Quasi-backsourcing in the Public Sector – The Challenge of Withdrawing from an Intertwined and Long-standing Relationship

ABSTRACT The aim of this study is to map the process involved in terminating a long-standing business relationship in a public context. It employed an exploratory study to interview 35 politicians, senior- and middle managers, ambulance- and dispatch centre staff and representatives of a regional alliance of patients. Growing criticism from a county council towards a contractor ended in a decision to bring services back in-house. However, terminating the contract failed and instead a state of quasi-backsourcing emerged. Only parts of the services were taken back in-house, with few reported benefits. The study suggests that caution is needed in relation to backsourcing services from a closely intertwined contractor.


Introduction
This article concerns the backsourcing of public services from a nationwide monopolist to a county council. The business relationship between principal and contractor developed over decades and became organisationally intertwined in the delivery of services. The aim was to map the process of backsourcing a long-standing contract from a well-established, semi-public contractor into a public organisation. The study provides advice for organisations that are planning to backsource public services and pinpoints obstacles that it is necessary to overcome. It also sends a message to contractors about the risk associated with being in a position of monopoly. The article is of interest for the field of managing changes to intertwined and long-standing relationships in public services. Furthermore, it illustrates the difficulties in calculating costs and measuring benefits prior to taking a service back in-house.
Sourcing includes processes from outsourcing services, i.e. contracting, to backsourcing, i.e. taking services back in house. Jansson et al. (2020) emphasise that outsourcing and backsourcing are associated and not independent phenomena in a dynamic sourcing strategy. Nevertheless, increasing numbers of public authorities have focused on the outsourcing of public services, with the aim of exploiting the rewards of market competition (Savas, 1987). Contracting-out has established a market of economic activities and become an important part of public service supply. (Bel & Gradus, 2017). According to Domberger & Jensen (1997), governments have saved some 20% on costs of services by putting them through a competitive tendering process. However, other studies have found mixed effects of contracting out services. Hodge (2000) found that expecting contracting to be a remedy for poor delivery of public services would be a mistake, especially in social services and health care. A systematic review of international studies on the outsourcing of public services during the period 2000-2014 shows that there is evidence of cost savings from contracting out in technical areas, but not in social services (Helby Petersen et al., 2018). In another review of literature concerning the public sector, the authors conclude their findings by emphasising that, where contracting entails a high degree of difficulty, outsourcing of services requires considerable attention to detail, which risks being neglected (Andersson et al., 2019).
A review of the literature on backsourcing in all sectors, including IT, business, human relations and public services, shows that it is often motivated by dissatisfaction with service performance, inadequate service quality, corruption, distrust and communication problems (Petalidis, 2018;Thakur-Wernz, 2018). Of these, unsatisfactory service and increased costs are deemed to be the most common reason to backsource (Damanpour et al., 2019;Whitten & Liedner., 2006). The advantages of outsourcing may have been overestimated and an overly positive attitude can make the calculation unrealistic prior to signing the contract (Young & Macinati, 2012). On the other hand, backsourcing can be expensive, with both parties adversely affected. One such example is the backsourcing of JP Morgan Chase & Co's and the IBM Corporation's IT services contract in 2004. Lack of preparedness caused projects to slow down and day-to-day deliveries to fail, resulting in a loss of income (Veltri et al., 2008;Whitten & Liedner., 2006).
There are a handful of studies explaining the reasons for backsourcing public services. In a review of the backsourcing of public fresh water supply, McDonald (2018) reveals that ideological reasons are a driving force behind remunicipalisation. He cites five main motives, the first being autocratic state capitalism, i.e. an ambition to reclaim state control for socio-political reasons. The next was to take the service back into public control as an effect of market managerialism. A third motive was to improve service and reliability and the fourth was an effect of anti-capitalist influences through trying to expand democratic control and equity. The fifth and final reason McDonald (2018) identified was to attain autonomy by applying local solutions and improving local services.
In another example of backsourcing public services, Wang et al. (2018) found that an expected market effect was as a driving force in the process of outsourcing public transport. Idealistic ideas such as assumptions of a free market, independence in business relations and potential financial and qualitative gains supported the outsourcing decision. However, the free market effect failed, with the outsourced services not performing as expected due to poor profitability, operators terminating their contracts, staff income remaining low, and the service eventually being backsourced. Young and Macinati (2012) studied cases of backsourcing in public health from Italy and Australia, where healthcare providers outsourced non-clinical services such as car parking, laundry, cleaning and catering. A small number of clinical services, such as medical imaging, pathology and radiology were also outsourced. However, after a few years, almost 15% of the Italian and 30% of the Australian healthcare providers had reverted to internal provision. One reason was incomplete contracts and information asymmetry between the parties, which fostered mistrust. As a consequence, monitoring was intensified, adding transaction costs. Young and Macinati (2012) conclude that public sector organisations' lack of expertise, experience and resources lead them to signing inappropriate contracts and expose them to opportunistic vendors. According to Madsen (2017), the trajectory of outsourcing is not always based on rational calculations, instead becoming more of a habit and sometimes even associated with New Public Management hype, which has been in vogue since the late-1980s.
Backsourcing is not a universal concept of taking services back from a contractor. Other concepts similar to backsourcing are bringing services back in-house, bringing services back, insourcing, remunicipalisation, relocating, reversed outsourcing, reclaiming public services, back into public hands and privatisation reversal (Wang et al., 2018). The term backsourcing can be traced back to Hirschheim and Lacity (2000), who realised that reclaiming services back in-house would be an emerging trend (Veltri et al., 2008). Backsourcing can be considered a countermovement based on diminished satisfaction in the outcome of outsourcing. In this study, the concept of backsourcing is based on Hirschheim and Lacity (2000) definition, i.e. a situation where a company takes an activity back inhouse.
The most common focus of studies into backsourcing has been the private IT sector (Damanpour et al., 2019). In contrast to private companies, public organisations can share semi-public contractors owned by clusters of public shareholders, for example, inter-municipal corporations (IMC) (Bel & Gradus, 2017). To the best of our knowledge, studies of backsourcing from IMC providers are virtually non-existent. This case study focuses on an attempt to break away from a long-term integration between a public principal, a Swedish County Council (CC), and SOS Alarm, an IMC-provider of public services in a position of superiority (Bel & Warner, 2015). In this paper, we focus on a case of backsourcing from a long-standing inter-regional collaboration between the CC and SOS Alarm. SOS Alarm is a company co-owned by the state and Sweden's 20 regions and 290 municipalities, which has provided the national emergency 911 service as well as dispatch services since the 1950s. Consequently, the backsourcing in the case fundamentally alters the division of responsibilities and institutional arrangements of existing production. Backsourcing results in the dismantling and reconstruction of old institutions and the emergence of new institutional arrangements. The heterogeneity during the processes suggests the choice of theories of institutional change (Damanpour et al., 2019;Madsen, 2017).

