The Survival of Family Firms: The Importance of Control and Family Ties

Lotti, Francesca; Santarelli, Enrico Working Paper The Survival of Family Firms: The Importance of Control and Family Ties Quaderni - Working Paper DSE, No. 461 Provided in Cooperation with: University of Bologna, Department of Economics Suggested Citation: Lotti, Francesca; Santarelli, Enrico (2002) : The Survival of Family Firms: The Importance of Control and Family Ties, Quaderni - Working Paper DSE, No. 461, Alma Mater Studiorum - Università di Bologna, Dipartimento di Scienze Economiche (DSE), Bologna, https://doi.org/10.6092/unibo/amsacta/4833 This Version is available at: https://hdl.handle.net/10419/159302 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc/3.0/ THE SURVIVAL OF FAMILY FIRMS: THE IMPORTANCE OF CONTROL AND FAMILY TIES1 by Francesca Lotti* and Enrico Santarelli** Abstract The aim of this paper is to analyze the survival patterns of a group of family firms which have already spent at least twenty-five years in the market. To this end, we use the Kaplan-Meier product limit estimator supplemented with qualitative information gathered by direct observation and discussions with entrepreneurs. The main findings of the paper are that small family firms which have reached their thirtieth year in the market face a very high risk of sudden exit, increasing with firm age. Further control carried out by means of interviews with entrepreneurs identifies problems connected with succession as one of the main causes of the decision to close down. Keywords: Family firms; Succession; Survival function; Kaplan-Meier estimator; Hazard function; Italy. JEL Classification: L20; C34; C41; M13. Corresponding author: Enrico Santarelli Department of Economics University of Bologna Strada Maggiore, 45 40125 Bologna ITALY [email protected].it 18 December 2002 1 We would like to thank Richard Caves, Matt Gentzkow, Giorgio Gobbi, Paolo Mistrulli, and Alberto Pozzolo for useful suggestions and comments. We thank the participants in the Industrial Organization Workshop at Harvard University (March 2002), the XXIX EARIE Conference (Madrid, September 2002) and in a seminar held at the Bank of Italy. Financial support from Assindustria Rimini and Fondazione Cassa di Risparmio di Rimini is gratefully acknowledged. Alessandra Giraldi provided outstanding research assistance. All remaining errors are our own. The opinions expressed do not necessarily reflect those of the Bank of Italy. * Bank of Italy, Research Department. ** University of Bologna, Economics Department. 2 1. Introduction For entrepreneurs who start a new firm, there sooner or later comes the moment when they decide, or are forced by circumstances, to retire. This decision gives rise to a succession problem which can be solved either by hiring professional managers or by appointing the founder's heirs to run the firm. If a professional manager or a heir is appointed, the founder (principal) can decide whether to stay and monitor him (which he usually does) or to give the agent acting on his behalf an incentive structure which is the same as his own, and thereby engage in delegation (Vickers, 1985). The resulting agent appointment game is more likely to be incentive compatible for the principal when he chooses an agent of his own type, namely a member of his family (who, as a natural consequence of family ties, can be supposed to have an incentive structure the same as his own). This solution to the appointment game is supported by our empirical evidence, which shows that those entrepreneurs who do not have heirs wishing to continue their activity (or do not have heirs at all) prefer to close down their businesses rather than hand over control of the firm they have created to an outsider. As a matter of fact, firms controlled by the entrepreneurs who started them often close down - rather than becoming Berle and Means (1932) corporations - when their founders are about to retire and no viable conditions exist for the persistence of family control after their retirement. According to the above theoretical considerations, succession may affect the likelihood of survival of family firms, even those characterized by the most favorable prospects of success. In-depth analysis of the succession event therefore provides the rationale for supporting the age dependence model put forward by Jovanovic (1982) (see also Cooley and Quadrini, 2001) which identifies age as the main dimension of heterogeneity among firms. When succession is considered, i.e. in the case of family firms which have already spent at least twenty-five years in the market, survival can be taken as dependent on age, since entrepreneurial firms experience different dynamics with respect to their managerial counterparts (which, by definition, are not affected by the succession problem) with the passing of time. Following the approach suggested by Borenstein et al. (1998), we combine the theoretical considerations put forward in this section with economic 3 evidence provided by statistical analysis and qualitative information gathered from discussions with entrepreneurs. Accordingly, Section 2 contains an extension of the results presented in Santarelli (2001) which uses the Kaplan-Meier product limit estimator to analyze the relationship between survival and age for a sample of small family firms in manufacturing, retailing, and the hospitality sector, the purpose being to identify whether around the thirtieth year in the market, i.e. when the succession problem usually manifests itself, this kind of firm is more likely to exit the market. Section 3 uses direct observation and interviews with entrepreneurs who have already faced or are facing the succession problem to set out the factors affecting their decisions concerning the future of the firm they have created. Finally, some concluding remarks are made in Section 4. 