M&As, Investment and Financing Constraints

Abstract

We use a panel data set of European firms to analyse the effects of domestic and international M&As on target firms’ investment, growth and financial constraints. Combining propensity score matching with a difference-in-differences estimator, our results indicate that upon acquisition, target firms obtain better access to external finance, are characterized by higher levels of tangible and intangible assets, and display lower dependence of investments and cash savings on the availability of internal funds. We also provide evidence that these effects are driven by acquisitions during the 2007–2009 financial crisis and relatively small target firms.

Keywords:

• Mergers and Acquisitions
• Financial Constraints
• Investment
• Firm Growth
• Financial Crisis
• Foreign Ownership

JEL CLASSIFICATIONS:

• L2
• F61
• F23
• G01
• G34
• L25
• D22
• D24

Notes

Acknowledgements

We would like to thank two anonymous referees, Andrea Ciani, Anna Gumpert, Jens Suedekum, and seminar participants in Düsseldorf, Göttingen, Maastricht, Munich, and Vienna for helpful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 See, for instance, http://www.wsj.com/articles/2015-becomes-the-biggest-m-a-year-ever-1449187101 and https://hbr.org/2016/05/so-many-ma-deals-fail-because-companies-overlook-this-simple-strategy (accessed 23 June 2017).

2 Erel, Jang, and Weisbach (2015) and Wang and Wang (2015) relate M&As to proxies for financing constraints. Recent contributions that analyse the effects of M&As on innovation, prices, productivity and other outcome variables include Guadalupe, Kuzmina, and Thomas (2012); Ashenfelter, Hosken, and Weinberg (2014); Braguinsky et al. (2015); Javorcik and Poelhekke (2017), to mention a few. See section 2 for a detailed discussion of related literature.

3 Previous research has also shown that acquisitions by financial companies such as private equity firms can lead to lower financing constraints and induce higher investment and innovation (Amess, Stiebale, and Wright 2016; Boucly, Sraer, and Thesmar 2011), but it is unclear whether these mechanisms also apply to M&As in general.

4 To obtain an accurate measure of total factor productivity, our baseline results are restricted to targets in the manufacturing sector. However, we find that our main results also hold for service firms when we conduct a separate propensity score matching exercise using a measure of labour productivity.

5 Erel, Jang, and Weisbach (2015) control for some observable characteristics at the firm- and industry-level in their baseline specification and conduct a robustness check where they match non-acquired firms based on size, industry and country. However, they do not allow for endogenous selection based on productivity and pre-acquisition levels of outcome variables, which might be important determinants of acquisition decisions in our sample.

6 The productivity advantage of multinationals has, for instance, been related to management practices (e.g. Bloom and van Reenen 2010) and differences in innovation and knowledge (e.g. Guadalupe, Kuzmina, and Thomas 2012). Besides knowledge transfer, foreign acquisitions might also benefit acquisition targets due to access to new markets (Guadalupe, Kuzmina, and Thomas 2012) or complementary assets of the acquiring firm (e.g. Nocke and Yeaple 2007, 2008). Several empirical studies have documented significant performance gains in the form of productivity improvements in target firms after international M&As (e.g. Arnold and Javorcik 2009; Chen 2011; Guadalupe, Kuzmina, and Thomas 2012) while other scholars have argued that the effects of cross-border M&As are not that different from other ownership changes (e.g. Gugler et al. 2003; Fons-Rosen et al. 2013; Wang and Wang 2015). There is a large literature on the effects of M&As on efficiency-related outcomes which either analyses domestic transactions or does not explicitly distinguish between domestic and international M&As. This literature indicates that domestic acquisitions can lead to productivity gains as well (e.g. Maksimovic and Phillips 2001; David 2013; Braguinsky et al. 2015).

7 We use the Gross domestic product at market price deflator for EU-28 countries.

8 We also distinguish between this rather broad and a more narrow crisis period in some regressions.

9 The distribution of target firms across countries is depicted in Table A.1 in the Appendix. Our sample of acquisitions is smaller compared with Erel, Jang, and Weisbach (2015) due to our focus on manufacturing industries and our much broader set of control variables which have missing values for some deals.

10 For variable definitions see Table A.2 in the Appendix.

11 The choice of matching with or without replacement can be seen as a trade-off between bias and variance. Since our sample of potential firms in the comparison group is large, we decide to perform matching without replacement.

12 There is no need to cluster standard errors by firms since equations (4)(4) $$\Delta y_{\mathit{it}+s}=\alpha +{\beta }_{\mathit{DiD}}M{A}_{\mathit{it}}+{ϵ}_{\mathit{it}}$$(4) and (5)(5) $$\Delta y_{\mathit{it}+s}={\alpha }_0+{\beta }_{1,\text{DiD}}M{A}_{\mathit{it}}+{\beta }_{2,\text{DiD}}(M{A}_{\mathit{it}}\times \mathit{smal}{l}_i)+{\beta }_{3,\text{DiD}}s\mathit{mal}{l}_i+{\eta }_{\mathit{it}}$$(5) are estimated separately by time period and therefore do not include repeated firm observations.

13 Empirical work, such as Opler et al. (1999), shows a negative relation between access to the capital market and the amount of cash held by the firm. More recently, Hadlock and Pierce (2010) also find that firms with more cash are more likely to be financially constrained.

14 Kim, Mauer, and Sherman (1998) test this explanation and provide evidence that the leverage ratio is negatively related to liquid assets.