Theoretical framework
According to Hinings and Malhotra (2008), periods such as collaborations during long-standing contracts are often stable and predictable, but might eventually contribute to stagnation and avoidance of promising alternatives. Van Nieuwaal (2011) calls the stage a situation of less room for manoeuvre, i.e. development is hindered by repetitive routines and outdated habits. Mahoney (2000) suggests that such monotony triggers dissatisfaction, which induces a pressure to make a change. The subsequent phase, deinstitutionalisation, paves the way for a new era, and constitutes a phase in when the status of the dominant institution is heavily criticised and disintegrated (Carlström, 2012).
The aforementioned bankruptcy of Lehman Brothers in September 2008 and its worldwide financial and economic aftermath was a spark for international criticism of the New Public Management (NPM) trend, with outsourcing behaviour dependent upon shared standards and broadly accepted. (Funck & Karlsson, 2020;Massey, 2019). Dissatisfaction was fomented and the habit of contracting-out was challenged (Appelbaum & Wohl, 2013). Hinings and Malhotra (2008) predict a stage of pre-institutionalisation in which alternatives to something heavily criticised are developed and appear more promising, in this case, taking services back in-house (Jacobs, 2005). New alternatives are discussed and questioned, with a range of possibilities proposed and new measures outlined (Carlström, 2012;Garud & Karnøe, 2001). Hinings and Malhotra (2008) describe the development of a new state of consensus. Few opponents remain during institutionalisation, and the development of new behaviour, such as backsourcing, cannot be stopped. A successful change process proceeds from diffusion to a fully established institution ( Figure 1).

