Biased by design – the case of horizontal accountability in a hybrid organization

Purpose – This studyexaminesunexploredhorizontal accountabilitytypes betweenpublic,privateand third sector actors within a hybrid organization. The case organization was applying a novel alliance model to generate service paths for heterogeneous clientele consuming cultural, educational, health and social services. It was first to do so in Finland. Design/methodology/approach – This research is on a case study that used documents and interviews to examine the design of the horizontal accountability. The descriptiveanalysisfocused on identifyingwhat type of formal accountability system was designed (i.e. who is the account holder, and who is accountable and for what and why). Findings – An imbalanced accountability system was identified because accountability obligations were unevenly distributed between public, private and third sector actors. The private sector was the most accountable for performance, and the third sector (i.e. voluntary sector) was the least accountable. As account holders, the public, private and third sector actors were judging their conduct as account providers. This created a biased horizontal accountability system. The hybrid ’ s accountability system was dynamic because the contracts made to establish the hybrid included opportunities to change horizontal accountability if future changes to the external environment affect too drastically the potential to achieve the hybrid ’ s goals. Originality/value – Three new concepts are proposed for studying dysfunctional accountability systems: dynamic, biased and horizontally imbalanced accountability.


Introduction
Public-private partnerships and contracting out are examples of hybrid organizations (Koppel, 2003;Doherty et al., 2014). In such organizations, the public sector is combined with the private sector, third sector or both (Johanson and Vakkuri, 2017). From the perspective of public service accountability, combining the three sectors offers both risks and possibilities. As accountability is a mechanism securing the responsiveness of public services, many have emphasized the importance of accountability in the context of hybrid organizations (Watson, 2003;Hodges, 2012). Here, accountability refers to the capacity of public, private and third sector organizations to check whether the performance of other collaborators responds to citizens' needs and is consistent with the goals of the hybrid.
The common fear with hybrids is that their complexity erodes accountability (Willems and Van Dooren, 2011), and this erosion has adverse effects on public services. This fear is Horizontal accountability in a hybrid organization legitimized by past research (e.g. Humphrey et al., 1993;Ebrahim et al., 2014). Hybrids can create barriers to accountability (Bloomfield, 2006) because they blend together different institutional logics, goals, funding arrangements, ownership and forms of control (Chua andMahama, 2007 andJohanson and; but see also Hofstede, 1981;and Kurunm€ aki and Miller, 2011). In fact, scholars have identified several potential and actual problems in securing accountability in hybrids (Sinclair, 1995). One such problem is the complexity of developing accounting tools that arises when different value logics must be mixed together in hybrids (Grossi et al., 2019). One way to summarize these negative findings from the previous literature is to state that "there is something wrong with accountability in partnerships" between the public and private sectors when hybrid solutions are used to provide public services (Willems and Van Dooren, 2011, p. 525).
There are also many who believe that proper accountability can exist in hybrid organizations (Willems and Van Dooren, 2012) if new solutions are created (Sands, 2006). Some even say that hybrid organizations can end up improving public accountability by introducing new forums for accountability (Willems and Van Dooren, 2012). The contradictory views have not lessened the need to study accountability in hybrid organizations (e.g. Grossi et al., 2017). In particular, the need for new research on accountability solutions in hybrids has been emphasized (Sands, 2006). New solutions in hybrid organizations can mean context-specific systems (e.g. Shaoul et al., 2012) or new ways to use the existing systems of accountability. New ways can lead to problematic accountability schemes, which are also underexamined (e.g. Grossi and Thomasson, 2015).
This article contributes to the studies focusing on the variations in which the existing accountability systems can be used. The study examines what type of formal horizontal accountability system was designed in a hybrid organization that sought to provide a wide range of welfare services with an alliance model incorporating public, private and third sector actors. The research question is further divided into the following four sub-questions: (1) Who is accountable and for what, who is an account holder and what are the forums for accountability in the horizontal accountability system?
(2) Does the proposed system meet the criteria for narrow accountability systems (e.g. Bovens, 2007)?
(3) What are the reasons for choosing the selected accountability system?
(4) How does the design of the hybrid's horizontal accountability system compare to the current theoretical understanding of accountability systems used in public-private partnerships utilizing purchaser-provider models?
To answer the research questions, a case study based on documents and interviews was undertaken. In total, 960 pages of empirical data describing the functions and accountability in the hybrid were analyzed by using descriptive analysis. The object of the study was selected based on its operational construct that offered a compelling theoretical case illustrating three previously unexplored horizontal accountability types in hybrids: biased, imbalanced and dynamic horizontal accountability. The identification and conceptualization of these three accountability types contribute to current theories that are lacking in describing these types of problematic forms of accountability. Moreover, the case organization applied many of the theoretical ideas presented in the literature that have been proposed to improve the accountability of the hybrid organizations (Forrer et al., 2010). As such, the study subject offers possibilities to test the validity of the theoretical ideas proposed by Forrer et al. (2010). The study found dynamic and biased accountability in the examined hybrid. Dynamic accountability is an unorthodox form of accountability because reorganizing governance AAAJ structures such as accountability systems is characterized primarily by blame avoidance and not accountability (e.g. Hood, 2011). In fact, accountability systems often require some level of stability because it is hard to define account givers and pass judgment if responsibilities and accountable actors constantly change. Biased accountability is considered to be unorthodox because the literature assumes that an account giver and account holder are different actors (Bovens et al., 2014), whereas in biased accountability, the account givers and account holders are considered to be the same actor as the public and private sector actors have both dual roles.
Next, the section two defines the concept of accountability used in this research. In section three, the article compares public, private and third sector accountability systems. This comparison provides a perspective on accountability systems that are being mixed in hybrids. In section four, the study steps into the world of hybrid accountability by first presenting the primary organizational characteristics of hybrids and by then explaining how those characteristics generate the anatomy of a hybrid accountability system. The fifth section presents the research method and the case organization and its design. The seventh section analyzes the designed hybrid accountability system and compares it to the criteria of narrow accountability and theoretical examples of purchaser-provider models. The seventh section also considers how hybridity explains the accountability system. The last section provides a discussion of the results.

On the concept of public accountability used in this study
To determine who in the hybrid was accountable to whom and for what, we use the definition of narrow accountability developed by Bovens (2006), although there are various other definitions in the study field (Bovens et al., 2014;Jayasinghe et al., 2020). The chosen concept is a study design choice that limits the scope. According to Bovens (2006), public accountability: (1) is a relationship between an actor and a forum (2) in which the actor is obliged (3) to explain and justify (4) his conduct (5) the forum can pose questions (6) pass judgment (7) and the actor may face consequences In Bovens' (2006) definition of accountability, the actor as an account giver refers to individuals (i.e. servants) or organizations. These organizations can be government agencies, local governments or other public organizations (Bovens, 2007, p. 450). The accountability forum is the organization, individual, institution or virtual entity, such as public opinion. For example, superiors, journalists and public agencies can act as accounting forums that examine how account givers have conducted their duties. At this point, however, we deviate terminologically from Bovens' definition by using the concept of account holder when we refer to the actor who receives the account. Moreover, an accountability forum refers to a forum in which the debate between account holder and giver takes place. The concept of the accountability forum we use in this study is adopted from Willems and Van Dooren (2011). Separating accountability forums from account holders makes the scientific communication clearer, as there are meetings that act as forums, but not everyone in the meeting is an account holder or account giver.

