The contracting of health care providers by health insurers based on the
Transaction Cost Theory of Management Control: a case study.
1. Introduction
At the 1st of January 2006, the system of Dutch healthcare totally changed due to the introduction of the Health Insurance Act (ZVW) and the Market Regulation Act. In the new system, the health insurer has an important role to play. Health insurers have to purchase high quality health care for low prices. To do so, the health insurer can offer customers an health insurance against a good price and quality, which is important to to achieve a growth rate of insured. For health insurers, health care purchasing is seen as one of the main goals which was confirmed by the Dutch minister of healthcare on the 6th of June 2014. This was in a discussion about the change of article 13 ZVW in which, from 2016, is arranged that health insurers get the possibility not to contract providers in secondary care (which is specialist-medical and specialist mental health services) anymore, and consequently not have to pay for not contracted care. At the NOS-Radio 1 journaal the Dutch minister of healthcare claimed:“The reason for existence of health insurers is the ability of them to purchase good quality care against a good price. With this they can keep the care affordable in the Netherlands.”
This quote shows the important role health insurers play in controlling the costs of health care in the Netherlands. Although this is an important role, it seems that the health insurers didn’t fulfill this important role as expected yet. The expenses of health care only rose since 2006 (Zorgkiezer, 2011). A way health insurers could fulfil this role is selective care purchasing. The idea of selective care purchasing is that health insurers oblige health care providers to deliver the quality that clients need and demand (RVZ, 2008). The change of article 13 ZVW is an important step for this. Selective care purchasing by health insurers is also one of the macro control instruments of the coalition Rutte II, to keep health care affordable (Schut and Varkevisser, 2013). Since the introduction of the ZVW in 2006, several changes within the ZVW took place. This includes changes of the package of care, for example the including of mental health in 2008 in the ZVW, the higher risk that health insurers bear by reducing the ex post risk adjustment mechanisms (Douven, 2010; Canoy et al., 2011) and the change of the regulations and methods of financing the care. For example the introduction of the financing of specialist-medical care by DOTs in 2012, the experiment with the free prices between dentists and health insurers and recently the introduction of the basic mental health and specialized mental health. All these changes affect the role the health insurer can take as health care purchaser, for the insured and, as it is seen by politicians, as an instrument to control the costs of health care. In this thesis, the role of the health insurer as health care purchaser is studied. This is executed by studying contracts between health insurer and health care provider, which are seen as part of a Management Control System (MCS) and conducting interviews with health care purchasers and experts from Health Insurer Eno. The most important elements of MCSs are collecting and using information to influence behavior and to evaluate performance (Horngren et al., 2005). This should be done by choosing and drawing the most effective MCS in the contract. In this thesis, it will be studied whether health care insurers use the most effective MCS in their contracts, based on the archetypes arm’s length control, machine control, exploratory control and boundary control of TCE-MC (Speklé, 2001). Important is, how the health insurer sees the purchasing of health care and what the difficulties are by contracting health care providers. If the contract is constructed in the most effective way, it can be an indicator if the health insurer can fulfill the role Dutch government is expecting from them.
2. The theoretical model
In this chapter, I describe MCSs in general. This is important for a proper understanding of the main theory of this thesis’ theoretical background: Transaction Cost Theory of Management Control(TCE-MC). Thereafter, I introduce generic Transaction Cost Economics (TCE) - reasoning. This reasoning can be interpreted as the foundation for TCE-MC, which will subsequently be described. Next to this, I introduce the types of health care that are studied in relation to the dimensions of TCE-MC. To conclude, I draw hypotheses about the most effective archetype of control with respect to the contracts between a health insurer and health care providers.2.1 Management control: what is it?
Another example are the control packages of Hofstede (1981). He defines six types of control of which the applicability is related to the organizational activities. The six types of control are: routine control, expert control, trial and error control, intuitive control, judgemental control, and political control. A limitation of the two studies mentioned, is that it still isn’t clear how to balance the different control forms/instruments in a given situation in order to establish an effective control system (Kruis, 2008). Speklé (2001a) realized that the problem of choosing the most effective governance structure also plays at organizational level and developed TCE-MC. As TCE-MC offers the possibility of choosing the most effective governance structure, in which the different control forms/ instruments are already balanced, in this study TCE-MC is used. This theory will be introduced in the next paragraph. As in this thesis it is the aim to study contracts, which lies outside the organization, the question arises if MC/ MCS can be applied also outside an organization. As Otley (1994) observed that “the scope of the activity of management control is enlarged and it no longer confines within the legal boundaries of the organization ”, which is also confirmed by the studies of Speklé (2001a, 2001b) and Vosselman (2002), there are no problems expected by applying MCS reasoning on contracts that are applied outside the organization. To be more specific, the current study focus is on contracting between health insurers and health care providers. As I described MCS in this paragraph and the possibility to apply MC reasoning outside the boundaries of an organization, it can be concluded that the theory is appropriate to use in the scope of this research. In the next paragraph, I introduce the Transaction Cost Theory of Management Control, which is used to balance the control instruments in order to establish effective control.
2.2 Transaction Cost Theory of MC
2.2.1 Transaction Cost Economics
can be on purpose misleading or deceiving. Some ways of control, which can prevent opportunistic behavior are a focus on monitoring and supervision of detailed norms and standards or benchmarking. When opportunism arises within a very complex environment, it will be difficult to discover opportunistic behavior. In this cases a common interest, which is costly to lose, is an option of control. Within the environmental characteristics there can be a distinction made between three characteristics: (1) asset specificity, in which degree human, material and capital inputs can be used for other purposes than originally intended, (2) uncertainty, unpredictable or unanticipated events that interfere with the ability to fulfill the terms of the transaction, and (3) the frequency, how often the transaction is repeated. The human dimensions and the environmental characteristics predict in which way the transactions can be coordinated in the most efficient way.