Methodology
The phenomenon studied is the process of backsourcing dispatch of ambulances in a Swedish county council (CC). In recent decades, the Swedish dispatch centre for ambulance services has been operated by an incorporated company, SOS Alarm, owned by the state, CCs and municipalities. The company provides a two-fold public service, the emergency 911 call service and dispatch services. The former, 911, is a commission from the Swedish state and the latter is a service designed by and provided for the CCs who operate the ambulances, in contrast to the 911 service which is financed by the government. The ambulance dispatch service is purchased by Swedish CCs. They have been loyal customers of the service since it was established in the 1970s. The relationship between SOS Alarm and the Swedish CCs is of such a long duration due to the fact that the ambulance dispatch service is closely interconnected to 911 and the CCs together with the state, and also that the municipalities are shareholders in the SOS Alarm company, even if contracts are terminated (SOS Alarm, 2008Alarm, , 2019.
The precursor to SOS Alarm was set up in 1956, with Sweden as the first nation in the world to introduce a national emergency telephone number. A national emergency dispatch centre was established in the Swedish counties in 1971, with geographic compliance between the dispatch centre and the catchment area of the CCs making it easy to use the SOS Alarm to dispatch ambulances once they received alerts from the public. SOS Alarm was in a monopoly position for decades, and continued to dispatch ambulances nationwide (SOS Alarm, 2008Alarm, , 2019. The emergency operators at SOS Alarm are professional telecommunicators with the task of receiving information and providing callers with assistance in the event of emergencies. The operators constitute the first line of emergency or crisis response, making them an important part of the overall preparedness organisation. The individual operator decides whether the situation merits an immediate response or not, and hence which emergency resources to dispatch. The process requires a high degree of continuity from 1) the 911 call to 2) prioritising and 3) directing resources. During this process an injured or ill patient, or the next of kin, has to be supported by phone, sometimes given instructions to resuscitate, preserve life or prevent escalation of the emergency. In Sweden, where the data for this study has been collected, that continuity (1-3) has hitherto been sustained by operators at the national 911 dispatch centre, SOS Alarm (Normark, 2002).
Though the ambulances were used from the outset as fast transportation to the hospital (load and go), this was subsequently criticised. The need to stabilise the patient on site before departure (stay and play) required staff skilled in assessment and treatment of traumas and the critically ill (Örtenwall, 1999). New legislation in 2000 led to ambulances in Sweden being staffed by Registered Nurses (RNs) capable of handling medical emergencies (Socialstyrelsen, 2017). Since then, as well as ambulances, assessment vehicles, single responders and emergency vehicles have also been staffed with pre-hospital trained RNs and physicians (Carlström & Fredén, 2017). The development of the emergency medical response slowly changed the requirement for dispatch centre operators who could rapidly operate ambulances based on generalised information concerning the degree of urgency to operators who were able to make advanced assessments, prioritise and choose between different medical resources to dispatch. CCs started to require registered nurses (RN) as the minimum degree of medical training for ambulance operators. However, SOS Alarm did not agree with the increased demand of having RNs as operators, arguing publicly that there was insufficient scientific evidence that assessment quality was enhanced when using RNs as operators at the emergency dispatch centre (Kjellin, 2015).
Growing criticism from Swedish CCs towards SOS Alarm gave rise to the notion that the dispatch services should be taken back in-house. The first CC in Sweden to backsource the dispatch service was in 2015, with an internal report by stakeholders claiming that the backsourcing was successful (Region Västmanland, 2020). This was followed by three more of the twenty Swedish CCs over the next three years. After a long process of internal debate, the second most populous CC in Sweden, from where the data for this study was collected, decided to start a process of backsourcing ambulance dispatch services in spring 2018.