Horizontal accountability in a hybrid organization
The actor's obligation to explain and justify his conduct is either formal or informal. Here, a formal obligation means that the actor gives an account regularly to specific account holders (i.e. courts, auditors, etc.). Press conferences, voluntary audits and casual briefings are all examples of informal ways to fulfill the obligation to explain and justify the conduct of the actor. By giving information about the process and task performance and the outcomes of these performances, the account giver informs his conduct to the account holder. Besides the information about the conduct, informing the account holder includes the provision of explanations and justifications. Justification for the conduct is required, especially when failures have occurred. The actual conduct can relate to financial accountability, legal accountability, political accountability and other types of accountability recognized in the literature. After informing, the account holder needs the opportunity to interview the account giver so that it can question the provided information or the justification for the conduct. Moreover, the account holder can "pass judgment on the conduct of the actor" (Bovens, 2007, p. 451). In this judgment, the account holder can accept the given account or condemn the conduct/policy. In the case of negative judgment, sanctions can be imposed on the actor (Bovens, 2007).
Accountability in public, private and third sectorsthe ingredients of the hybrid's mixture Accountability has been a controversial concept (Sinclair, 1995), but accountability means answerability and expectation management in this research. Here, answerability refers to the state where one has to answer for his/her actions to an individual or body (Flinders, 2001). Prior expectations for such actions are preconditions when one is holding someone answerable (Willem and Van Dooren, 2011). Therefore, answering for one's own actions becomes expectation management.
To whom are public sector actors accountable is an old question. Accountability was retained at the political level because politicians needed to confirm whether the public administration had executed the orders derived from political deliberations (Goodnow 1900;Wilson 1887Wilson /1987. As the complexity of the public administration increased in the 19th century, scholars needed to look beyond the politics-administration dichotomy and consider what bureaucratic accountability for administrators meant (Williams and Taylor, 2013). As the years went by, many other stakeholders demanding accountability from the public sector were identified (Almquist et al., 2013). Today accountability is seen as a method that corresponds to the various "expectations generated within and outside the organization" (Romzek and Dubnick, 1987, p. 228). Thus, accountability is a "multidimensional product of many forces operating in the accountability environment. . ., best described as emerging from their multi-stranded interactions" (O'Connell 2005, p. 86).
Accountability in the public sector can be achieved in many ways, such as financial reporting (e.g. Oulasvirta, 2014). According to Finer (1941), accountability is generated by using external controls such as laws and political means. In contrast, Friedrich (1940) argued that internal control (i.e. professional codes of conduct) creates accountability. Polidano (1998) mentions indirect and direct accountability. Hodges (2012) talks about vertical and horizontal accountability, whereas Willems and Van Dooren (2011) add that there are political mechanisms (e.g. political debate within political parties and mass media), administrative mechanisms (i.e. supervisors demanding accountability from subordinates), judicial mechanisms (courts prosecuting lawbreakers) and societal mechanisms (grievance procedures and complaints) driving accountability in the public sector.
In the private sector, accountability has traditionally focused on ensuring that a company makes a profit while operating within the established legal framework (Williams and Taylor, 2013). However, the forms of corporate governance have varied depending on the context (Shaoul et al., 2012). The private sector has applied professional accountability (Elms and Phillips, 2009) and grievance procedures similar to the public sector (Mulgan, 2000). Similarities do not end there, as the public sector has implemented private sector internal controls for administrative control purposes (Van der Kolk, 2019). In addition, the corporate social responsibility movement argues that to be legitimate, companies cannot focus only on customers and shareholder profits. Therefore, companies seeking for more legitimacy must consider wider groups of stakeholders and use interactive dialogues (Brennan and Solomon, 2008). Interestingly, the mechanisms to implement corporate social responsibility resemble the accountability mechanisms of the public and third sectors (Hohnen, 2007). Overall, the change driven by the corporate social responsibility movement has made private sector accountability more similar to public sector accountability. Although the corporate social responsibility movement is strong, activities "improving accountability are intended not so much to extend the mechanisms of public accountability to private organizations as to establish separate, parallel mechanisms mimicking to some extent those found in the public sector" (Mulgan, 2000, p. 95).
Despite the similarities, private sector accountability is not identical to public sector accountability. In companies, there is no politics-administration dichotomy, political opposition or administrative accountability of public servants. Moreover, public scrutiny is not such a pressing issue in the private sector compared to the public sector (Mulgan, 2000), and there are differences in accounting practices in these two sectors that can be seen in reporting practices and calculus techniques (Oulasvirta, 2014). This indicates that hybrids have to mix accountability mechanisms.
Potential organizational forms and service areas in the third sector put accountability in a challenging position. On the one hand, to get resources for their basic functions, third sector organizations have to act like profit-making companies. This means they have to keep strategies private, sustain financial solvency and ensure their funding by out-performing their competitors in fund-raising (Williams and Taylor, 2013). On the other hand, third sector organizations can function similarly to quasi-governmental democratic organizations because they are ensuring fairness, equality, participation, accountability in decisionmaking (Ackerman, 2004) and adherence to laws and regulations (Phillips and Levasseur, 2004). The balancing act between private and public accountability depends mostly on the operational context, organizational size, image, production capacity, service area, availability of voluntary workers, and financial solvency of both donors and the third sector organization (Dees and Anderson, 2003).
Studies of third sector accountability demonstrate that the concept of accountability is contingent and arises from social deliberation occurring between organizations and their stakeholders (Williams and Taylor, 2013). Thus, the stakeholder-based approach to accountability appears to be as relevant in the third sector context (Najam, 1996;P€ arl et al., 2020) as it was in the private and public sectors. The initiative to be accountable can be internal or external (Christensen and Ebrahim, 2006) because third sector actors can be held responsible by staff and volunteers (lateral or administrative accountability), clients (downward accountability) and others (upward accountability) (Williams and Taylor, 2013). Moreover, the third sector can be an active player constructing accountability systems and defining what accountability is in the third sector. There are similarities between public, private and third sector accountability mechanisms. However, there are also significant differences. Compared to the public sector, there is no dichotomy of politics-administration, political opposition or administrative accountability in the third sector. The differences between the private sector and the third sector are also clear; investors and shareholders do not dominate the accountability framework in the third sector. When compared to public and private sector organizations, the third sector's accountability system often serves communal needs not addressed by companies or public organizations.

Horizontal accountability in a hybrid organization
Anatomy of hybrid accountability Typically, hybrid organizations blend elements of different established organizational categories to create new organizational forms Dorado, 2010, p. 1419;Scott, 2008). Johanson and Vakkuri (2017) claim that the results of this mixing of elements are evident in many aspects of the hybrids, such as goal incongruence, different institutional logics (Pache and Santos, 2013;Skelcher and Smith, 2015;Gebreiter and Hidayah, 2019), differentiated forms of economic and social control (Jordana and Levi-Faur, 2004), mixed ownership structures (Thynne, 2011) and the variety of financing sources used by hybrids (Hodge and Greve, 2007). According to Evers et al. (2002), hybrid organizations blend resources, goals, interests, decision-making structures, incentive systems and cultural practices.
So far, the literature has argued that hybrid organizations mix those things that are different in the public, private and third sectors (Kastberg and Lagstr€ om, 2019). However, listing those different organizational aspects is, to some extent, a futile exercise because of the differences between sectors change over time. Lately, the public sector has become more like the private sector due to the New Public Management developments (Funnell and Cooper, 1998). Enumerating all the differences between the three sectors is difficult because there can be so many (Rainey et al., 1976). Based on previous studies (Billis, 2010), it is crucial to know where the main differences exist. To define the differences, this study used the list of critical principles summarized in Table 1. In Table 1, the authors of this study have added content from previous literature to the principles identified by Johanson and Vakkuri (2017).
In Table 1, the last row addresses forms of social and economic control and includes the accountability mechanisms that separate the public sector from the private and third sector organizations. Although Table 1 seems to suggest otherwise, it should be emphasized that different forms of social control do not capture all aspects of hybrid accountability. To understand hybrid accountability more comprehensively, one needs to be aware of the connections between other features of hybrid organization and accountability. To provide a preliminary idea about the mixture of public, private and third sector accountability systems that are blended in a hybrid organization, Figure 1 portrays some of these connections.
As we review Figure 1, it is easy to accept the notion that hybrids can be complex (Menard, 2004;Krauss et al., 2020). The complexity may introduce problems (e.g. Costa and Andreaus, 2020). For example, the hybridity has eroded the traditional notion of accountability and caused difficulties by weakening political control while emphasizing performance function of accountability in which performance is measured by the extent to which it achieves stated goals (Willems and Van Dooren, 2011).
Lack of political control is not the only problem with hybrid accountability reported in the literature. Think here situations where performance measurement systems often operate within organizational boundaries limiting the performance measurement of hybrids (Kurunmaki and Miller, 2006). Information systems operating within the organizational boundaries can also erode accountability as essential information for accountability purposes cannot be provided. Thus, a common criticism is that hybrids lack transparency because commercial confidentiality inhibits the sharing of the necessary performance information (Coghill and Woodward, 2005).
Power has been used to explain the lack of common information systems in hybrids (Rajala et al., 2020). Power, however, is related to the features of hybrid, such as ownership, governance, operational priorities, distinctive human resources, financial sources, goal incongruence and institutional logics. For example, when the public sector does not own the company or majority of its shares, it cannot use the owner's power to decide that the joint information system between public and private sector actors produces shareholder value, and thus it should be adopted. As employees are paid to generate shareholder value through sales, the inability to determine what produces shareholder value or increases market sales AAAJ indicates a lack of power over company employees. The company cannot force legally backed agencies of the public sector to adopt a common information system that does not produce public value according to collective choice of citizens concerning the right market interventions producing public value. The company does not have the power to override the collective choice made in elections.
Features of hybrids also relate to hybrids' well-functioning accountability systems. According to Forrer et al. (2010), establishing a well-functioning accountability system in hybrids depends on clarifying responsibilities in the collaboration. The act of clarifying responsibilities is an integral part of social control. Forrer et al. (2010) add that six dimensions require attention in setting up the hybrid accountability system. First, to ensure the functionality of the hybrid organization in the long run, there needs to be sufficient support in political and social spheres. Political and social support are directly related to ownership, governance and operational priorities as features of hybrids. Second, an in-depth costbenefit analysis should be conducted to see whether establishing the hybrid is rational.  (Billis, 2010) Paid public servants in a legally backed agency Paid employees in a managerially controlled company