TCE furthermore features three generic clusters of governance which can be used to coordinate transactions: (1) markets, (2) hierarchies, and (3) hybrids.
In a market, the coordination of transactions is established horizontally between organizations with the price. This is done by a classical contract, while the price reflects all the information about the transaction. In the market there is a lot of competition, which is the incentive for an organization to deliver high performance in quality and costs. When the contract partner doesn’t deliver the right performance, the other partner can change immediately to another partner.
In a hierarchy, the coordination of transactions is established vertically within one organization with administrative controls. While there is no possibility to change contract partners, there is no price mechanism and another way of transaction coordination has to be used. This is done by administrative controls; these include written policies, employee training, monitoring and career rewards and punishments. A hybrid is a governance structure in which both horizontal and vertical coordination is used for the transactions between two separate organizations. In this case, additional institutional arrangements are used next to the contract. This is especially necessary when the price doesn’t contain all the necessary information. These transactions are in general complex and there exist more uncertainty about the transaction. Examples of institutional arrangements are the duration of the contract, quality agreements and reporting about performance. Speklé (2001a, 2001b) has used TCE to develop the Transaction Cost Theory of MC, in which he describes how, depending on the situation, to choose the most effective governance structure. He introduces four archetypes which are explained in the next section. Speklé (2001a, 2001b) uses the transaction characteristics described in TCE before, although he uses them slightly different.
2.2.2 Transaction Cost Theory of Management Control
of the transaction. When characterizing the activities of the transaction the best fitting archetype can be determined and in this way the most effective form of control.
Figure 1 provides an overview of TCE-MC.
Figure 1: Matching control problems with control solutions (Kruis, 2008)
The three dimensions which define the control problems are: (1) uncertainty, (2) asset specificity, and (3) ex post information asymmetry. These dimensions will be described in more detail in the next subparagraph.
2.2.3 The activities which define the control problems
The framework of Speklé is built on the human dimensions of TCE (Bounded rationality and Opportunism) and on three dimensions which define the nature of the activities which have to be controlled and the control problems to which they give rise. These three dimensions are also based on the TCE theory, but instead of the dimension ‘frequency’ Speklé introduces ex post information asymmetry. The dimensions of Speklé will now be described.
The first dimension is uncertainty, or as Speklé (2001a, 2001b) describes; the extent to which the activities and desired contributions are amenable to ex ante programming. Put differently: how much knowledge is available about the best way to carry out the activity and how much knowledge is available about what results to expect. When the uncertainty is high, this knowledge isn’t available and when the uncertainty is low, this knowledge is available. Uncertainty can be grouped in two categories: (1) programmable activities, and (2) non-programmable activities. Programmable activities are activities for which the organization possesses sufficient knowledge and information to decide in advance on how they have to be executed in order to achieve success, or activities for which the outcomes that may realistically be expected can be defined ex ante. Uncertainty is low in these cases. Non-programmable activities are activities for which the organization lacks knowledge and experience to relate them to outcomes. Uncertainty is high in this case.
The third dimension is the level of ex post information asymmetry, i.e. the extent to which the organization is able to observe and to assess perceptively the true quality of actually delivered contributions, with hindsight (Speklé, 2004). This dimension is only of interest when uncertainty is high. In case of ex post information asymmetry there isn’t enough information available about the activities performed. Low ex post information asymmetry can dissolve over time, when the organization is learning from the activities, which are performed, and obtains knowledge about the activities. When this knowledge is commonly known in the organization (or at least by those involved in the process), ex post information asymmetry is also relatively low. Information asymmetry remains high when the information about the process of delivery stays unknown to those who are involved in the activities. This can occur in situations with highly specialized information (expert information) or when it’s not possible to protect the information from opportunistic manipulation by the sender at acceptable cost. Then, the organization is effectually unable to assess the quality of performance, even after it has been delivered (Speklé, 2004).
2.2.4 The control solutions: archetypes
The TCE-MC offers a framework of four archetypes, which are a combination of different control instruments and which offer an effective control solution. Depending on the scores on the dimensions asset specificity, uncertainty, and ex post information asymmetry, the most effective archetype has to be chosen. The four archetypes of TCE-MC are: arm’s length control, action/ result oriented machine control, exploratory control, and boundary control.
An introduction to the archetypes follows. Table 1 provides an overview of the main characteristics of the archetypes.
Table 1: Characteristics of the control archetypes (Speklé, 2004)
performance conforms to market standards. Another characteristic is that there are detailed and reasonably complete contracts in which arbitration is the option to resolve conflicts. Arm’s length control is associated with generic, relatively unspecific activities for which an outside market exists (Speklé, 2004). Machine control is considered as effective in controlling highly programmable activities (low uncertainty) and when there is high asset specificity. The machine control archetype is associated with mature programs and routine activities. Within action oriented machine control, the actions are monitored closely and there is a standardization of behaviour which results in detailed work instructions. There is not much autonomy for the other party and the compliance with the pre-defined rules and standards will be closely monitored and are the basic for the performance evaluation. There is no direct link between performance and rewards. Within result oriented machine control of the result oriented kind, the other party has clearly defined responsibilities. The main focus is on the achievement of clearly predefined performance targets instead of compliance with rules and regulations. Within the result oriented variant, a bonus will be rewarded when the contracted party achieves the targets, to provide contributors incentives to achieve the targets.
Exploratory control is the best fitting archetype when activities are non-programmable and ex post information asymmetry is low. Within this archetype, due to high uncertainty, there is a lack of ex ante work instructions and rules. The contracting party defines general goals and gives the contracted considerable freedom to act upon them. Pre-defined performance targets are absent and emergent targets take over. During the carrying out of the activity, information about what can be achieved becomes available for evaluation purposes and is spread through the organization. Subjective judgements are used to assess performance, also based on long term performance. Rewards will be given through good long term prospects instead of short term bonuses. Boundary control can also be considered as ‘control of the last resort’. Activities are non-programmable, asset specificity is moderate or high and ex post information asymmetry is high. The contracted party is working autonomous and the main focus is on preventing unwanted behaviour, instead of stimulating wanted behaviour. This is done by using prohibitive guidelines or boundary systems, budgets with authorization of (maximum) expenditure, tie in of agents through hostages and external audits. There are no performance targets or work related instructions to coordinate the work. Non-compliance has serious consequences for the contracted party.