2. Data and Econometric Issues 2.1 Data From the empirical viewpoint, our first step was to study, by means of statistical analysis, the likelihood of survival after the 25th year following start-up of small family firms registered with the Chamber of Commerce of Rimini, in the Emilia-Romagna region of Italy, during the period 1950-1965 and which had survived for at least 25 years. The “family” nature of these firms is confirmed by the fact that they are all single proprietorship firms - which is clear evidence of unlimited liability - and that in 78 per cent of them at least one relative of the owner, besides the owner himself, occupies a crucial managerial position. At the end of 1999 such firms were aged between 34 and 49, and therefore had already faced (or were about to face) the succession event. Their small employment size is confirmed by the fact that at start-up time they had an average of 3.53 employees, whereas the current average size of surviving ones is 7.68 employees2. Thus, firms in our analysis satisfy the narrowest family business definition commonly accepted by scholars of family business (Astrachan and Shanker, 1996) which requires direct family involvement in daily operations, more than 2 At the industry level, average start-up size was 2.9 employees in manufacturing, 2.2 in retailing, and 6.0 in hospitality. In 1999, the average size of survivors was 8.5 employees in manufacturing, 3.7 in retailing, and 8.2 in hospitality. 4 one family member with significant management responsibility, and multiple generations involved. Emilia-Romagna is one of the most economically advanced Italian regions and is characterized by the large number of family firms, most of which belong to industrial districts (cf. Forni and Paba, 2002). For these reasons, the Province of Rimini, in the southeastern area of this region, is an ideal observatory for study of the patterns of survival of such firms. We focused our analysis on manufacturing, retailing, and hospitality services, which are still the most important activities in the local economy. The total number of single proprietorship firms registered during the 1950 - 1965 period and still alive after 25 years was 908, most of which (63.10%) were in the retailing sector, whereas manufacturing and hospitality services accounted for respectively 19.94% and 17.95% of the whole sample (Table 1). Nearly two thirds (62.44%) of the initial firms which survived for at least 25 years were still active at the end of 1999; that is, between 9 and 24 years after completion of their 25th year in the market. The percentages of survivors are high in all three industries, although hospitality services (70.55%) appeared to perform better than manufacturing and retailing. This is probably connected to the spatial agglomeration of hospitality firms in the Rimini area, which is favored by the endowment of tourist amenities (such as beaches, discotheques, amusement parks, etc.) and is likely to result in a higher likelihood of survival for firms in this industry. Table 1 NUMBER OF FIRMS REGISTERED IN THE CHAMBER OF COMMERCE FILES AND ACTIVE FOR AT LEAST 25 YEARS, AND NUMBER OF SURVIVORS IN 1999 Number of firms born in the 1950-’65 period and active for at least 25 years Percentage Number of firms still active in 1999 Percentage of Survivors Manufacturing 172 19.94 107 62.21 Retailing 573 63.10 345 60.21 Hospitality 163 17.95 115 70.55 Total 908 100 567 62.44 5 2.2 Econometric Issues Among many problems concerning survival analysis, it is impossible to make complete measurements of the life spans of all the subjects in the sample. Some individuals, firms in this case, can be dropped from the sample without their having necessarily exited from the market.3 In the case of business firms, this event may be voluntary liquidation, bankruptcy, or other defined adverse events. Accordingly, for each individual we can observe either the time to failure or the time to loss (or censoring). This means that for the censored individuals, we know only that the time to failure is greater than the censoring time. With these incomplete observations, the estimation of a survival function cannot be a simple description of the sample (Kaplan and Meier, 1958). In the presence of a complete sample one could easily estimate the empirical survivor function via maximum likelihood procedure as: (1) () () () tFtTtTtS −=≤−=>= 1Pr1Pr)( where T is the random variable representing failure time. This survival function gives the population probability of surviving beyond time t. Due