Data collection and analysis
The study was based on interviews collected during 2018 and 2019 from politicians, senior-and middle managers, ambulance-and dispatch centre staff and representatives of the regional alliance of patients. Questions were asked about reasons, expectations and effects of backsourcing. Questions were designed on the basis of the aim, i.e. surveying the backsourcing process after decades of using a single contractor. We asked questions such as: "Tell me about the reasons for backsourcing the services"; "Tell me about the backsourcing process"; "Was the outcome of the backsourcing as you expected".
We interviewed 35 individuals, 17 woman and 18 men. A total of 20 interviews were conducted, 9 of them in groups of 2 to 4 participants ( Table 1). The interviews lasted for 30 to 80 minutes.
Participants were selected based on the expectation of acquiring different points of views in order to provide a picture of the reasons, process and effects. However, SOS Alarm was reluctant to participate in the study, and, after several reminders and requests for interviews with staff in various positions, they decided to put one person up for interview. However, a total of five individuals from SOS Alarm were included, with four of them approached individually and asked to participate. Three of them were former SOS Alarm staff, one a stakeholder, one a senior manager and one an operator. In order to ensure richness of data, the participants were chosen from different parts of the CC, for example, the ambulance RNs interviewed were from the CC's four main geographic areas. Only participants with three or more years of experience took part. With one exception, the managers and politicians had more than ten years of experience.
The interviews were recorded on tape. Repeated replaying identified certain phenomena, constituting a pattern (Lo Lacono et al., 2016) of four themes: dissatisfaction with the contractor; meta-level obstacles; failure of termination; and potentially better services (Kvale & Brinkmann, 2009 The results were interpreted using the theoretical models. The backsourcing process was compared to the various phases suggested in the theoretical framework. The respondents' statements played an important role in this part of the interpretation (Deakin & Wakefield, 2013). Finally, the data was reviewed again to ensure consistency with the overall picture (Corbin & Strauss, 2008).

Dissatisfaction with the contractor
During the last century, SOS Alarm and the CC have been in conflict over the necessity of RNs as operators in the dispatch centre. The lack of interest shown by SOS Alarm in using RNs as operators was a well-known issue among CC participants. The CC used an updated contract to compel SOS Alarm to employ RNs as emergency operators. However, due to a shortage, SOS Alarm was not able to recruit enough RNs. The CC even set up a special control unit to register the attendance of RNs at the dispatch centre. SOS Alarm did, however, continue to use staff as operators with an inferior educational background than RNs. Over the years the CC imposed several fines on SOS Alarm for breaching the contract. According to the CC participants, the lack of competence resulted in poor assessments and unnecessary dispatches.
According to the CC managers, poor routines and lack of competence often led to the patients waiting for up to four hours before a pick-up. SOS Alarm simplified the routines by dispatching the closest ambulance rather than the most applicable healthcare resource such as assessment vehicles, single responders or emergency vehicles staffed with physicians. According to the CC managers, the effect was an overuse of ambulances and overcrowded A&E Departments. They felt that an in-house dispatch centre would save health care resources and distribute patients more effectively than SOS Alarm was able to. SOS Alarm has been focused on meeting the time objective rather than grasping the whole picture. They still don't have staff who can distribute the patients to the right type of service.
(Senior manager, CC) Another critique was the accusation that SOS Alarm over-triaged patients. CC participants felt that SOS Alarm overused the highest priority, level 1 of 4 possible levels. The ambulance staff considered that there were too many occasions when they drove at high-speed but arrived at a patient that did not need urgent care. Another example of over-triaging was an overuse of ambulance helicopters. The staff were frustrated because they were repeatedly sent on missions that were recalled because of over-triaged assessments.

Even though the ambulance is ours, SOS Alarm controls the steering wheel.
(CC ambulance officer) The CC respondents felt that the operators at SOS Alarm did not take into account the consequences of overusing ambulances in terms of costs and staff overtime. The high pace and short breaks between call-outs made the ambulance staff tired and the managers had to deal with their dissatisfaction. However, SOS Alarm was autonomous and not a part of the CC. According to CC participants, they were reluctant to address internal CC problems such as this.
It is easy to understand, it is less problematic to dispatch resources you don't own than those you possess within your organisation.
(Senior manager, CC) Some of the CC managers felt that a situation in which an organisation (SOS Alarm) is in the position of controlling another organisation's resources (the CC's) will end in conflicts and poor cost control. According to several managers, the contract should be strictly regulated and any deviation from the agreements should result in consequences for the contractor. If no effective corrections were implemented, the equitable and correct distribution of resources would be problematic.
The respondents emphasised overuse of ambulances as the first in a chain of overuse of health care resources, from A&E Department to unnecessary hospitalisation. More skilled operators would refer to other types of services, such as primary care centres, health advisory dispatch centres and social service centres. In the responders' opinion, the dispatchers of a national organisation such as SOS Alarm had neither the incentives nor capabilities to stay up-to-date regarding all regional services and to make their actions specific to local options.