As described which archetype is relevant on controlling in the most effective way depends on the activities.
Figure 2 provides an overview of the archetypes and the activities they control best. Figure 2: Archetypes of control and their habitat (Speklé, 2001b)
As the archetypes of TCE-MC and the activities which they control best are described, now I will introduce the types of health care and the characteristics of the activities. Based on this activities hypothesises will be determined on which archetype fits best to the contract of the type of health care.
2.3 Types of Health care
For this thesis, I have chosen to describe three types of health care, which are part of the Health insurance act (ZVW) in the Netherlands. Because of comparability between different types of health care within the ZVW and a practical point of view, there has been chosen to limit this study to three kinds of types of health care. The following types have been chosen: general practitioner care, physical therapy and specialist-medical care. A further description of these types will be given next. Thereafter a hypothesis is drawn about the most effective way of controlling this type of health care by a health insurer. This is done by the health insurer through a contract between him and the health care provider, which can be seen as the MCS. Thus, for each type of health care, the dimensions of TCE-MC have to be defined. In case of uncertainty and ex post information asymmetry the activities of the health care provider are considered and it is first determined if they are programmable or not. This means, if the health insurer has enough information about the performed activities by the health care provider. If not, ex post information can make it possible to gain enough knowledge about the performed activities by the health care provider. In this, the activity is considered as the treatment the health care provider performs. In this study, asset specificity is considered as the opportunity of a health care provider to use the investments he made for other purposes. In other words, can he switch to a contract with another health insurer easily?
2.3.1 General practitioner care
The general practitioner (GP) is seen as the key position in primary care and has an important role in keeping Dutch health care affordable (Vektis, 2013). The GP is also seen as the gate keeper in the Dutch health care system. (NZa, 2012a; Vektis, 2013) This means that patients need a referral from a GP before they can go to a medical specialist. Other tasks of a GP are the assessment and treatment of medical problems. The GP is the first one to turn to when having medical problems and the GP treats most of the problems himself. (Vektis, 2013) A GP can also prescribe medicines and for diagnosing purposes, the GP can ask for a lab test (for example blood or urine). Because of their important role in primary care and, as described, as well in the costs of the whole Dutch health care system (most costs in Dutch ZVW are in second line care, wherefore patients need a referral), I’ve chosen to add GP care to my study. Health insurers contract GP’s using the representation model (NZa, 2012a). This means that the biggest health insurer in a region negotiates with the GP’s within this region about the conditions of the contract. Other health insurers follow the contract of the health insurer who negotiates. Health insurers use their own contracts for their own region and use follow contracts for GP’s outside their region. Any customization can be contracted by modernization and innovation (M&I) modules. 98% of the GP’s have a contract with a health insurer (NZa, 2012a). By looking at the contract with a GP from the position of a health insurer using the three dimensions uncertainty, asset specificity and ex post information asymmetry of Speklé (2001a, 2001b), the following question arises; what is the most effective MCS for a health insurer for a contract between the health insurer and a GP? Therefore the dimensions of TCE-MC have to be known. These will be decided next. The activities of a GP can be considered as non-programmable and therefore uncertainty is high. For a health insurer it is unknown before which activities a GP should perform because the activities a GP has to make can/ must be different for every patient. A health insurer has no knowledge about this activities as there are many possible treatments and beforehand it can’t be decided which treatment has to be given. This makes controlling the activities of a GP difficult, as they are not known before. As mentioned, before the GP is the first who sees a patient, and there is no information available yet about the health situation of the patient. This makes it difficult to plan a treatment and for the health insurer to control the activities. For example when a patient goes to the GP beforehand, it is uncertain whether the patients need a consult for less or more than 20 minutes. Every patient is unique and thus, every treatment. Therefore, I consider GP care as non-programmable. I consider asset specificity in case of a GP as moderate. This while a GP and the health insurer are active in a market and it is possible for the health insurer not to contract a GP and vice versa for a GP it is possible not to accept a contract of a health insurer. The investment a GP has to make, doesn’t make him dependent on a health insurer. Especially in the market of GP’s, the patients decide to which GP they go and there are almost no limitations in the reimbursement of the health insurer for the performed activities of the GP. Only the earlier mentioned customisations in the contract are not paid for by the health insurer in the case, when there is no contract at all. For these reasons, asset specificity is classified as moderate.Ex post information asymmetry between the health insurer and the GP is considered as high. This because the treatment of a GP is something between him and his patient and the health insurer isn’t involved in the treatment. Therefore the length of a treatment or the exact treatment is only known by the GP and the patient. One of the reasons health insurers aren’t involved and can’t be involved in the treatment is the fact that health is a very personal issue and information about someone’s health can be very valuable for companies. It can be used by companies to decline products or to ask higher prices for their products. This for example would be the case, when health insurers would know the health situation of a person and would be declined a voluntary health insurance (VHI) or asks a higher price. Therefore medical confidentiality, in which medical confidentiality is the privacy of a patient, is very important in the Dutch health system and it is written in law (ZVW). It is therefore not allowed for a health insurer to examine the treatment of a GP. There are some exceptions. These are when there is permission from the insured, there is a suspicion of fraud or wrong invoicing, or in case of medical material control (MMC; medisch materiële control in Dutch). In these cases, the treatment can only be examined by a physician, who’s employed at the health insurer, but who isn’t allowed to give any information about the content of the treatment or the health situation of the insured. The physician employed at the health insurer is only allowed to give information about if the treatment was done in a proper way and if it took place, so no medical information can be given (Eno, 2013b; Zorgverzekeraars Nederland, 2012). The knowledge of a GP can be typified as expert knowledge, and according to Speklé (2004) this is one of the characteristics where ex post information symmetry appears easily. This ad to the situation that ex post information asymmetry between a health insurer and a GP is high. From the text above can be concluded that within the contracts between a health insurer and GP’s, the variables which define the control problems can be classified as high uncertainty, moderate asset specificity, and high ex post information asymmetry. The health insurer has in this case very limited information about the activities performed by the GP (high uncertainty) and has also very limited possibilities afterwards to find out which activities were performed (high ex post information asymmetry). Therefore, the most effective control solution is boundary control. This while boundary control is the most effective archetype when it is very hard to measure performance. It focuses on preventing unwanted behaviour and results, and is doing this by proscriptive control. Figure 3 provides an overview of the position of GP care within the dimensions of TCE-MC. According to Speklé (2001a, 2001b) boundary control is the most effective control solution. This leads to the following hypothesis:
Figure 3: Position of GP care within the dimensions of TCE-MC
2.3.2 Physical therapy
Physical therapy is a health care profession primarily concerned with the remediation of impairments and disabilities and the promotion of mobility, functional ability, quality of life and movement potential through examination, evaluation, diagnosis and physical intervention (NZa, 2013a). For example patients can go to a physical therapist after a knee surgery to improve the strength of the muscles and the coordination and balance of the movement system.Physical therapy was the first type of care in which there was free contracting and negotiating possibilities between health insurers and the physical therapists. This was allowed by the Dutch Health Care Authority (NZa) and started in February 2005 as an experiment for two years (CTG, 2005). It stayed like this, since. Also the referral from a GP isn’t obligatory anymore since 2006 (the year of the introduction of the ZVW) (NZa, 2013a).
Because of the free negotiation and contracting possibilities, this type of health care is relevant to add to this study.
There has to be made a distinction between chronically physical therapy (article 2.6 ZVW and Appendix 1 of article 2.6 ZVW) (also referred to as physical therapy on the “Lijst Borst” named after a former minister of health) and non-chronically physical therapy. Chronically physical therapy is paid from the ZVW after the first twenty treatments when a person is 18 years or above (Zorginstituut Nederland, 2013; Article 2.6 ZVW; NZa, 2013a). There are maximum terms for the treatment of chronically physical therapy (Appendix 1 of article 2.6 ZVW). The first twenty treatments can be paid from a voluntary health insurance (VHI) or when the insured doesn’t has this; he has to pay for it himself. Non-chronically physical therapy isn’t covered in the ZVW when the person is 18 years or above. When the person is younger than 18 years, also the non-chronically physical therapy is covered by the ZVW. This is for the first 9 treatments, which can be extended with at maximum another 9 treatments. Chronically physical therapy at persons younger than 18 years is fully covered by the ZVW.
Physical therapists sent in their declaration to a health insurer a 4-digits code, a so-called diagnostic code, which indicates if it’s chronically or non-chronically and a 2-digits code, called the code type of indication (CSI), to indicate which care is delivered and in what stage the care is (Eno, 2014). For example a CSI code of 01 means that it is covered by the ZVW, while it’s after the first 20 physical therapist treatments and a CSI code of 09 means it isn’t included in the ZVW coverage, as it is not chronically. Around 90% of the Dutch physical therapists have a contract with one or more health insurers (NZa, 2013a). Similar to the GPs, the three dimensions in case of physical therapy are considered next. The activities of a physical therapist can be considered as non-programmable and therefore, uncertainty is high. For the health insurer the activities of the physical therapists are unknown before, and as described above, the treatments by a physical therapist are not known before either for the physical therapist. This is similar as in the case at GP’s, and at the start of the treatment, most of the time the problem isn’t known yet. Although there are protocols for physical therapist to use when he made a diagnosis, still every patient can react different on this treatment. So he has to evaluate this treatment and can choose to change the treatment method during the whole treatment. During this process the health insurer doesn’t has any information, just the information it gets from the invoices. For the health insurer it is uncertain which treatment is chosen and it is unknown if the chosen treatment is effective. As well this means that the length or the amount of treatments is unknown before the treatment starts. This makes uncertainty high for the activities of a physical therapist.
So within physical therapy, health insurers should use the same MC structure for a contract as the health insurer uses for GP care. This while uncertainty is high, asset specificity is moderate and ex post information asymmetry is high as well. Figure 4 provides an overview of the position of physical therapist care within the dimensions of TCE-MC. As can be concluded boundary control is the most effective form of control for a health insurer for physical therapy care. This leads to the following hypothesis:
Hypothesis 2: Health insurers contract physical therapists using boundary controls.
Figure 4: Position of Physical therapy care within the dimensions of TCE-MC
2.3.3 Specialist-medical care
DOTs. The insight of the delivered care has mainly been lost and the expected costs for 2012 were unknown because of this; as well it is difficult to transfer a price of a known DBC to a price of a newly introduced DOT (Zorgverzekeraars Nederland, 2013). A specialist-medical care provider registers its activities on a patient in a treatment and has to close the treatment after a certain amount of days (depending for example on the specialism and if there is a surgery involved) by using the registration rules which have been introduced by the NZa. A DOT can be open for a maximum of 365 days. After a treatment has been closed, because of finishing the treatment or because the maximum number of days the DOT is reached, a national grouper translates the activities of the treatment in a DOT which is sent to the health insurer of a patient. Dependent on the contract between the specialist-medical care provider and the health insurer a price is paid for this DOT (NZa, 2013b). The system of specialist-medical care financing is very complex and for the purpose of this study, the brief description above is sufficient, as it isn’t the aim here to provide a full description about specialist-medical care and how it’s financed. For the B-segment, specialist-medical care providers and health insurers can freely negotiate about a contract and the conditions. The insurer is free to choose which kind of payment it provides (for example lump sum, or price * quantity) and which kind of quality it expects and how the specialist-medical care provider has to prove it complies with this quality (Zorgmarktadvies, 2011). The choice of a health insurer not to contract a specialist-medical care provider can limit the choice of the insured, as the insurer can decide not to reimburse or just partly reimburse a treatment of an insured. Therefore most health insurers are still restrained in respect with selective contracting, as it is important to be attractive for possible insured.