to the censored nature of our sample, it is not possible to know the exact number of observations with duration greater than t. The Kaplan Meier estimator modifies the estimated survival function, keeping the maximum of the available information. Suppose that we have a sample of N individuals each with a time to failure, or lifetime, T1, T2,…, TN. If the observations are not complete, we can only have observed lifetimes, defined as: (2) () iii LTt ,min= Ni ,...,2,1= where Li represents the censoring time (or the limit of observation). These censoring times can be constants or values of other random variables: in any case, they must be independent4 from Ti. In this way, the individuals in the sample are divided into two 3 Of course, this can be due to several reasons. For instance, an individual can be dropped from a sample if its characteristrics change over time or simply because, at a certain point, the measurement has to be stopped. 4 Censoring times may differ from individual to individual or they may be the same, also according to the sampling techniques employed. 6 mutually exclusive classes: deaths (or failures) and losses. Let us partition the age scale into intervals (0, u1), (u1, u2), …, (uk-1, uk), and then let us denote with δj the number of pure exits and with λj the number of losses during interval (uj-1, uj). The relationship between N, nj, nj’, λ j, and δ j is described in Figure 1 below. Figure 1 Number of individuals N n1 n 1’ n2 n 2’ … Number of losses (λj) and deaths (δj) λ0 δ1 λ1 δ2 λ2 u0 u 1 u 2 … Division points of intervals Accordingly, the number of individuals still in life after uj-1 is denoted by nj and δ j, deaths are observed in the interval (uj-1, uj), and the estimated conditional probability for interval j is defined as: (3) j j j jj jn n n n p ' ˆ= − = δ Once we take the k intervals into account, the Product-Limit estimate (or Kaplan Meier Estimator) is given by: (4) () ∏ =        =k jj j n n tS 1 ' The resulting estimator turns out to be consistent for S(t) (Lancaster, 1990). Another important measure when dealing with survival analysis is the hazard function, or the age-specific failure rate. Let us define first the hazard rate as: (5) () () () () tS tf t TtttTt tt= ∆ ≤∆+<≤ =→∆ |Pr lim 0 λ where f(t) is the density function whose c.d.f. is () () tStF −=1. For each time t, the hazard rate λ(t) is the probability of exiting in the period tt ∆+ , given that the individual is still alive at time t. For our purpose, we use the integrated hazard function, defined as: 7 (6) () () ∫ =Λ tdttt 0 λ which can be consistently estimated by means of the Nelson-Aalen estimator (Aalen, 1978 and Nelson, 1972), defined as: (7) () ∑ =        =k jj j n tH 1 δ 3. Results The Kaplan-Meier estimator applied to the sole proprietorship firms born in the period between 1950 and 1965 and which have survived for at least 25 years in the territory of the Province of Rimini provides a clear picture of the time in the firm's lifetime at which mortality rates have the strongest impact. It should be pointed out that the focus of our analysis is the survival patterns of these firms just as a function of their age, independently of the period in which such firms were born. This approach has two main advantages: first, it does not capture any “cohort effect”, and second, the impact of the business cycle (if any) is spread over a longer period. Figure 2 reports the results for all the relevant industries. It will be observed that after about 30 years in the market (i.e. 5 years after the firm's 25th birthday), the likelihood of sudden exit starts to increase dramatically, suggesting the strong dependence of liquidations on the owner's retirement. Both the manufacturing and retailing sectors exhibit patterns of survival which do not differ significantly from those revealed by computation of KM for all industries. In manufacturing, however, the likelihood of exiting the market is much lower after the fortieth year following foundation, and this suggests that the manufacturing firms in our sample may have dealt with the succession event either by hiring professional managers or by leaving the firm in the hands of a direct heir. Conversely, in hospitality services (including hotels, restaurants, and catering firms), the mortality of aging firms starts to be very high just after their thirtieth year of activity, although in the following two decades the likelihood of survival remains higher than in the overall economy. 8 Figure 2 Kaplan-Meier survival estimates. 0 5 10 15 20 25 30 35 40 45 50 55 0 0.2 0.4 0.6 0.8 1 Age, in years Estimated Density Kaplan-Meier Survival Estimates All Industries Manufacturing Retailing Hospitality Threshold The results from estimation of the integrated hazard function are reported in Figure 3. Inspection of the hazard curves supports the evidence from the KM estimator. The probability of exit, given that the firm is still active in the previous period, tends to increase exponentially with age, after the 30th year in the market. Again, the hospitality industry is an exception, since it shows higher hazard rates between the 30th and the 35th years of age, but thereafter the risk of failure significantly decreases.