Members of volunteer associations
Distinctive other resources (Billis, 2010) or financing sources (Johanson and Vakkuri, 2017) Taxes, prices, and transfer payments from other public sector actors

Sales and fees
Dues, donations and legacies Goal incongruence (Johanson and Vakkuri, 2017) Create public value, that is, achieve the result of a production process whereby inputs are allocated to output production and used to generate outputs leading to desired outcomes (Van Dooren et al., 2015) Create shareholder value (Barsky et al., 1999) Create social value, that is, societal betterment through social inclusion in the field of activities lacking government or private sector production (Akingbola et al., 2019) Institutional logic (Johanson and Vakkuri, 2017) Intervene in markets when needed (Samuelson and Nordhaus, 2005), serve the interests of the nation (Liu and Tang, 2011) Compete for customers and market share against other firms to generate revenue (Porter, 2008) Goals of individuals, emotional attachment of members, credibility (Williams and Taylor, 2013), mission Forms of social control (Johanson and Vakkuri, 2017) Political control and administrative control of public servants (Willems and Van Dooren, 2011) Corporate governance, private sector accounting standards, and shareholder pressure (Brennan and Solomon, 2008;Oulasvirta, 2014) Rules of membership and guidance provided by donors and stakeholders (Williams and Taylor, 2013;Ackerman 2004;Hyndman and McConville, 2018) Table 1.

Features of the hybrid
Horizontal accountability in a hybrid organization Rationality and cost-benefit analysis depend on goals, and goals relate to goal incongruence. Third, ensuring that goods and services are meeting expectations and that someone is accountable for their provision means that the allocation of risk is balanced. This relates to social control. Fourth, flexibility in the alliance and clear expectations for coordination have to be explicitly stated in hybrid arrangements. Flexibility and coordination are features of governance.
Fifth, the organizations forming the hybrid must clearly communicate the expertise and performance they expect from the other partners while being open-minded about the knowledge and experience of other organizations. The fifth point in Forrer et al.'s (2010) list connects to goal incongruence while setting up demands relates to ownership and human resources. Sixth, confirming the hybrid's execution of public policy may require the measurement of performance. Measurement, on the other hand, is a form of social control. The hybrid organization studied in this research meets the criteria set by Forrer et al. (2010), as can be seen from Appendix 3. Therefore, this study provides an opportunity to test the ideas of Forrer et al. (2010).

Research method
This empirical research was based on a case study design (Patton, 1990) because the research area is new and current knowledge is inadequate (e.g. Eisenhardt, 1989). The goal of the case selection was to serve the theory-building associated with dynamic, biased and horizontally imbalanced accountability (e.g. Thomas, 2011). The theory is being built in this study deductively in Figure 1 by joining together different research results and through inductive reasoning in the empirical analysis (Niiniluoto, 1999). The study was cross-sectional by nature, as we observed only a snapshot of the accountability system designed in the hybrid organization. The empirical data were assembled from two sources: documents (n 5 44) and interviews (n 5 9). In total, 873 document pages and 87 pages of interview transcripts were

Goal incongruence:
Reveals that account givers may have to take into consideraƟon different values in their tasks and account giving and account holders need to consider these values when they pass judgement

InsƟtuƟonal logics:
Provide interpretaƟve schemes for account givers acƟons and reporƟng and for account holders judgment (e.g., schemes guiding to generate and report revenues and maximize profits versus policies encouraging to intervene in markets and report well-being outcomes of such intervenƟons)

DisƟncƟve other resources and financial sources:
Reveals accountability forums (poliƟcians vote taxing policies, customers vote with their wallets and donators with their donaƟons) Hybrid accountability as social form of control Interactions between hybrid accountability and other features of the hybrid AAAJ analyzed. Documents were the main source of information, whereas the interviews were used to confirm and supplement the findings from documents. The interview questions were pretested before interviews were conducted in the summer and fall of 2018. The interview questions addressed goal setting, implementation responsibilities, incentive systems and monitoring systems. Questions about the incentive systems and monitoring systems were mainly used to obtain information about account holders, accountability forums, rewarding and sanctioning functions whereas implementation responsibilities often revealed account givers. The semi-structured interviews lasted about an hour, were conducted face to face, audiotaped and transcribed verbatim. The interview process could be described as snowball sampling because the interviewees were chosen in collaboration with the city of Tampere and its representatives and members of the hybrid organization. As shown in Appendix 1, we achieved good coverage from different accountability forums, and several interviewees attended multiple forums. The interviewees also represented different policy fields. The documents were collected from publicly available archives of the public, private and third sector organizations to get good coverage of different perspectives.
The quality of interviews can suffer from subconscious bias and potential inconsistencies (Brown, 2001). In this research, we attempted to control any biases of interviewees by using document data to triangulate the findings from the interviews. In this triangulation, we determined there were few contradictions between the documents and interviews. When there were contradictions, we relied on the information in the official documents, such as contracts with clear descriptions of organizational responsibilities. A researcher who did not conduct the interviews controlled the possible bias resulting from the interviewer by using recordings of the interview. No biases relating to the interviewers were observed during peer evaluations of the recordings.
The descriptive analysis applied two analytical frameworks. The first framework is presented in Tables 5 and 6 in the results section. The first framework was used to answer two questions: (1) Who is accountable and for what, who is an account holder and what are the forums for accountability in the horizontal accountability system?
(2) Does the proposed system meet the criteria for narrow accountability systems (e.g. Bovens, 2007)?
In Tables 5 and 6, narrow accountability existed if all five conditions mentioned by Bovens (2007) were present in an accountability forum in the hybrid. The second framework is presented in Figure 1. It was used to answer the research question three, asking what are the reasons for choosing the horizontal accountability system. The last part of the descriptive analysis uses a theoretical example of a public-private partnership to study how does the design of the hybrid's horizontal accountability system compares to the current theoretical understanding of accountability systems used in public-private partnerships utilizing purchaser-provider models. Overall, the descriptive approach can be used when the researcher is attempting to describe what exists (Bickman and Rog, 2008). Descriptive research cannot be used to identify causality. Therefore, the explanations that clarify why the accountability system evolved the way it did should be considered as descriptions of how things occurred in the case organization. These events can be case-specific and represent an anomaly.