Specialist-medical care providers are diverse and especially after 2001 specialized Independent Treatment Centres (ITCs; in Dutch Zelfstandig Behandelcentra ZBC’s) entered the market of specialist-medical care (NZa, 2012b). ITCs mainly focus on not-clinical care with high quantity, which can be planned (so not emergent), and which is less complex, and which makes the treatment possible in day-care. Stand-alone specialisms like orthopaedics, ophthalmology and dermatology are common for ITCs (NZa, 2012b). This means that hospitals are left with the more complex cases (Groot and Krabbe, 2010). Because the difference between the complexities of care delivered between ITCs and hospitals, it is necessary to differentiate between these two types of providers within specialist-medical care. This is also supported by the fact that hospitals offer a much wider range of treatments then the mainly stand-alone specialism at ITCs, especially after the enlargement of the B-segment from 34% to 70% in 2012. That is to say, the variety of offered care in hospitals is much higher and more complex than it is at ITCs, therefore I suppose that the dimensions of TCE-MC are influenced. This makes it necessary to distinguish between these two providers within specialist-medical care. As done at GP care and physical therapists care, I now consider the three dimensions in case of the B-segment of specialist-medical care in which ITCs and hospitals are considered separately. First the dimensions of the ITCs are described and secondly the dimensions hospitals are described.
The dimensions in the case of ITCs:
which are performed. This is partly due to the limited types of DOTs performed at an ITC, because of the stand-alone specialism within an ITC. For example for an orthopaedic ITC there is enough information at a health insurer about the activities that should be performed at an arthroscopy or at an anterior cruciate ligament reconstruction. It can be concluded that at ITCs actions are programmable and that uncertainty is low. Asset specificity is considered as moderate. This because ITCs operate in a market. They actually try to outperform general hospitals in costs and quality and try to get contracts with health insurers in this way. Health insurers can decide to contract or not to contract them. The investments made by ITCs can be used for other patients as there exists a market in which it can be used and where different health insurers try to contract the care with the highest quality against the lowest price.
At ITCs the most effective MC structure is the MC structure which fits best at a situation where uncertainty is low and asset specificity is moderate. Ex post information asymmetry is not considered as the health insurer possesses enough information already about the activities. Figure 5 provides an overview of the position of ITCs within the dimensions of TCE-MC. Arm’s length control is the most effective form of control for a contract between health insurer and ITCs. The ITC is autonomous and there are relative complete contracts in which for example the expected quality is written. This leads to the following hypothesis:
Hypothesis 3a: Health insurers contract ITCs using arm’s length controls.
investments are general, as every hospital needs them, to be able to be a hospital. Therefore asset specificity is considerate as moderate. Ex post information asymmetry between hospitals and health insurers is high. As explained above health insurers don’t know which activities lead to the invoiced DOT and reverse it’s also not possible for a health insurer to deduce the invoiced DOT to the activities which took place. The only ones who know which activities took place are the treating doctor and (mostly) the patient. As well the very complex registration rules and the use of a national grouper make it almost impossible to control the correctness of a DOT for a health insurer. In hospital care, the earlier mentioned medical confidentiality is a problem for controlling and for solving the problem of ex post information asymmetry as well. Hence, ex post information asymmetry between hospitals and health insurers is high. This leads to the conclusion that in hospital care boundary control is the most efficient MC structure for a contract, to control hospital care from a health insurer’s point of view. Uncertainty and ex post information asymmetry are high and asset specificity is moderate. Figure 5 provides an overview of the position of hospital care within the dimensions of TCE-MC. The hypothesis for hospital care can be drawn as follows:
Hypothesis 3b: Health insurers contract hospitals using boundary controls.
Figure 5: Positions of specialist-medical care within the dimensions of TCE-MC
Figure 6: Positions of the types of care introduced within the dimensions of TCE-MC
3. Methodology
This chapter contains the description of the research method used and the introduction of the case that is studied. As well the method of data collecting is proposed here. Firstly, the research method and the way of data collecting are introduced and secondly, the case is described.3.1. Research method
In this study, I use qualitative research techniques. In chapter 2, I draw my hypotheses on which archetype of TCE-MC is most suitable to control the actions of a health care provider within the contract from a health insurer based on existing literature. I’m performing a hypothesis-testing study. This is conducted through a case study. Although hypothesis-testing studies in general are done by quantitative studies (Cooper et.al, 2003) a case study can be used for hypothesis-testing (Flyvberg, 2006). The research method used is a case study. A case study can be categorized as a research strategy which focuses on understanding the dynamics present within single settings (Eisenhardt, 1989). Furthermore, “case studies offer us the possibility of understanding the nature of management accounting in practice; both in terms of the techniques, procedures, systems, etc. which are used and the way in which they are used” (Scapens, 1990). Although Scapens (1990) is focusing on management accounting, it is also applicable on MCS. This is confirmed by the definition of Yin (2003) as he describes a case study as “an empirical enquiry that investigates a contemporary phenomenon within its real-life context”. As I study the contracting of health care providers by health insurers, there is a single setting. As well, I explore how contracts are designed, which fits in the description given by Scapens (1990) and Yin (2003). This is the nature of MCS in practice. Therefore, it can be said that a case study is appropriate in this study. This case study can be classified as an embedded single case study (Yin, 2003; Grünbaum, 2007). This, while the unit of analysis is the contract between the health care insurer and the health care provider and in this case I examine four types of contracts. The case in this study is the health insurer Eno. This will be introduced in the next paragraph. Eno has been chosen because I had full access to all the necessary documents and because I’m employed there, I had access to all the necessary people who I had to interview.
