, secondly, the data is not rich enough to account for all possible confounding factors.