15 We argue that increased leverage after acquisition is consistent with relaxed credit constraints as firms are able to rely more on external financial funds, in line with, for example, Bellone et al. (2010) and Boucly, Sraer, and Thesmar (2011). In contrast, Wang and Wang (2015) interpret a decrease in the leverage ratio and an increase in the liquidity ratio as a reduction in financing constraints.

16 Other authors, e.g. Khurana, Martin, and Pereira (2006) and Hadlock and Pierce (2010), also provide empirical evidence that the cash flow sensitivity of cash is related to a firm’s financing constraints.

17 However, their approach has been criticized by Riddick and Whited (2009). They show in a dynamic framework that cash holdings are in fact negatively related to cash flow when accounting for measurement error in Tobin’s q. In addition, they argue that the amount of cash savings is not only related to a firm’s financing constraints, but also (and to a greater extent) to its income uncertainty.

18 Note that the common support condition is fulfilled for all acquired firms in the sample.

19 This is also true for industry and country dummies which are not displayed separately in the table. An F test for joint equality of industry and country dummies produced a p-value of 0.965.

20 See Table A.3 in the Appendix where we separately analyse the change in long-term debt and current liabilities, the two components of leverage.

21 As a robustness check, we perform the DiD estimation for the log of intangible assets as an outcome variable (see Table A.3 in the Appendix). Since around 22% of acquired firms have zero intangible assets in the pre-acquisition year, the log of intangible assets is calculated as ln(Intangible Assets + 1). The estimates are highly significant and similar in magnitude.

22 The main results are very similar when taking into account macroeconomic fluctuations with the inclusion of country-year fixed effects in the DiD estimation (see Appendix Table A.4).

23 See Appendix Tables A.6 and A.7 for the results of the Probit estimation and the propensity score matching. Note that, in contrast to the manufacturing sample, we include a firm’s lagged labour productivity (i.e. sales over the number of employees) instead of TFP in the Probit estimation to estimate the propensity score.

24 Again, we perform a one-to-one nearest neighbour matching without replacement. Firms not involved in M&As are used as a control group. The Probit estimations and balancing tests are given in Tables A.9, A.10 and A.11 in the Appendix.

25 These results are reinforced when calculating heterogeneous effects by the targets’ pre-acquisition size (see Table A.12 in the Appendix). We find that part of the estimated effects in the crisis are driven by small target firms, which are likely to be particularly financially constrained. However, some of these estimates are quite noisy, probably due to the relatively small number of acquisitions during crisis periods. We also experimented with heterogeneous treatment effects according to pre-acquisition leverage and cash reserves following Duchin, Ozbas, and Sensoy (2010). However, these additional tests did not indicate a robust pattern. A likely explanation is that in contrast to the sample used in Duchin, Ozbas, and Sensoy (2010), our sample contains many non-listed firms. Within our heterogeneous group of target firms, firm size seems to be a more important predictor of financial constraints than pre-acquisition leverage and cash.

26 We would like to thank an anonymous referee for suggesting this additional test.

27 We exploit equation (5)(5) $$\Delta y_{\mathit{it}+s}={\alpha }_0+{\beta }_{1,\text{DiD}}M{A}_{\mathit{it}}+{\beta }_{2,\text{DiD}}(M{A}_{\mathit{it}}\times \mathit{smal}{l}_i)+{\beta }_{3,\text{DiD}}s\mathit{mal}{l}_i+{\eta }_{\mathit{it}}$$(5) replacing small with an indicator variable intra which is equal to 1 if the acquiring and the target firm operate in the same two-digit NACE Rev. 2 industry. The regressions include the interaction term MA × intra but not the dummy variable intra itself as this variable is not available for control firms by construction.

28 For about 3% of acquired firms (=20 deals), we have no information on whether the deal is cross-border or domestic. These observations are excluded from this analysis.

29 Again, we apply one-to-one nearest neighbour matching without replacement using non-acquired firms as a control group. The balancing condition is tested in Table A.14 in the Appendix.

30 We also analysed if the effects for cross-border deals differ by the acquirers’ country or region but did not find a systematic pattern of heterogeneity.

31 As the number of potential control firms (i.e. domestic-acquired firms) is only somewhat larger than the number of treated firms (i.e. foreign-acquired firms), we allow for propensity score matching with replacement. In addition, the common support condition is imposed, i.e. foreign-acquired firms which are of common support are not included (corresponds to 30 deals in our sample). Tables A.15 and A.16 in the Appendix display the results of the Probit estimation and the propensity score matching.

32 More precisely, we perform the DiD estimation in equation (4)(4) $$\Delta y_{\mathit{it}+s}=\alpha +{\beta }_{\mathit{DiD}}M{A}_{\mathit{it}}+{ϵ}_{\mathit{it}}$$(4) , while additionally controlling for the log of age and the cash flow ratio in period t–1, as well as their respective interaction with the treatment indicator MA (results available upon request).

33 To estimate separate productivity effects for different deal types and time periods, we draw on the results of the separate matching procedures (see Chapter 5.1).

34 The key variables include cash ratio, leverage ratio, capital stock, employment, and intangible fixed assets over fixed assets.

35 .This applies to four pairs of acquired and matched control firms. Unfortunately, for 276 firms (196 acquired firms, 171 control firms) we do not have information on their region.

36 We would like to thank an anonymous referee for suggesting this additional robustness test.

37 We obtained similar results when we proxy size with sales instead of employment (results available upon request).