As the organisation is structured, SOS Alarm cannot be updated on local and regional options. Geographically, our calls can end up in any centre because of the distribution of call technology. How could an operator in northern Sweden possibly know anything about the healthcare facilities in a small town in southern Sweden?
(SOS Alarm operator)

The failure of termination
The decision to terminate the contract with SOS Alarm and move the dispatch centre in-house seemed to be based on CC managers' critique of SOS Alarm's services. However, a reluctance to backsource was apparent in several politicians from the CC.
We thought everybody would applaud the idea of taking services back in-house and instituting our own dispatch centre. All of us, from decision-makers to operators, were aware of all the breaches of the contract registered over the years in relation to SOS Alarm. The internal resistance to backsourcing services was therefore surprising.

(CC manager)
Even though middle and senior CC managers were critical of SOS Alarm, few of the politicians agreed. The fact that Swedish CCs owned 25% of SOS Alarm's shares made the politicians doubtful about backsourcing. Moreover, some of the politicians were involved in SOS Alarm and one was a former member of the board. Several of them were concerned about emergency preparedness at national level. Additionally, SOS Alarm was dependent on the CC co-financing national emergency preparedness, and the CC was dependent on the unique technology developed by the contractors. The establishment of autonomous dispatch centres in the CCs was therefore considered a risk, especially if a major disaster or an international conflict involving Sweden was to occur. SOS Alarm was set up in order to ensure preparedness from a national perspective, but the CCs were organising their operations from a regional and local perspective. Local technical solutions could consequently be detrimental to the national redundancy of emergency preparedness systems.

If a situation of mass calls and communication overload arises, there will be no possible way to support each other in different parts of the country. SOS Alarm has the capacity to connect Luleå to support the Stockholm dispatch centre if it is overburdened because of a terrorist attack.
(Operator SOS Alarm) The reluctance of CC politicians affected the backsourcing decision. A compromise was made with the effect that the expected termination failed. SOS Alarm operated a substantial proportion of the dispatch services following the backsourcing process, and the chain of operations was divided between CC operators and the staff at SOS Alarm. The two dispatch centres, SOS Alarm and the CC were still closely connected to each other.
As a contractor for the national 911 monopoly, an incoming 911 call was received by SOS Alarm and, after assessment, transferred to the CC operator. The double dispatch centre functions were time consuming and increased the risk of misunderstandings, especially when SOS Alarm and CC operators provided differing assessments. Furthermore, the indecisive CC politicians concluded that the CC should do the prioritising, but that SOS Alarm would perform the dispatching. As a result, the calls were transferred three times between SOS Alarm and the CC operator to triage, estimate priority, choose the type of service, and eventually make the dispatch. Technically, the operation become confusing, and though the majority of the SOS Alarm and CC operators simplified the process through consensus, the outcome was that SOS Alarm and the CC were more intertwined than ever.
The patient dials 911 and is answered by an SOS operator who decides whether it is a healthcare matter, a police matter, a fire matter or something else, the calls concerning healthcare are then transferred to us and we determine the priority, after which, according to the rules, it should go back to the SOS operator who is supposed to dispatch the ambulance. This is nuts, it is time consuming and confusing. We normally discuss it with the SOS operator and simplify it together, suggest this and that resource, and it works well in most cases.

(CC operator)
The CC's senior managers were all highly critical of the way the indecisive politicians in the CC allocated tasks. They all emphasised the fact that the structure was counterproductive and impossible to maintain in the long-term. According to the CC managers, the CC dispatch centre was not supposed to be reduced to a priority centre.