Case organization
The examined hybrid organization was an inter-organizational alliance formed by a municipality, private company and third sector organization. This alliance was supposed to Horizontal accountability in a hybrid organization provide a wide variety of cultural, educational, health and social services to the 20,000 inhabitants of the suburbs where it was located. In public-private partnership literature, this type of hybrid is known as an advantaged partnership model (Willems and Van Dooren, 2011). The case study was in the city of Tampere, Finland.
Next, different aspects of the hybrid organization are described in three parts. In the first part, the reasons for the adoption of the alliance model are described. The second part describes the complex design process of the accountability system in the hybrid organization. Illustrating the design process makes it clear how the criteria in Forrer et al. (2010) for good accountability were reflected in the examined hybrid. The third part of the analysis reviews the responsibilities in the hybrid and analyzes which actors can be potential account givers and account holders based on their competencies and roles in the development and implementation phase of the hybrid.
The rationale for forming the hybrid organization The municipality did not apply the heavily criticized purchaser-provider model to the case organization because of the poor results achieved with past purchaser-provider arrangements. In fact, the hybrid was created by applying an alliance model. Thus, the case study organization represented a change from one existing system of accountability to another one in the municipality's welfare service production. The municipality forming the hybrid was previously successful with the alliance model in public-private partnerships focusing on construction projects. It was, in part, the success of new infrastructure and the desire to solve the problems of purchaser-provider models used before in cultural, educational, health and social services that encouraged the municipality to test the alliance model in the delivery of welfare services. The problems with the purchaser-provider model and the solutions to these problems provided in the alliance model are summarized in Table 2.
Designing the hybrid organizationthe complex planning process Privatization, contracting-out, concession public-private partnerships and alliance publicprivate partnerships are different types of service production models (Willems and Van Dooren, 2011). The alliance model applied to the case study organization differed from the alliance public-private partnership model described by Willems and Van Dooren (2011) because Tampere incorporated a third sector actor in its alliance. Tampere had used the public-private partnership alliance model before, but it had not been applied to educational, cultural, social and healthcare services before. In this sense, the hybrid organization was a welfare alliance that opted to use a variation of the existing accountability system used in previous applications of the alliance model in Tampere. The alliance model served as a framework guiding the establishment of the hybrid constellation formed by a private company, a municipality, and a third sector organization. Here, the private company was a consortium.
The alliance model is basically a service production and planning model guiding procurement, contracting, planning and implementation processes. The alliance model enables cooperation by using contracts encouraging the purchaser and providers to work toward common goals through close cooperation. Trust, commitment, collaboration, joint contracts and incentive schemes, a virtual organization made to track the hybrid's costs, shared development and risks and a focus on preventative services are critical features of the alliance model. These features are used to tackle the challenge arising from sub-optimization that can occur in public-private partnerships (Document 4). In general, the alliance model is used in Finland for complex projects. Using it in simple projects is not recommended because the model requires a lot of effort and collaboration (Interviewee G).

AAAJ
The development of the hybrid organization began with an independent preparatory phase in which the purchaser evaluated its activities using data analysis and knowledge management techniques. The primary goal of the data analysis was to identify those social and health services that suffer from a lack of skilled staff and are complex, challenging and even risky by nature. The data analysis also investigated the markets for social and healthcare services, drew up preliminary goals and requirements for hybrid's services and modified the alliance model used in construction projects to fit the purposes of the welfare center. Part of the preparatory phase included citizen hearings as well as surveys and workshops for personnel. These were used for the purpose of innovation.
Old purchaser-provider model Problem with the purchaserprovider model Alliance model and its solution to problems with the purchaserprovider model Purchaser is separated from provider and purchaser orders the services produced (1) There is no joint information system because of the separation (2) Purchaser's and providers' solutions about the ordered services did not succeed adequately (1) There is a joint information system provided by the municipality (2) Purchaser and provider form a joint informal organization and make unanimous decisions about service production together Purchaser organizes the tendering and announces the services to be provided and the goals for the service production (1) The provider did not have any say relating to the goals of the hybrid (2) Service outcomes and customer needs received little attention while outputs were under the focus (1) Joint goals are set by the purchaser and provider (2) Customer needs and service outcomes become center stage in operations The purchaser chooses the provider and uses contracting to steer the actions of the provider that can be public, private or third sector provider (1) The emphasis is on sanctions (2) Contracting is a game of risk transfer (3) Sub-optimization is typical (1) The emphasis is on rewards (2) All either win or lose together indicating that risks and benefits are shared which minimizes sub-optimization Customers choose a service provider or go to an available service provider assigned to them and use the service (1) Sectoral thinking does not encourage cross-sectoral service paths that would be useful to the service user (2) Customers are not involved enough in development (1) There is joint responsibility related to the development of cross-sectoral service paths (2) Co-production applications are utilized frequently in planning the facilities and services of the hybrid as well as in arranging social events in community cafe Provider reports about the results and the purchaser monitors the results (1) The purchaser is kept in the dark in important matters by using the commercial confidentiality (2) The monitoring of the services is lacking; learning possibilities are limited and quick changes are not possible (1) The alliance applies an open book principle that makes rewards, planning and billing information transparent and available to the partners (2) Joint measures between the collaborators (3) Continuous co-development, quick experiments in the service production model, and moderating operating models are enabled in a hybrid organization A competitive tendering phase followed the preparation phase, and the timetable for this phase is shown in Appendix 3. The purpose of the competitive tendering was not only to choose a service provider but also to develop and refine the future of the hybrid organization. The competitive tendering was implemented as an interactive procedure through the public procurement process. Companies and third sector organizations were invited to attend co-development sessions (see Appendix 3). The service producer was selected after competitive bidding ended, and the producer was designated a partner rather than a service producer in the contract language. That language was important because the partnership also reflected accountability relationships, as we demonstrate later in the empirical analysis.
The design of the hybrid organization continued in the development phase in which the contracting parties jointly defined the goals for the hybrid organization, planned the implementation phase, agreed on project costs and identified the key performance indicators (see appendix 3). A large part of the development phase was the development of the so-called commercial model. The commercial model is a contracting model defining what needs to be in alliance contracts. One of the biggest aims of the commercial model is to promote transparency of costs and service production by using open books. The commercial model also generates a common incentive and risk-sharing system for the purchaser and providers and enables dynamic actions needed to respond to changing conditions. In practice, the commercial model encourages service providers to act in harmony with the organizer's interests (Document 2). Of course, the commercial model is described in a favorable light in the contract documents, but it remains to be seen whether it works in practice. The Pirkanmaa regional government and Sitra (an independent trust fund set up by the central government to promote future development and economic growth) served as external consultants in the development phase. Based on the development phase, the purchaser decided to proceed with the long-term implementation phase. Thus, the hybrid is set to operate the next ten years based on the implementation contract made in the development phase. We examined the hybrid organization in the early stages of implementation but were able to focus our review only on the designed horizontal accountability system.

Roles and responsibilities in the hybrid organizationagents and principals
The actors who could be accountable for the conduct of the hybrid organization were organizations and individuals. This study focused only on the organizations as accountable actors because they are responsible for attaining the goals set for the hybrid organization according to the empirical data. There were three possible account givers in the hybrid organization: the municipality, private company (i.e. Mehil€ ainen) and the third sector organization known as Setlementti. Considering the different services provided by the hybrid organization (see Table 4), many of the goals in Table 3 are not relevant to all service providers. For example, the library can affect customer satisfaction, but it does not create treatment plans. Thus, the goals place more pressure on those service providers who provide health services because most of goals relate to health.
To understand who is accountable to whom and what they are accountable for, the roles and responsibilities of the municipality, private company and third sector organization need to be identified. The municipality "has several distinct roles" (Document 2, p. 9). The first is the role of the owner or purchaser (Document 4). As a purchaser, the municipality is responsible for ordering and financing the services from the hybrid organization (Documents 1 and 2). The municipality and the private company determine through the contract what is ordered and at what cost. Purchasers also participated in defining goals, as reported in the following quotation:

AAAJ
We had a working group focusing on goals and outcome measures. There were experts from Tampere [the municipality], Mehil€ ainen, region of Pirkanmaa, and Sitra. . . in these working groups, we started thinking and defining the key performance areas . . . (Interviewee A) The second role of the municipality is a service provider (Document 7). The municipality provides a library, youth center, Domiciliary care services and employment services as well as services for children and families (see Table 4). A walk-in clinic is also a municipal service provided in the hybrid organization. The third role is the one in which the public sector operates as a customer receiving the services produced by the hybrid organization (Document 1). In the role of customer and purchaser, the municipality can act as an account holder in a hybrid organization. However, operating as a service provider, the municipality is an account giver.
The primary role of the private company was to act as "the service provider" (Document 2, p. 7). The service provider role included general health, oral health and enhanced housing services. In this role, the company can act as an account giver. The competitive tendering was based on the goals of the purchaser, whereas the contract used to establish the implementation of the hybrid organization was based on goals developed jointly by the public and private sector actors (Document 1). The arrangement used to create goals for the contract made the private sector company a principal in the hybrid organization. This means that Mehil€ ainen could have dual roles as both account holder and account giver in the horizontal accountability forums.
The municipality and the third sector organization had a separate contract between them. That contract required two things from the third sector organization: they should run a community caf e and take people from work rehabilitation programs to work in the caf e (Interviewee E). This contract between the public and third sector actors indicated that the third sector could operate as a service provider in the hybrid organization. According to other Goals Performance measures* 1. The welfare of the children and youth (1) Survey tracking life satisfaction, perceived health and oral health of fourth and fifth graders (2) Survey tracking perceived health, symptoms and sicknesses, sexual health, oral health and functionality of family life.

Coverage of major national diseases
(1) Treatment plans are made for coronary heart disease and diabetes patients between the ages of 18 and 65 (2) Treatment plans are made for adult oral health care patients (3) Percentage of long-term blood glucose and LDL cholesterolbalanced patients with type 2 diabetes and coronary heart disease patients 3. Welfare of elderly people 4. Health service use of the elderly 5. Elderly's ability to live at home (1) Elderly's ability to function and live at home (2) The welfare of the elderly and reduced need to use emergency services (3) Per-patient cost of hospital services for the elderly 6. Curbing the cost of special medical care (1) Per-patient cost of special medical care 7. Customer satisfaction 8. Collaborator satisfaction (1) Customer satisfaction for services (1) Collaborator satisfaction 9. Stay within the budget (the four-year budget was 56,914,399 euros) (1) The combined overall cost of units in the welfare alliance Note(s): *There is clear performance standard set beforehand for each measure and this standard is taken from past performance, but this table does not include these standards  Table 4. Actors and hybridity in the case organization AAAJ documents, the third sector organization arranged social events in the welfare alliance and provided a community caf e. Moreover, the third sector organization took care of these space reservations of the citizens who could reserve spaces from the welfare alliance for their activities.
The third sector organization was not involved in the goal-setting process, which was overseen by the municipality and Mehil€ ainen (Document 1). Not taking part in goal setting indicates that the third sector was not in the role of principal. However, the third sector was included in the hybrid organization steering group that acted as an account holder when it awarded bonuses. The following quotation shows this: Health center, oral healthcare, maternity and child health clinic, domiciliary care, and housing service unit, library, youth services, and then this info desk and the third sector. From all of these, we have named one person who is in charge of the activities, and these persons form the steering group.
Based on this reporting, the third sector can have dual roles because it attends the meetings of the steering group, where it acts as both account giver and account taker when bonuses are awarded from the savings.

Empirical analysis
The empirical analysis is divided into three parts. In the first part, the study examines the formal horizontal accountability system in the hybrid to provide answers to the following research questions: (1) Who is accountable and for what, who is an account holder and what are the forums for accountability in the horizontal accountability system?
(2) Does the proposed system meet the criteria for narrow accountability systems?
In the second part, the analysis provides explanations about the horizontal accountability system by using the theoretical framework from Figure 1. The second section answers the third research question asking the following: What are the reasons for choosing the selected horizontal accountability system? The third part of the empirical analysis compares the designed accountability system in the alliance model to a theoretical model of accountability in public-private partnerships to understand what changes in the alliance model. The third part contributes to the literature by describing new ways to use accountability systems in public-private partnerships. The new application of the public-private partnerships is demonstrated by answering the fourth research question: how does the design of the hybrid's horizontal accountability system compare to the current theoretical understanding of accountability systems used in public-private partnerships utilizing purchaser-provider models?

Formal horizontal accountability in a hybrid organization
In the hybrid organization, there are five possible accountability forums for the municipality, private company and third sector organizations. These are the development group, the communication group, sectoral staff meetings, steering group and the alliance management group. The results from horizontal accountability in staff meetings and development and communication groups are presented in Table 5. The development group involves representatives from the health center, oral healthcare, maternity and child health clinic, domiciliary care and housing service unit, library youth services and community caf e (Interviewee D). The representatives of the service units bring information on performance to the sessions, but the reporting is informal. The development group can pose questions to the Horizontal accountability in a hybrid organization Possible formal accountability forums in the hybrid Examined question relating to Boven's (2006) conditions of narrow accountability Communication group Development group

