Family Ownership Influence on Cost Elasticity

Abstract

AbstractThis study explores the relation between family ownership and cost elasticity. Using a sample of 1746 European firms, we first find that family ownership, a prevalent ownership type with unique characteristics, is associated with greater cost elasticity. Further, we use four empirical settings to increase our confidence that a higher cost elasticity is attributable to family ownership. We also document that family firms achieve greater operating cost elasticity primarily through modifying SG&A costs in response to changing sales, but not by hiring or firing employees. These findings extend prior studies on ownership effects on cost structures, suggesting that family ownership matters in understanding firms’ cost elasticity choices.Keywords: Cost behaviourCost elasticityOwnershipFamily firmsJEL Codes: M41D24L23D10 AcknowledgementsWe gratefully acknowledge constructive comments from Ashiq Ali, Eli Amir, Dan Amiram, Yakov Amihud, Eli Bartov, Jan Bouwens, Verdan Capkun, Mark DeFond, Shane Dikolli, Shai Levi, Luc Paugam, Suresh Radhakrishnan, Herve Stolowy, and participants of the 2019 Financial and Managerial Accounting Research Conference, 2020 MAS Midyear Meetings, JMAR Online Brownbag, and seminar participants at Bocconi University, HEC Paris, University of Florence. We acknowledge financial support from our respective institutions and from the Raya Strauss Center of Family Business Research. Any errors are entirely our own.Disclosure StatementNo potential conflict of interest was reported by the author(s).Notes1 Cost elasticity is the percentage change in costs in response to percentage change in activity level (or volume).2 See Kreps (Citation1990) and Varian (Citation2014).3 Family businesses generate an estimated 70–90% of global GDP annually (Family Firm Institute, Citation2017), while 85% of startups worldwide are established with family money (Family Firm Institute, Citation2017 – Global Data Points).4 The Bvd Electronic Publishing (BvDEP) Ownership Database is a source of owner and subsidiary links worldwide.5 Information on proprietary linkages from BvD is collected directly from official bodies, if any, or from national and international providers. In case of conflicting information among providers covering the same country, the Ownership Database is updated according to the latest available report.6 In these cases, the GUO type is “Industrial Company” (type C), “Foundation/Research Institute” (type J), and “Mutual & Pension Fund/Nominee/Trust/Trustee” (type E). Following Faccio et al. (Citation2011), we exclude 69 firms for which the government is the GUO (type S, i.e., public authority, state, government), because governments’ risk-taking preferences are typically different from those of private investors (Faccio et al., Citation2011).7 According to DeFond et al. (Citation2017, p. 3630), “CEM is essentially a variation of exact matching, but instead of matching on the exact values of covariates, it matches on a coarsened range (or strata) of covariates. By stratifying covariates, CEM alleviates the significant demands that exact matching imposes on the data. CEM also directly matches on the multivariate distributions of the covariates instead of matching on a single scalar (i.e., propensity score). As a result, CEM does not rely on the functional form and discriminative ability of a first-stage propensity score model, and considers higher moments of the covariate distributions.”8 Matching by country and industry allows us to account for (invariant) unobserved heterogeneity within countries and industries, which subsumes common factors such as quality of institutions, labor regulations, and idiosyncratic industry characteristics and therefore could affect our results.9 Our matching procedure follows Barber and Lyon (Citation1996). They advise researchers to use a performance variable (e.g., earnings) as a matching variable and argue that commonly used research designs (including matching by using firms’ attributes other than performance) yield test statistics that are misspecified in cases where sample firms have performed either unusually well or poorly. Particularly, they find that matching treated firms to control firms on industry and performance is generally more important than matching them on industry alone (or on only country or size). Huson et al. (Citation2004) advocate the matching method suggested by Barber and Lyon because it controls for potential mean reversion of accounting performance time series, which may affect measures of change variables (such as sales) in our study.10 We replicate the analyses using the actual values of the variables rather than their natural logarithm. The results are essentially the same.11 In sensitivity tests, we extend the model in equation (1) to control for a potential differential influence of firm age, operating cycle, national cultural dimension of long-term orientation as in Kitching et al. (Citation2016), and country legal protection using the investor right index from the World Bank’s website. We also estimated equation (1) using lagged sales as the denominator of the percentage change in operating costs, as suggested by Balakrishnan et al. (Citation2014). All the findings (unreported for brevity) further support our hypothesis. In line with prior cross-country studies, however, we cannot perfectly rule out that some country-specific features affect our findings. To partially alleviate this concern, we test whether the findings are driven by the observations from the UK, which represent 26% of the entire sample. Results from estimating equation (1) excluding observations from the UK (unreported for brevity) remain unaffected.12 We cannot rule out, however, that other unobservable variables predicting both the choice of the successor and the subsequent change in cost elasticity may exist. Also, the study is limited in ruling out other explanations: for example, a change of cost elasticity in family firms after a death event might be the result of the founder’s traits and managerial ability, not only family ownership.13 Aboody et al. (Citation2018) require at least eight years of data for each firm, Kallapur and Eldenburg (Citation2005) require at least 16 years of data to be available for each hospital department, and Lev (Citation1974) requires at least 20 years of data for each firm.14 We use the z-statistic for testing the hypothesis (β3 full sample = β3 sample excluding founder firms). The result from the statistical test indicates rejection at the 1% significance level following Holzhacker et al. (Citation2015b, Table 7) in using the z-statistic to test for differences across groups, as in Clogg et al. (Citation1995) and Cohen (Citation1983).15 We acknowledge that founder firms can differ in other ways that affect cost elasticity. They are likely to be younger firms in earlier lifecycle stages, they are likely to focus on growth, and the founder might be more overconfident than subsequent generations. Some of these differences encourage family firms to take risks, working against our hypothesis.16 The incidence of eponymy in our sample, 17.8% (150 / (150 + 688)), is similar to that reported in Belenzon et al. (Citation2017), i.e., 19%.17 As a robustness test, we also match 1,080 eponymous family firm observations with the corresponding 1,080 non-eponymous non-family firm observations. The coefficient estimate on the interaction term, β3, is 0.100% per 1% change in sales, and is significant at the 5% level, supporting our conclusion.18 We test the hypothesis (β1+β3)FF = (β1+β3)NON-FF at the 1% significance level following Holzhacker et al. (Citation2015b, Table 7) by using the z-statistic to test for differences across groups, as in Clogg et al. (Citation1995) and Cohen (Citation1983). We also estimate a triple-interaction model using the full sample for testing the hypothesis (β1+β3)FF = (β1+β3)NON-FF. Untabulated results confirm our results.19 This study focuses on cost elasticity and our partial insights on cost stickiness are limited. Prabowo (Citation2019) documented SG&A cost stickiness in US family firms. Recently, Abudy and Shust (Citation2022) used a small Israeli sample and report anti-sticky SG&A costs in family firms. SG&A costs reflect different resources than operating costs, and our large sample of 13 EU countries does not include either US nor Israeli firms. Additional research is necessary to examine cost stickiness in family firms.20 As a real-life example of actions that firms take to modify their cost structure, Guess, a family-owned apparel marketer founded by the Marciano family in 1981, reports strengthening its “ability to control variable costs such as cost of sales and payroll and, in some cases, renegotiate lease costs” (Guess 10-K 2017, 56).