This case study is only a small sample from which it is difficult to make a statistical generalization. Another problem of this case is that the costs of specialist-medical care in 2012 (the year of the introduction of DOT) are not known yet by health insurers, which makes a quantitative analysis impossible still.
In the case study, firstly the contracts between Eno and a health care provider in the certain field of interest (for example, general practitioner) for the year 2012 are studied and are checked if the MC structure complies with the expected structure of TCE-MC.
Secondly, semi-structured personal interviews with experts in the field of contracting health care providers are conducted. Semi-structured interviews are used, so follow-up questions can be asked if the researcher needs more information. The capacities as experts for a certain field of activity is the most important when performing an expert interview (Flick, 2009). As all the interviewees are working at a health insurer and are related with the contracting or control of health care providers, they are considered as experts, because of their experience in, and their knowledge about health care and control or contracting of health care providers.
Before the interviews, no information was given to the interviewees about the aim of the study and the drawn hypothesis. Just a neutral introduction was given in order not to influence the opinion of the interviewees and not to influence the answers of the interviewees. Table 2 provides a list of the interviewed persons and the length of the interview.
Table 2: list of interviews The interviews took around 1 hour each, were done in Dutch and were tape recorded. This is done, so afterwards could be checked, in case of doubt, if conclusions made during the interview were made right. For the interviews, an interview protocol is used. During the interviews first the interviewees were put at ease, so he/ she felt free to give the information needed from the interview. Subsequently, questions based on the dimensions of TCE-MC were asked and the focus lied on these dimensions, as it is the aim to find out if the hypotheses were drawn correct. An example of one of these questions is of the performance and the quality of the delivered care is easily measurable. After the questions about the dimensions of TCE-MC were answered, the focus lied on how the current contract is constructed, how control takes place, how the most favorable contract according the interviewee would be constructed and what minimal should be included in a contract to effectively monitor the performance. This is included to get an answer of how the four dimensions of MCS are integrated in the contract between health insurer and health care provider. The four dimensions are: (1) the allocation of decision rights, (2) use of standards, rules, and regulations, (3) performance evaluation, and (4) rewards and incentives. There had to be given an answer for all four dimensions, before the interview could be finished. An example is the question about how health care providers should be rewarded, within a specific type of care. As interviews took place with health care purchasers and non-health care purchasers, two interview protocols were used. This while the health care purchaser has knowledge about one specific type of care, as they are purchasing this type of care. The interview protocol is based on this. The non-health care purchasers, who are more general experts about contracting and control, needed another interview protocol. In this protocol questions are more general and don’t go too much in depth about one type of care. For example, the interviews with the CFO and the manager zorgcontrol where more focused on the control and the financial impact of the health care purchasing and how they define the activities (within the model of Speklé) of the health care provider. The interview protocols can be found in Appendix I & II. The results of the interviews are presented in the next chapter.
3.2. The case introduction
Eno is from origin a regional health insurance company within the region of Deventer (Salland) and is founded in 1860 in Deventer. Eno has therefore the main group of insured in the region of Deventer. At the end of 2012, 151 FTE where working at Eno and over 2012 a net profit of 5,6 million Euro was presented (Eno, 2013a). In 2006, Eno was transformed from a sick funds in to a health insurance company, because of the newly introduced ZVW. Eno is the smallest independent health insurance company of the Netherlands with around 150.000 (estimated) full time insured in 2014 (in a market of around 17 million people). Eno has three different insurance labels. These are Salland (approx. 95.000 insured) and focusing on the traditional region of Deventer, HollandZorg (approx. 45.000 insured) which is the leader in the market for labor migrants from mainly the former east of Europe like Poland and Czech republic and Energiek (approx. 10.000 insured), which focus is on women. Energiek is a relative young label which was introduced in the market in 2010. As Eno is the smallest health insurer in The Netherlands, it only purchases the care and contracts in its main region Deventer, because it is too expensive and complicated to contract outside the region. For the contracting outside the main region, Eno participates in Multizorg VRZ, which is a purchaser
Position Length: (in minutes)
combination of four health insurers (ASR, ONVZ, Eno, and Zorg & Zekerheid). This makes influencing the contracts outside the region more difficult as there should be consensus between the four health insurers about the content of the contracts.
4. Results
This section describes the major findings resulting from the interviews and from the studied contracts. This leads to findings about which archetype is used by Eno when contracting GPs, physical therapists, ITCs and hospitals. The results will be confronted then with the hypothesis. Consequently, I will describe whether Eno is using the most effective MCS (archetype) according TCE-MC. The results will be presented for every kind of health care separately, starting with GP care, following physical therapist care, and last specialist-medical care.4.1 GP care
According to the interviewees, the interpretation of the dimensions of GP care are considered in the same way, as proposed earlier in paragraph 2.3.1. This means that uncertainty is high, asset specificity is moderate and ex post information asymmetry is high. The interviewees stated that it is unknown for the health insurer which activities a GP will perform when a patient comes with a problem to the GP. According the interviewees this is for example due to the subjectivity of a GP and the different interests between GPs. A GP care purchased explains as follows:“Although the GP has protocols about how to diagnose a patient and the treatment, this depends on the GP. Some have specialized themselves in some diseases or parts of the human body and will treat the patient in another way as his colleague who isn’t specialized. This is unknown for me, as is the exact treatment of the GP.” (GP care purchaser)
As a general introduction is given about the contracts, now the four dimensions of MCS are described based on the contracts and interviews.
4.1.1. Structure in GP care
Within the contract, the only thing included about the structure of the contact is that the GP has to perform an activity which is allowed and which is general accepted as GP care. These activities are listed in the contract, and he can only invoice these. Therefore, based on the contract, the structure of GP care can be classified as an autonomous structure. The GP decides which activities (within a given range) he performs without influence of the health insurer. The conclusion from the contract was confirmed in the interviews, as the following quotes from the interviews show.