Pros and cons in the aftermath of backsourcing
The participants interviewed reported pros and cons in the aftermath of the backsourcing process. Cost-control and arguments concerning quality were particularly difficult to follow-up due to lack of hard data.
The skills of the CC's contracting department were still in doubt, with all CC managers critical of the contracting department's ability to perform risk assessments. The department seemed to expect the collaboration to be free of obstacles.
Unfortunately, the people writing the contracts are a long way from the front line. They don't know much about the service provided . . . In administering the agreement on a day-to day basis, we have to make interpretations of the contract, which is really challenging. Of course, there are conflicts between us and the contractor.
(CC middle manager) One severe challenge was the shortage of registered nurses in the labour market. Ironically, the new CC dispatch centre had difficulties in staffing the centre with RNs and had to use less staff with inferior training. The shortage prompted a controversy on whether trained RNs were in fact more or less skilled to perform triage than the operators at the dispatch centre. SOS Alarm still questioned the need to staff the operation with RNs.
There are a lot of myths. There is no research that shows that a nurse is more skilled than an operator.
(Stakeholder SOS Alarm) Another issue was whether backsourcing would affect end-users, i.e. the patients. The participants were unsure about whether the patients were satisfied with the new backsourced function. This was also verified by the representatives of the regional alliance of patients who were interviewed. There were still no reports from patients who had experienced the effects of backsourcing.
However, the ambulance helicopter staff reported an improvement in the use of the helicopter after backsourcing had been implemented. Senior CC managers checked the figures and concluded that helicopter missions had been reduced by almost 50% six months after backsourcing, rendering the need to invest in another ambulance helicopter redundant. The conclusion made by the CC operators was that they did not over-triage to the degree that the SOS Alarm operators did.
According to the CC managers, in contrast to SOS Alarm, the CC operators were experienced in critical and prehospital care, well-informed about service options within the CC and motivated to find the best individual solutions. The redistribution of cases from high to low level of urgency made room for planning and logistics.

