Cost (In)Efficiency and Institutional Pressures in Nursing Home Chains

Abstract

Over the past 10 years, merger activities in the private for-profit nursing home industry have been increasing in Europe. In this paper, we investigate chain affiliation’s influence on the performance of lucrative nursing homes. We measure performance using a cost frontier estimated by stochastic analysis on a sample of 370 French for-profit nursing homes. We find that cost efficiency decreases with the number of facilities in a chain. We also identify different external actor types in nursing homes’ institutional environment and test their influence. We show that nursing home chains’ cost efficiency improves when local governments and shareholders exert pressure. Our results are robust to alternative model specifications and another definition of costs. Overall, our findings inform researchers, as well as standards setters, of the relevance of chain affiliation and of the role of institutional pressures regarding cost containment at the nursing home level.

Acknowledgements

We are grateful to our anonymous reviewer and the two editors for their helpful suggestions. We wish to thank Mélina Ramos-Gorand for sharing her skills in geography and for providing us with several variables included in this paper. We also thank Karen Earl-Erpelding and Simona Kepalaite for their editing help.

The research was carried out at Direction de la Recherche, des Études, de l’Évaluation et des Statistiques (DREES), under a contract ‘convention prêt de données’. It allowed us to access most of data used in this paper.

Notes

¹Department for Research, Studies, Assessment, and Statistics.

²For-profit organizations own almost one-third of beds created between 2007 and 2011 (DREES, 2014).

³For-profit ownership accounts for 50% of nursing homes in Canada (Comondore et al., 2009), 67% in the USA (Grabowski, Feng, Hirth, Rahman, & Mor, 2013), and 76.8% in the UK (Forder & Allan, 2011).

⁴See Sections 2.2 and 3.2 for more details.

⁵Some studies investigate differences in performance across the non-profit category (see Dewaelheyns et al., 2009 or Eldenburg & Krishnan, 2003, for instance).

⁶Source: own computation from Ernst and Young (2008) and 2011 Établissements d'Hébergement pour Personnes Âgées (EHPA) survey (DREES).

⁷Source: own computation from Ernst and Young (2008) and 2011 EHPA survey (DREES).

⁸Relationships with local authorities are considered to be strategic. Several large groups possess internal services entirely dedicated to these activities.

⁹In 2012, Orpea acquired Artevida in Spain and, in 2014, Senevita in Switzerland and Silver Care in Germany.

¹⁰Source: own computation from 2011 EHPA survey (DREES).

¹¹Some facilities are entirely regulated: their lodging price is also set according to a cost-based payment mechanism. Several non-profit nursing homes are in this situation. Conversely, only very few for-profit nursing homes are totally regulated. To ensure relevant comparisons across for-profit nursing homes, we omitted entirely regulated nursing homes from our final sample.

¹²Source: Les Échos, 278, April 2011.

¹³A Conseil Général operates at the county (département) level. Since 2015, a Conseil Général is named a Conseil Départemental. The French metropolitan territory is divided into 96 départements.

¹⁴For instance, Orpea has been listed on Euronext Paris since April 2002, Le Noble Age since June 2006, Korian since November 2006, and Medica since February 2010 (Korian and Medica merged in March 2014).

¹⁵Top managers receive stock options and other remunerations linked to the share price in compensation packages.

¹⁶French National Institute for Statistics and Economic Studies.

¹⁷ and are estimated using the formulation suggested by Jondrow, Knox Lovell, Masterov, and Schmidt (1982).

¹⁸Explanatory environmental variables of inefficiency are, depending on the regressions, either the chain size or variables in the institutional environment.

¹⁹This approach is better than a two-step one where the following steps are successively performed. First, a stochastic cost frontier is estimated, assuming that inefficiency scores are independently and identically distributed. Second, inefficiency scores are regressed on a collection of explanatory variables. The second step is just to contradict the assumption of independent inefficiency terms, which could lead to inefficient estimates (Battese & Coelli, 1995) or biased estimates of inefficiency scores (Wang & Schmidt, 2002).

²⁰Support staff members are low-skilled and provide no care services.

²¹When they are present, wages of high-skilled managers (e.g. physicians) represent a substantial part of staff costs. However, under certain conditions, their use of accounting information systems for cost control, and thus their increased cost consciousness, may lower costs (Abernethy & Vagnoni, 2004).

²²These residents benefit from the Health Area Social Allowance (see Section 3.2).

²³To isolate the effect of building aging and equipment aging, we conducted a robustness check where ActivityLength is replaced by BuildAge (see Section 5.4).

²⁴With the global option, the reimbursed health benefit basket is large, because it includes physicians’ fees, radiography, and biological analyses. With the partial option, residents first pay for these medical care services and are then reimbursed by their own health insurance.

²⁵Using another sample of facilities in France, Dormont and Martin (2012) show that they are significantly correlated with residents’ health status.

²⁶We do not include replacement staff, so as not to assimilate high-quality and high-absenteeism level.

²⁷The corresponding needs are established for the formal home care sector. The quantity of full-time equivalent staff members needed per day is equal to 1 for a RUG 1 person, 0.84 for a RUG 2 person, 0.66 for a RUG 3 person, 0.42 for a RUG 4 person, 0.25 for a RUG 5 person, and 0.07 for a RUG 6 person (Ratte & Imbaud, 2011).

²⁸We excluded nurses and physicians.

²⁹This category includes nursing homes from the three largest groups in France: Korian-Medica, DVD, and Orpea.

³⁰We could also have considered used capacity (rather than mere capacity) to build the MarketShare variable. However, including such a variable could generate an endogeneity bias (Gulley & Santerre, 2007). In addition, this variable differs very little from the capacity we use. Indeed, the average occupancy rate was equal to 94.1% in the private for-profit sector in 2011 (Martin, 2014).

³¹The (by road) 20 km radius seems to represent the size of the relevant market. When for-profit nursing homes are considered, 75% of the residents head toward nursing homes located within a 21 km radius (Martin, 2014).

³²Independent nursing homes are used as the reference category.

³³This is a relative score computed from the comparison of for-profit nursing homes. The comparison of these institutions with public or private non-profit facilities could lead to higher or lower inefficiency scores. However, such a comparison is not possible, since we do not have homogeneous cost data for all statuses.

³⁴The average price for a resident-day is €93.9 for independent nursing homes, €107.6 for medium-sized groups, and €119.7 for large chains. Sources: own computation from DIANE and 2011 EHPA survey (DREES) (sample of 370 nursing homes).