“I can’t decide which treatment a GP has to give. He is the expert and I don’t want to pretend I have sufficient knowledge to decide which treatment is the best.” (Purchaser GP care)
“A health insurer shouldn’t and can’t play the role of a health care provider.” (CFO)
These quotes show that according to Eno, a GP should operate autonomous of the health care insurer. The autonomous position of a GP in relation to the health insurer is consistent with the archetype boundary control, and so consistent with the hypothesis about GP care.
4.1.2. Standardization in GP care
Within the contract, there are articles about the delivering of care and how it is standardized. Examples are an article about the possible care that can be delivered and the prices that can be invoiced. As well, an article state that the GP should deliver care with taking in account the most recent guidelines, standards and protocols of his type of healthcare. Furthermore, an article is found about the consideration of the costs and the benefits of the care and that the delivered care shouldn’t be inefficient. These articles are general constructed and are based on what is acceptable in the “market”. There is not much emphasis on behavior to avoid in the contracts, other than the mentioning of fraud and that the care shouldn’t be unnecessary expensive. The contract is structured most consistent with the archetype arm’s length control as it is includes general articles based on what is standard in the market. From the interviews no new information was gained. Therefore, I consider the standardization in GP care not consistent with the archetype boundary control. Standardization is most consistent with the archetype arm’s length control.
4.1.3. Performance evaluation in GP care
information is sent to the GPs and is discussed during the contract negotiations or during visits of the GP care purchaser. In this meeting the GP can be asked to explain the differences of the benchmark. Generally there are no consequences for the GP when performing different compared to the average GP. The GP care purchaser felt that benchmarking is a weak performance evaluation system, but the only possible one. He felt that it is weak, because there can be plenty of reasons a GP performs different than the average. According to four of the seven interviewees, it is not observable if the GP is using the most recent guidelines, standards and protocols and if he doesn’t makes care unnecessary expensive. So therefore no monitoring takes place on this. The described evaluation method used at Eno is not consistent with the archetype boundary control. Based on the findings of the interview it is most consistent with the archetype arm’s length control. As benchmarking is a way to evaluate performance relative to the “market”. So the hypothesis about boundary control is not supported.
4.1.4. Rewards and incentives in GP care
In the contract with the GPs is included what the GP can invoice. All the possible activities and the prices are either included in the contract or in the “tariefbeschikking huisartsenzorg” of the NZa. The structure is based on a subscription fee per registered client, depending on the characteristics of the insured (age and neighborhood), which can be invoiced quarterly and the activities the GP performs. For example, a consult or a consult more than 20 minutes. In the contract an article is included that in case the GP doesn’t fulfill the contract, it can be terminated. So, in the contract there is an incentive for the GP to invoice the more expensive activities, as there is not enough information and possibilities to control if the invoiced care is really delivered. This is consistent with the principles of boundary control. This incentive was confirmed in the interviews.
“I don’t know if the GP was visiting for more than 20 minutes, or for less than 20 minutes. I was not in the room with the patient and the GP.” (GP care purchaser)
contract, the structure is not based on boundary control, as the termination has no consequences. Because of the empirical findings, the hypothesis about boundary control has to be rejected as arm’s length control is the main archetype.
4.1.5 GP care summarized
The hypothesis that GP care is contracted using boundary control has to be rejected. The empirical findings show that the health insurer is using a control structure which is using different control instruments from different archetypes. This means that the health insurer doesn’t use the most effective MCS, according to the theory of TCE-MC. Within TCE-MC the most effective MCS should be consistent with the dimensions of MCS and should be structured consistent. Table 3 provides a summary of what the empirical findings are about the contracting between Eno and GPs and how this compares with the archetypes of TCE-MC.Table 3: summary of empirical findings at GP care compared with the archetypes of TCE-MC
4.2 Physical therapy
During the interviews, first the interpretation of the dimensions of physical therapy was discussed. According to the interviewees the dimensions of physical therapy are interpreted as high for uncertainty, moderate for asset specificity and high for ex post information asymmetry. This is consistent with the assumptions made in paragraph 2.3.2. The most important reasons why uncertainty is considered high are the lack of insight in the combination of used CSI and diagnostic code and the given treatment and the subjectivity of physical therapists. The interviewees argued that the health insurer doesn’t possess enough information about the treatment before it starts.“When the patient starts a treatment, I don’t know which problems he has and how the treatment and the plan will be constructed by the physical therapist. This is depending on his professional view and on his experience with different kinds of physical therapy. For example shockwave therapy or hydrotherapy.” (Physical therapy purchaser)
Since 2014, Eno has made it obligatory for patients to have a referral from, for example a specialist or a GP, if the care has to be paid out of the ZVW. This is done, to obtain more control and insight about the care which is paid out of the ZVW. Although the referral can’t be checked without physical therapy added to the MMC. Still then there should be a reason to individual files. As mentioned earlier, control afterwards is considered very difficult. Therefore ex post information asymmetry is high.