Discussion
The case illustrates the difficulties involved in backsourcing public services from a supplier that has become integrated through a mutual long-term tradition. The close collaboration did not appear to ensure that no conflicts would occur in the future. This is in contrast to what Damanpour et al. (2019) suggest, i.e. that a gradual roll-out or incremental implementation contributes to improved adaptation and is a stimulus for 'managers and employees to cooperate, learn about managing the relationship with suppliers, and acquire knowledge for sustainable use' (p. 6).
One reason for the obstacles to backsourcing may be a reversed relationship between the principal and the contractor. In a free market, a principal will normally reap the benefits of market competition in the hope of both reducing costs and providing more service choices for citizens. The contractors are dependent on the outsourcer to secure extended contracts. Regions and municipalities can therefore normally rely on contractors to do their best to perform their undertakings and be responsive to new demands. If they are not attentive to their clients' needs, the services might be backsourced before the contract period ends due to a lack of risk assessment. However, in the case studied, conflicts were fomented but no simple backsourcing process was in sight. Instead of being a dominant client in relation to multiple service providers, the CC was one of many members of a superstructure organisation. SOS Alarm operated nationally under the government and had a monopoly position. As a result, the case studied was an attempt to end a contract from an inferior position rather than from 'above'. SOS Alarm had the status of an IMC contractor. IMC contractors can have the form of super-organisations, larger and potentially stronger than each of their members (Bel & Gradus, 2017). With an increasing number of collaborative partners, each partner's control over service production is potentially weakened, and when relations sour, divorce can potentially become a huge step with unforeseen consequences (Peteraf, 1993). In the case studied, monopolist characteristics, for example, deciding what the customer needs but being indifferent to what the customer wants (Hirschmann, 1970;Kucuk, 2019), were triggered by powerful shareholders and supported by politicians in the principal organisation.
The principal and the contractor studied were intertwined from the outset. The contractor was dependent on the CCs co-financing emergency preparedness at national level, and the CC was dependent on the unique technology and long experience of the contractors. The symbiotic relationship was a case of mutualism, i.e. both organisations, principal and contractor, benefitting from each other over a period of several years. This transformed into a one-sided dependence, with the contractor seeming to misuse the CC's resources by over-triaging and lopsided dispatching routines, producing growing mistrust. One example of this growing mistrust from the case studied was the establishment of a CC department to monitor the actions of the contractor in order to fine every breach of the contract.
The CC politicians regard SOS Alarm as an important partner and were very reluctant to backsource. They consequently delayed the decision process by questioning the reasons for backsourcing the services. The CCs position as shareholders in SOS Alarm meant that the dissatisfied managers had to strengthen their arguments to justify backsourcing. These arguments focused on promises of improved quality of services and financial gains. However, the scope of such improvements was difficult to predict, with very few figures presented comparing the situation before and after backsourcing. Even though poor accounting, flow, quality and cost-control were still important arguments in applying pressure to make a change.
The challenge of predicting financial and quality effects during backsourcing is well-known. Petalidis (2018) emphasises backsourcing as expensive, and seldom fulfilling expectations. Even though tenders and contracts put measures in place to ensure information transfer between a principal and a contractor, Petalidis (2018) found it to be difficult. The contractor may provide the principal with some information, but data can be concealed or put across informally in a format not useful for calculations. Based on Petalidis (2018), backsourcing will include surprises and non-predictable obstacles.
Most of McDonald's (2018) suggestions regarding reasons to backsource public services were invalid. There were no obvious ideological reasons such as sociopolitical, anti-capitalism, democratic control or equity arguments when the dispatch centre was backsourced. Instead, the case studied may be explained using the framework of Mahoney's (2000) explanation of how long phases of organisational monotony in a changing context induce less room for manoeuvre and trigger dissatisfaction (Van Nieuwaal, 2011). The concept of deinstitutionalisation suggested by Hinings and Malhotra (2008) illustrates the conflicts between CC managers who were dissatisfied and CC politicians who defended the contractor. The decision to backsource the services was, however, considered by the CC managers to be a promising restart in promoting an emergent pre-institutionalisation phase, with improved services and fresh alternatives.
Nonetheless, the interdependent technology and non-separable routines fed frustration (Van Nieuwaal, 2011). It was impossible to terminate the contract and the services were even more intertwined after backsourcing than when they were outsourced. For example, calls were transferred three times between SOS Alarm and the CC operator in order to triage, estimate priority, choose the type of service and eventually make the dispatch. It seemed to be difficult to reach a broad agreement. The expected state of institutionalisation was delayed and new agreements were made with SOS Alarm. One conclusion to be drawn is that backsourcing that concerns two separate politically-controlled public organisations where one is a principal and the other a contractor is not a smooth and unproblematic process. Taking public services back in-house in a quasi-NPM market may lead to quasi-backsourcing. For example, the company contracted remained under contract, the CC was still a shareholder in it, and the collaboration between principal and contractor became more in-depth during dayto-day operations (Figure 2).
The quasi-backsourcing was a consequence of difficulties in maintaining a common policy. Public organisations are often divided between steering committees consisting of politicians with a broad societal involvement and managers loyal to a single organisation. The politicians can also be board members in collaborating organisations and publicly-owned contractors such as IMCs. A mix of loyalties can make negotiations and contracting incompatible. Change processes like backsourcing may be frozen in a frustrating stage of de-institutionalisation based on opposite values and lack of management power (Van Nieuwaal, 2011).
The case studied appears to be the opposite of idealistic expectations of backsourcing. Such expectations may be based on the assumption of a free market, consensus in relation to cancelling the contract, independence in business relations and potential gains (Wang et al., 2018). In the case studied there was no market, no consensus on terminating the contract, the business relation was interdependent, and few obvious gains were reported (Table 2) One important question is whether the case studied is unique, or if the results are transferable to other instances of backsourcing public services. More studies are required to provide principals as well as contractors with practical implications of backsourcing procedures in public services (Damanpour et al., 2019).
A few normative implications may be tentatively drawn from this study. Public organisations which are planning to backsource services from a closely intertwined contractor have to consider whether a divorce is possible or if backsourcing will end in a half-measure. If so, will such a half-measure be sufficient? If not, are there other options available? It may be better to monitor the contractor, refine the contract or even integrate staff and technology. If backsourcing seems to be the only alternative, data has to be collected in order to make realistic predictions of cost and quality. Collecting such data can delay the process but might ultimately be crucial for the outcome.
In order to prevent backsourcing, monopolist contractors may be responsive and listen to requests from the principal. They may also be aware of what is sometimes contradictory and ambiguous management control in organisations governed by committees composed of politicians (see Thakur-Wernz, 2018). Having support solely from politicians and not from managers and staff can be detrimental for the monopolist. An unaware monopolist, relying on the support of politicians in central positions, might be surprised by rapid changes in contexts and procedures.

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

Funding
This study was supported by the Swedish research Council (Grant/Award Number: 2016-01680).