Staff meetings
Is there a relationship between the account giver and account holder?
Public, private and third sector actors have a relationship, but it is not a formal accountability relationship Is the account giver obligated to explain and justify his/her conduct in the accountability forum?
Public, private and third sector actors are not formally obliged to justify their conducts in the forum Can the account holder pose questions?
Public, private and third sector actors cannot pose questions about the performance as there is no formal reporting or account giving in this forum Can the account holder pass judgment?
Public, private and third sector actors cannot pass judgment on the performance in the role of the account holder as this is done in either steering or management group Can the account giver face consequences?
Public, private and third sector actors cannot face formal consequences in the forum Examined question relating to Boven's (2006) conditions of narrow accountability Steering group Management group Is there a relationship between the account giver and account holder?
There is a relationship between third sector actor as an account giver and public, private and third sector actors as account holders There is a relationship between the public and private sector actors who have dual roles as account givers and holders Is the account giver obligated to explain and justify his/her conduct in the accountability forum?
The third sector actor is not formally obligated to explain and justify its conducts The private sector is obliged to explain and justify input, output and outcome results and public sector is obliged to explain and justify input results Can the account holder pose questions?
Public, private and third sector actors cannot pose formal questions to third sector actor as the third sector does not provide formal reports to the forum The public sector can pose questions on input, output and outcome results and private sector can ask questions about the input results Can the account holder pass judgment?
Public, private and third sector actors can pass judgment on the performance of the third sector actor The public sector can pose judgment on the private sector's input, output and outcome results and private sector can pose judgment on the public sector's input results Can the account giver face consequences?
The third sector can receive rewards if the hybrid achieves budget savings and steering groups decides to reward the third sector The private sector can receive rewards or sanctions from input, output and outcome results whereas the public sector can receive rewards or sanctions from input results Table 5. Study results relating to accountability design in staff meetings and communication and development groups Table 6. Results of the study relating to the accountability design in steering and management group AAAJ representatives and pass judgment on past activities but only in an informal manner. Lastly, no formal sanctions or rewards to the public, private or third sector organizations can be given in the development group (Document 1). Thus, public, private and third sector organizations are not formally accountable in the development group.
The role of the communication group is to go through issues relating to the whole welfare alliance by keeping "monthly briefings that are informal and kept in the community caf e" (Interviewee F). The communication group does not act as an accountability forum because account givers do not need to report anything formally and account holders cannot impose rewards or sanctions on public, private or third sector organizations in the forum. Sectoral staff meetings typically occur within the three organizations and include workers from only one sector, making it impossible to exercise horizontal accountability between public, private and third sectors. The staff meetings on oral care are an exception because there are both public and private sector workers. In oral care staff meetings, account givers are present, conduct is explained, and questions can be asked in an informal manner, but formal sanctioning and rewarding the public, private or third sector organizations is not possible. As such, staff meetings do not make the municipality, private company or third sector organization horizontally accountable in the hybrid.
"The steering group manages the operational activities" (Document 2, p. 5), and "gathers together every two week" (Interviewee D). In the steering group, "financial matters are reported monthly" (Document 4, p. 32). It has representatives from "health center, oral healthcare, maternity and child health clinic, domiciliary care and housing service unit, library youth services and then this info desk and third sector" (Interviewee A). The results concerning the steering group are presented in Table 6.
The steering group is a formal accountability forum where account holders can pose questions to the account givers. Those questions can address costs, outputs and outcomes. Account holders are representatives of the service units participating in the steering group meetings, whereas the account giver is the third sector organization. Thus, the third sector operates in dual roles in the steering group because it is at the same time account giver and account holder. The obligation to explain and justify actions in the steering group is an informal one and is concerned more with explaining the conduct of the service units (Interviewee B). Costs and outputs are often explained, but outcomes are addressed less. Formal reports would be too demanding for the third sector organization as it does not have any formal measurement systems for tracking their performance. As interviewee E stated: "we [workers in community caf e] cannot do anything other than observations." The steering group can pass informal judgments, but it cannot give formal sanctions to the three organizations. Still, the steering group generates some form of accountability for the hybrid organization. For example, if the total costs of the welfare alliance are under budget in the budgetary season, the steering group decides whether rewards to the staff of the hybrid organization are given. Moreover, the steering group can choose " to use this sum of money to reward the third sector actors" (Document 2, p. 21). How to reward the third sector is not specified, leaving those options to the steering group. The option to reward or not reward the third sector actors makes the steering group an account holder.
To summarize, the steering group can reward staff and the third sector, but it cannot reward public or private sector organizations. Therefore, the third sector organization is the only one accountable in the steering group. However, even they cannot face any sanctions if output and outcome goals are not met or when costs exceed the budget. In addition, accountability does not exist in the way defined by Bovens (2006) because the third sector organization is part of the forum that decides whether or not the third sector is being Horizontal accountability in a hybrid organization rewarded. The third sector also justifies its conducts in an informal manner as it does not have a performance measurement system.
The management group is the main decision-making body in the hybrid organization. The duties of the management group include the following (Document 1): (1) Resolve disagreements and exercise the ultimate authority in the hybrid organization (implementation role: account giver).
(2) Accept changes in goals and provide strategic guidance to the hybrid organization (strategic role: account holder).
(3) Define the authority for the service unit managers and specify procedures (implementation role: account giver).
(4) Appoint a manager for the hybrid organization and accept proposals from the service unit managers who participate in the steering group (implementation role: account giver).
(5) Communicate with the purchaser and stakeholders (reporting role: account giver).
(6) Review and approve any changes to the service plan and management system described in the contract used to establish the hybrid organization (implementation role: account giver).
(7) Monitor the policies and performance of the hybrid organization (account holding role: account holder).
As can be seen from Tables 5 and 6, the management group is the primary accountability forum in the hybrid organization because it develops all the organization's goals and monitors their achievement with performance measures. Because all the decisions of the management group must be unanimous, it is considered to be an integrated account holder. The management group has "five people from the city [refers to the municipality] and five people from Mehil€ ainen" (interviewee A). There are public managers from youth and family services, elderly and oral care, and culture and leisure time services. One of the public managers is from the central administration. From the private sector, there are managers responsible for general healthcare, oral healthcare and enhanced housing services. One private manager is from the business unit and one private manager is in charge of the welfare center. All representatives from the municipality and private sector act as account holders in the management group. In contrast, no one from the third sector is included in management group meetings, which means that the voluntary sector organization is not an account holder or account giver in this forum (Document 4). The account giving is formal in the management group, and conducts are explained and justified by using reports from the databases. The management group receives appropriate information through a virtual unit in the enterprise resource planning software that monitors the costs of the hybrid organization. Output and outcome measures are also taken from existing databases. The "information systems used in here [in welfare alliance] are ones used in the city" (Interviewee A). The hybrid also applies an open book principle, which enables confirmation of the production costs in the hybrid. The confirmation is produced in an inspection conducted by an independent operator.
Both public and private sector actors can pose questions in the management group. However, the judgment process that occurs in the management group raises questions because the municipality and the private company are judging their own performance. Indeed, public and private sector actors operate as account holders and givers at the same time, creating a biased horizontal accountability system. The literature on public-private AAAJ partnerships have addressed self-monitoring but not described how self-judgment in accountability systems occurs (e.g. Shaoul et al., 2012). The alliance model applied in the hybrid incorporates the idea that public and private sectors as partners share the duties of account givers and account holders, which raises the self-judgment on the center stage in the horizontal accountability system.
The self-judgment may work as the sanctioning and rewarding process is straightforward and "clear and easy [to understand]" because measures and performance standards are set in the contract (Interviewee A). The management group can impose sanctions and rewards based upon the numbers seen in cost, output and outcome measures. However, the management group can propose changes to cost and performance standards in the middle of the contract term in the event of exceptional or uncontrollable factors or unanticipated risk in the operating environment. Additionally, some conditions in the operating environment can change output and outcome goals in the middle of the contract term. These conditions are identified in the contract as risks and opportunities that may materialize. The purchaser also has the right to expand services in a hybrid, which changes the goals and performance measures used in the hybrid (Document 1). The ability to change the accountability system can generate a dynamic horizontal accountability system that changes constantly which inhibits the accountability if these changes make it difficult to determine who is accountable and for what and to whom.
What comes to sanctions in cases where the hybrid organization overspends, the costs are divided equally between the municipality and the company. The company pays these costs from its profit margins (Document 1). However, the private company can only pay the same amount of sanctions that it has received profits (Document 2). If the total costs of the hybrid organization are less than planned for in the budget, both the municipality and the private company receive each 47% of the savings as reward money while the rest of the savings are given by the steering group to personnel in the hybrid organization and/or to the third sector (Document 4).
When considering other goals, it became evident that the private company was the only one accountable for the output and outcomes shown in Table 3. As interviewee A noted: those goals (refers to output and outcome goals] are tied to bonus and sanction model. . . That is the difficult thing as the bonus, and sanction model only applies to private sector actor. Each output and outcome measure presented in Table 3 can provide rewards or sanctions to the company, depending on whether the measures show desired or undesired results. The scoring method applies different weights based on the various output and outcome measures. The amount of the reward or sanction given to the company is based on the sum of the weighted scores assigned to each output and outcome measure. However, the total amount of sanctions the company gets from the input, output and outcome goals cannot exceed the profit margins paid to it. Thus, there is a cap on the sanctions that the private company may receive (Documents 1, 2, and 4).
In conclusion, Bovens (2006) narrow accountability was not achieved in the hybrid organization for two reasons. The first of these reasons point to the fact that the judgment process was not unbiased because account holders and account givers were the same actors in all the accountability forums. However, this might not be a problem as the judgment process can be a straightforward process due to the clear interpretative scheme to the judgment. If citizens can confirm the judgment by examining the criteria on which the judgment was made, the potential for a scenario in which the judges are adversely judging their own cases will be minimized. The second issue hampering the narrow accountability was the inadequate reporting from the third sector. The third sector did not have a performance measurement system. This explains the inadequacy.