Arm's Length Control Machine Control (Action Oriented)
Machine Control (Result
Oriented) Exploratory Control Boundary Control Empirical findings
Most consistent with
Structure
Rel ati ve autonomy; i nvol vement hi gher l evel manage ment l imited long as performance is s a tis factory
Well -define d ta s ks ; s tri ct hi erarchy; li mited room for dis creti onary beha vior
Dece ntrali zed with cl early defined areas of res pons i bi lity and accounta bil ity
Rel ati vel y fla t hi erarchy; fl uid and permeable ma tri x-li ke project s tructures ; vague res pons ibi li ti es
Rel ati ve autonomy within defined boundaries Autonomy wi thi n defi ned boundaries Boundary control Standardi zati on Market-rel ated outcome requirements ; external performance benchmarks
Sta ndardizati on of beha vior; deta il ed rules , norms and i ns tructi ons
Predefi ned performa nce targets of admini s trati ve origi ns
No ex ante s ta ndars and targets ; 'do your bes t'; eme rgi ng s ta ndards
Pros cripti ve codes of conduct; boundary s ys tems ; emphas is on behavior to be avoided
Guidel ines , regulati ons and protocol s which are bas ed on bes t practi ce in the "market". Arm's Length Control Monitori ng a nd performance eval uati on Pe rformance as s es s ment relati ve to 'the market' Moni toring and s upervi s i on to ens ure compl iance to norms and s ta ndards
Monitori ng focus ed on ta rget a chie vement; performa nce as s es ment rel ati ve to
Bas ed on emergi ng s ta ndards ; s ubjecti vely as s e s s ed
contri buti ons to l ong
Focus ed on compl iance; obs ervance of interdicti ons ; external
Benchmark between GPs Arm's Length Control Reward & Incenti ve s tructure Pe rformance depe ndent bonus es No direct l ink between performance and rewards Performance dependent bonus es Career pros pect depende nt on long te rm pas t perfomance; peer pres s ure
Emphas i s on thre at of puni s hment of rule-bre aking behavior; tie-in through 'hos ta ges ';
Pay per a cti vi ty + s ubs cri pti on fee. So incenti ve to "produce"
Contracts between Eno and the physical therapists are negotiated within the region of Eno by the health care purchaser of Eno. For outside the main region Eno uses Multizorg VRZ for contracting the physical therapists. Eno distinguishes between physical therapist practices. Depending on the score of a questionnaire, which is based on the categories quality, service, treatments, reporting and innovation, the physical therapist gets a rate. There are four categories (A, B, C and D) in which the highest category has the highest price for invoicing the activities. The physical therapist has to fill in this questionnaire and the answers can be tested by Eno. At this moment, the interviewees told that it is not done yet, but it is considered as a safeguard. Eno trusts the physical therapist and the way how the questionnaire is filled in. The contract between Eno and the physical therapists consists of 2 parts. The first part is the general terms and conditions of purchase and the second part is the contract 2014 between Eno and the practice of the physical therapist. The general terms and conditions are the same as at the GPs. The contract 2014 is a prescriptive contract in which is described what Eno expects from the physical therapist. In the contract 2014 is also the questionnaire added as an appendix. Next to this, the price list is added with the four categories of physical therapists. As done at GP care, subsequently the four dimensions of MCS are described based on the contracts and interviews.
4.2.1 Structure in physical therapy
In the contract is defined which treatments the physical therapist can invoice and that he only can invoice physical therapy as how physical therapy is described in law. There is not more included in the contract. From the contract analysis, it can be concluded that the structure of physical therapists at Eno is an autonomous one. Within the interviews this was confirmed, as the same arguments as in case of GP care were used. In general, the physical therapist is the expert and decides which treatment to give. Eno does not influence this. The autonomous structure of a physical therapist in relation to the health insurer is consistent with the archetype boundary control, and so, consistent with the hypothesis about physical therapy.
4.2.2 Standardization in physical therapy
The care of the physical therapist is standardized in the contract by articles about the quality of the care, the activities which can be invoiced and how the physical therapist should register the care. An example of this is article 6.8 of the contract. This article describes that the physical therapist has an electronically physical therapist reporting system. Another example of the standardization is article 6.7. This article states that it isn’t allowed for a physical therapist to treat at schools. The focus of the contract lies therefore on the standardization of behavior where Eno ads rules, norms and standards to the contract. This is confirmed by the earlier mentioned questionnaire, in which norms are defined for good quality. Eno clearly makes a choice about, what they feel, the standard should be. This focus on the standardization of behavior and setting norms and rules is consistent with machine control (action oriented).
“In the questionnaire of physical therapy we ask them general information about their practice and indicators about, what we feel, influences the quality of care they can deliver to our insured. These indicators are standards, what we feel a good practice should meet up with. For example, the gym or specializations within the practice.” (Head health intelligence)
Although the three interviewees that confirmed it, didn’t think that Eno was putting limitations for the physical therapist by this. They think it is a way to get some control as they don’t see any other option for doing so. The conclusion is not consistent with the expected archetype of boundary control and is consistent with machine control (action oriented), which focus on detailed rules, norms and instructions.
4.2.3 Performance evaluation in physical therapy
In the contract there is written that Eno is benchmarking the physical therapists. In the contract is included that when the average number of treatments lies lower or higher than 20% of the Eno average, Eno will look at the effectiveness of the delivered care. The physical therapist has to give an explanation for the deviation. When the physical therapist is opting for option D in the questionnaire, Eno asks for an external audit report as well. As this is not done for everyone it is not considered as a general performance evaluation. So from the contract, only benchmarking was found as a performance evaluation method. In the interviews, it became clear that it isn’t possible for the years 2012 and 2013 to benchmark the physical therapists, as Eno had problems with their administration of physical therapy and therefore the available data is not reliable. From 2014 benchmarking has started again, based on reliable data. The interviewees confirmed that the only performance evaluation used is benchmarking. As mentioned earlier, the standards and norms asked in the questionnaire aren’t evaluated. This leaves the benchmark as the only performance evaluation method of Eno. Benchmarking is not a consistent control instrument for boundary control, but for arm’s length control. The main archetype is therefore arm’s length control.
4.2.4 Rewards and incentives in physical therapy
In the contract, the price per activity is included. This is dependent on the category the physical therapist is categorized in. The category depends on what is filled in at the questionnaire. In the questionnaire no performance related measure is included. In the contract is also included which consequences irregularly behavior can have (article 12.2.e). This is; reclamation of the already paid activities, putting the physical therapist in a lower category, or as a last option, the termination of the contract. As it is at GP care, termination of the contract has no real consequences. Eno still has to pay for the activities performed. The price in case of no contract is equal to the price of an A- contract.