Horizontal accountability in a hybrid organization
Hybridity as an explanation for a biased and lopsided accountability design According to the results, the accountability system in the hybrid organization is imbalanced in terms of who is responsible for output and outcome goals. Eventually, only the private sector can face sanctions and rewards from the results of achieved output and outcome. The features of hybrid described in Figure 1 offer explanations for this imbalance.
From the perspective of ownership and goal incongruence (e.g. Thynne, 2011;Pache and Santos, 2013), the municipality wants to provide public value to citizens through output and outcome goals, whereas the company's main target is to produce shareholder value. As an example, the annual report of Mehil€ ainen emphasizes the importance of profits and states that "operating profit increased to 81.8 million euros in 2019" (Document 42: 37). By assigning the achievement of output and outcome goals to the private sector and penalizing the failures to achieve these goals, the public sector makes private-sector profits dependent on the public value produced. This creates a win-win situation or lose-lose situation when public value for citizens is not produced, and the profits of the private company are lost as a consequence. This arrangement attempts to tackle the problems related to goal incongruence and mixed ownership (e.g. Johanson and Vakkuri, 2017). In the below quotation, the head of the executive board of the alliance group explains how goal incongruence is dealt in the hybrid (document 43): "The measures developed together make sure all the participants are committed to the same goals. The alliance shares the profits and risks of the project and all decisions are made together, taking the customers' needs into consideration" When considering institutional logic and financing sources (e.g. Hodge and Greve, 2007), gaining experience in this kind of hybrid is desired by the private company for enhancing the possibilities of achieving larger future market shares (e.g. Porter, 2008). Mehil€ ainen (Document 43: 19) claims to be "the forerunner in public healthcare service" as it "is the market leader among the pilots' service providers and the number one choice for the Finns with over a 30% market share". The hybrid examined in this study is a pilot project where Mehil€ ainen operates as a pilot service provider. If passed in the Finnish Parliament, proposed social and health reform could generate these types of hybrids all over the country. The experience gained from the welfare alliance could be a competitive advantage in attempts to grow the market share. Because the hybrid offers an opportunity for profits while potential losses are minimized with the sanction cap, being responsible for output and outcome goals is a low-risk strategic opportunity to gain a competitive advantage for future market shares. By assigning output goals to the private sector, the public sector steers the private company to conduct particular market interventions (e.g. the treatment plans) on its behalf (see Table 3). In addition, the cost goals encourage the private sector to save tax money while producing public value.
The aspect of operational priorities (e.g. Billis, 2010) emphasizes that hybrid produces public services and these services cannot be left at the mercy of market forces as the public sector has to be able to fix dysfunctions in public service provision. Placing full responsibility for output goals to the private sector lends more control to the public sector as it is clear whose responsibility the goal attainment is when public services need fixing in the hybrid. As interviewee A noted: "those goals (refers to output and outcome goals] are tied to bonus and sanction model," which "only applies to private sector actor". Because market forces tend to punish poor services, this situation is also familiar modus operandi to the private company making it more acceptable (Porter, 2008), especially when the company cannot face losses due to the sanction cap. "In the contract period, the maximum sanctions cannot exceed the profit margins paid to the company" (Document 2: 22).
The fact that output and outcome goals were derived from strategic goals set by politicians elected in public elections introduces political and administrative control (see Willems and Van Dooren, 2011) to the private company's operations when the company is responsible for achieving these goals. As the company has negotiated a deal where it cannot face losses, it seems that market situation, shareholder pressure and corporate governance have counterbalanced the political and administrative control in the hybrid. To conclude, the imbalance in horizontal accountability demonstrates the negotiation power of the public sector as the monopsony market situation lends market power to the public sector. However, the private sector used its power to negotiate a contract that is rather low-risk in terms of losses. Mehil€ ainen had market power because it was "the only company submitting an offer" in the tendering (Document,44: 4).
Besides the private company's responsibilities for output and outcome goals, the role of the third sector also contributed to the imbalance seen in the horizontal accountability system in two ways. First, the third sector cannot face any sanctions from the company or municipality, although it may hamper the achievement of performance goals established for these two actors. For example, the third sector plays a major role in customer satisfaction as they run the customer information desk in the hybrid, but the company will bear poor results in customer satisfaction. Second, the third sector is not accountable at all if no savings occur in the hybrid or the steering group does not reward the voluntary sector for its efforts when the costs of service production are less than budgeted. Again, the third sector contributes to the positive or negative results but it does not receive any sanction from its contribution and it may not receive any reward. However, the imbalance in accountability relating under-and overspending is not that significant as the third sector does not use much money in the hybrid. From the perspective of hybrid characteristics, the operational priorities (e.g. Billis, 2010) explain why the third sector was not accountable. The third sector was recruited to the hybrid with separate negotiations securing that the mission of the third sector was in line with public services produced in the hybrid, and thus goal incongruence would not cause suboptimization.
Because creating win-win and lose-lose solutions in co-design sessions required that private and public sector actors became account holders and givers, the alliance model aiming for win-win solutions created the biased accountability system. In the alliance model, win-win and lose-lose solutions attempt to tackle the problems arising from financial sources, goal incongruence, governance, mixed ownership and different forms of social control (e.g. Johanson and Vakkuri, 2017); this is accomplished by simultaneously allowing budget savings, sales profits in the spirit of corporate governance and elected politicians' goals driving public value production for citizens. The adoption of the alliance model was driven by the possibility of making better interventions in markets and achieving a dominant market position, and the alliance model lead to biased accountability. According to document 43, the hybrid focuses on "measuring the effectiveness of the services." This indicates that better interventions in markets are important. To understand the motives of the private company, it is useful to know that "growth in public healthcare services in 2017" was achieved by Mehil€ ainen and this was a desired development (document 43). Thus, market share seems important to the company. Joining with the hybrid offers lessons on how to gain competitive advantages through better services. This can be seen from the following quotation: "Why did Mehil€ ainen join a risky project? The main reason was to learn to do services in a new way. Gaining experience from the alliance model will be useful in the future.
This statement demonstrates that features of the hybrid (e.g. Billis, 2010) incentivized the adoption of the alliance model, and these features related to the problems solved by public and private actors in the development phase of the hybrid organization.

Horizontal accountability in a hybrid organization
Welfare alliance versus contractual public-private partnershipwhat changes in accountability? In a typical public-private partnership, the public sector is the principal, and the private sector is the agent. As a principal, the public actor defines what is to be done, and the private actor executes it as the agent (Wu et al., 2016). After completion, the public actor in the role of principal judges the private actor's performance but does not judge its own activities because the public actor is not involved in task execution (Willems and Van Dooren, 2011). Compared to the public-private partnership model, the alliance model used in Tampere changed the accountability setting by introducing shared responsibilities in planning and production (e.g. appendix 3). As the examined case organization demonstrated, public and private sector actors define what is to be done, and public and private sector actors execute the task. Thus, both public and private sectors have dual roles; they both operate as principals and agents, and they both judge their own cases, unlike public-private partnership in which the public actor is the principal who judges the private actor operating as an agent (e.g. Shaoul et al., 2012). In the alliance model, public and private sector actors are principal-agent hybrid actors. As noted in document 44: The alliance is not contracting out or privatization. An alliance is like marriage. . . The Alliance has a common budget. The parties also share the benefits and risks. . . and produce services together The alliance model includes a specific type of power imbalance that is not present in typical public-private partnerships. In the public-private partnership, task execution is conducted by the private sector, and the public sector has no part in the production (Hodge and Greve, 2007). Here, the private actor cannot argue that it was the public providers whose actions led to failure (Forrer et al., 2010). In contrast, in the alliance model, the company's performance depends on the performance of the public sector provider, and the company needs the public sector provider to play its part if operational goals are to be achieved (see Table 3). However, the company cannot impose any sanctions on the public sector if they fail to perform, and the company also loses its profits, even though the company did everything as expected (see Tables 5 and 6). This is a power imbalance related to the ability to sanction the production partner resulting in poor performance. In the alliance, "the public sector can sanction the private partner if the private provider is not performing to the required level" (Document 44), and there are no provisions in the contract allowing the private partner to negotiate new conditions due to changes in the operating environment.
Because there is no shared production between public and private actors in the traditional public-private partnership (Hodge and Greve, 2007), the imbalance between public and private sector service providers seen in the alliance model cannot occur. In the traditional public-private partnership, the power imbalance is of a different type, the type in which the principal can impose sanctions if the effort fails due to the poor performance of the private actor (Forrer et al., 2010). However, the private actor cannot sanction the principal if the public actor in the role of the purchaser caused the failure and damage to the company's public reputation by ordering services with incomplete service specifications (Willems and Van Dooren, 2011). Understanding the differences between the alliance model and public-private partnership model can help us understand the imbalanced, balanced, dynamic and static types of accountability described in Figure 2.

Conclusion and concluding remarks
This research focused on the question of what type of horizontal accountability was designed within a hybrid organization that utilized alliance model. By using a case study and descriptive analysis, the study shows that the private company had more accountability obligations than the municipality while the third sector organization had the least AAAJ obligations. In fact, the third sector organization could only be rewarded, and even this reward was optional and based on the discretion of the steering group responsible for managing the hybrid organization. The private company was accountable for the outcome and output goals. The public and private sectors together were accountable for the input goals. The studied accountability system is not in balance because it selects the private sector as the primary account giver while assigning fewer accountability obligations to public sector service providers and mostly relieving the third sector from accountability obligations. This result demonstrated a power imbalance relating the rights to sanction different service units and supports the previous literature that identified several possible and actual problems in assuring accountability in hybrids (see Sinclair, 1995). Therefore, this study proposes a balanced and imbalanced accountability system model for hybrids (Figure 2). This model describes a situation where horizontal accountability is lopsided because some actors are more accountable than others. Similarly, not all actors are equally important account holders in the hybrid.
In a way, the finding concerning the imbalance is odd, considering the concept of shared risks embedded in the alliance model and the fact that public, private, and third sector actors are all accustomed to horizontal accountability in their systems (e.g. Williams and Taylor, 2013). Although failures of the municipal service providers may cause the private company to lose its rewards, the private company cannot impose sanctions to the public sector providers. However, understanding the features of the hybrid relating ownership, operational priorities, human resources, financing sources, institutional logics, goal incongruence and social forms of accountability offers multiple explanations why this imbalance is a preferred state in the case study organization. For example, profit logic motivates the private company to accept the accountability system, and the public value logic drives the municipality toward this type of imbalance to the extent that the municipality has enough tools to steer the hybrid to the production of public value. The imbalance between the third sector and the other sectors resulted from mission alignment between the public and third sectors. The public sector did not have to fear that the third sector would deteriorate public value.
The uneven distribution of accountability responsibilities found in the hybrid organization raises questions about how different actors should be accountable to each other in horizontal relationships within the hybrid organization. The past literature has not

StaƟc accountability
Imbalanced and staƟc accountability: Accountability system is fixed for the whole contract term and only one principal-agent hybrid actor can set goals and pose sancƟons and rewards to the other principal-agent hybrid actor who can set goals but cannot sancƟon and reward the other principal-agent hybrid actor Balanced and staƟc accountability: Accountability system is fixed for the whole contract term and both principalagent hybrid actors can set goals and pose sancƟons and rewards to the other principal-agent hybrid actor Balanced and dynamic accountability: Accountability system can change during the contract term and both principalagent hybrid actors can set goals and pose sancƟons and rewards to the other principal-agent hybrid actor Imbalanced and dynamic accountability: Accountability system can change during the contract term and only one principalagent hybrid actor can set goals and pose sancƟons and rewards to the other principal-agent hybrid actor who can set goals but cannot sancƟon and reward the other principal-agent hybrid actor Figure 2. Examples of dynamic and imbalanced accountability in biased two-actor hybrids Horizontal accountability in a hybrid organization focused on horizontal accountability in hybrid organizations that have public, private and third sector actors providing a wide range of public services (e.g. Luke, 2010;Willem and Van Dooren, 2011). This means that questions arising from our results cannot be answered with the results of past studies. To encourage future research in response to the questions raised, the imbalanced accountability systems were suggested as a concept.
We also found that proper accountability can be jeopardized because of the accountability system. In all forums, there were double roles as account holders, and account givers were the same actors. These double roles make accountability problematic in the hybrid organization as actors are justifying, rewarding and sanctioning their own conduct (e.g. Hodge and Greve, 2007). This finding implies that there is a need to develop theoretical models describing the formation and functioning of biased horizontal accountability models in hybrids. The word bias refers to judges judging their cases in horizontal accountability systems. The publicprivate partnership literature talks about self-monitoring (Shaoul et al., 2012), but current theories of dual accountability roles in hybrids are in a nascent state. By proposing a biased horizontal accountability system, we sought to challenge the notion of accountability systems that is in line with the traditional definitions of accountability (e.g. Bovens, 2006). As a concept, the biased accountability system does two things: (1) It criticizes previous conceptual frameworks of accountability that do not consider dysfunctional accountability systems.
(2) It highlights in the hybrid context the relevance of the following of old Latin phrase used in legal doctrine: Nemo judex in causa sua (no one is the judge in his own cause).
In the hybrid, an attempt was to prevent or mitigate the problems relating to double roles by using clear standards of performance that were set beforehand. Clear standards are a good practice in hybrids (Forrer et al., 2010). However, the accountability system included several scenarios in which risks and opportunities materializing the operational environment open up possibilities to refute the performance criteria, allowing new performance standards to be adopted. This indicates that we should have dynamic horizontal accountability models that examine how horizontal accountability systems are dismantled and generated again as a part of risk management procedures in a hybrid context. The possible dynamic nature of accountability in hybrids emphasizes what Grossi and Thomasson (2015) stated when they argued that the way in which the accountability system is used affects the ability to bridge the accountability gap existing in hybrids. If the accountability system is dismantled all the time, the accountability gap will probably increase as it becomes uncertain who is accountable and for what in hybrid activities. Moreover, setting clear criteria for performance as a good practice suggested by Forrer et al. (2010) may not be sufficient if there is a possibility of negotiating new performance criteria whenever circumstances change. Dynamic accountability models attempt to challenge the theoretical notion that the system is stable because accountability is otherwise challenging to attain according to blameavoidance literature showing how changing organizational structures complicates accountability (e.g. Hood, 2011). As accountability in hybrid organizations is an under-examined topic (e.g. Hopwood, 1996;Caperchione et al., 2017), scholars have called for new accountability solutions for hybrids (Deleon, 1998). The alliance model used to construct the hybrid sought to tackle the problems arising from sub-optimization by creating a new version from already used accountability systems in alliances between public and private sector actors. Unfortunately, we cannot answer how this type of accountability system works in practice because this study concentrated only on design solutions of such a system, and the subject hybrid has just started its operations. Thus, we recommend that future studies investigate how this type of horizontal accountability system works.

AAAJ
This case study increases the current knowledge of accountability in cross-sectoral alliances between public, private and third sector actors. Currently, there are very few studies on public-private partnership alliances that cross administrative sectors (e.g. Tomkins, 2001;H akansson and Lind, 2004;Neu et al., 2010). However, there are even fewer on cross-sectoral alliances incorporating actors from the public, private and third sectors. Unfortunately, previous research has focused primarily on contracting out, government-owned enterprises and public-private-partnerships while ignoring the hybrids incorporating all three sectors (e.g. Hansson and Longva, 2014). The void in the research makes the examination of the case organization and its design process valuable because knowledge on the functions of crosssectoral alliances providing welfare services is lacking. The studied hybrid accountability system, the features of the hybrid driving such system, and the design process of the system all support the studies stating that accountability systems can be complex in hybrids and, therefore, they can erode accountability (Willem and Van Dooren, 2011). Because of the hybridity and complexity, the development of examined accountability system took a long time and required a lot of efforts.
As a methodological contribution, this research has shown one way to study horizontal accountability in hybrids. However, the most prominent limitation of this study is that a crosssectional case study offers very little basis for generalization of the results. Thus, the safe way to use the findings of this study is to apply them only to initial development of theory. However, in applying the results of this study to other studies, one can see that certain patterns do emerge. For example, the problematic nature of the accountability and the existence of accountability gaps in hybrids was confirmed in this study and in other studies (e.g. Flinders, 2005).
Overall, this study raised several interesting questions for future research. First, how does public sector actors' accountability system change when they become a part of the hybrid organization. It would be interesting to map out all the changes in accountability induced by the hybrid constellations. Second, is an imbalance in accountability typical of hybrids involving actors from the public, private and third sectors, and services from multiple administrative sectors? Third, what factors in hybrid organizations typically encourage horizontal accountability systems that unevenly distribute accountability responsibilities? Finally, we need more research on why hybrids create biased horizontal accountability systems, what type of hybrids design those types of biased systems, and how account holders that are judging their own conduct operate in hybrid settings. All the above questions open up interesting research opportunities that hopefully will inspire new studies on the important topic of hybrid accountability.