The role of the World Bank and the International Monetary Fund in the healthcare financing reforms in Croatia: Transfer of ideas and limited coercion

The paper investigates the influence of policy ideas from the World Bank and the International Monetary Fund (IMF) on healthcare financing policy in Croatia during the 2002 reform. It contributes to the global social policy literature by providing evidence that the influence of international organisations primarily stems from non-coercive instruments to control the policy agenda, for example, dissemination of ideas, technical assistance and consultations with the recipient government. Even though Croatia was facing economic and political difficulties which weakened its bargaining position vis a vis IOs, the paper shows that impact of coercion and conditionalities attached to international aid was limited. It explains the lenient stance of international organisations by their mission to aid and adjust to a country’s needs as well as their self-interest to lend money, to stay in the reform game and to prolong their influence in the future. Consequently, international organisations are willing to bargain and make trade-offs with the recipient government.


Introduction
After the collapse of the communist regimes in Central East Europe (CEE), countries in the region faced major challenges due to political, economic and social transition. Post-Yugoslav countries were under particular strain because of the additional task of nation building (Ramet, 2010). Furthermore, some post-Yugoslav countries like Croatia suffered severe damage during the 1991-1995 war which followed the disintegration of Yugoslavia. This made Croatia heavily reliant on international aid and consequently susceptible to international organisations' (IOs) influence.
The existing literature suggests that IOs such as the World Bank (WB), the World Health Organization (WHO), the International Monetary Fund (IMF) or the European Union (EU) were heavily involved in reforming the different sectors of social policy in post-communist CEE, including healthcare (Cerami, 2006;Deacon et al., 1997;Grabbe, 2006;Grielen et al., 2000;Kaasch, 2015). The WB and the IMF appear as the most influential organisations to have shaped social policy in postcommunist countries (Deacon et al., 1997). They seized the opportunity to offer policy advice, technical expertise and money (Cerami, 2006;Deacon et al., 1997;Kaasch, 2015). Regarding healthcare, the goal of the WB and the IMF was to improve the efficiency, productivity and financial sustainability of the healthcare systems in CEE (De Beyer et al., 2000). The same is observed in Croatia where the WB and the IMF were the most important IOs involved in the transition reforms and post-war reconstruction effort in the aftermath of the 1991-1995 war (Stubbs and Zrinščak, 2007).
Healthcare was one of the sectors severely affected by transition processes and war. The financial problems in healthcare were temporarily addressed by wide-ranging reforms in 1993, but resurfaced in the late 1990s because of low healthcare insurance contribution levels, increased unemployment, a large informal economy, population aging and large general budget deficits. In the early 2000s, major reforms were initiated, focusing on the three healthcare dimensions: financing, provision and regulation (Rothgang et al., 2010). The reforms were prepared and coordinated together with the IMF and the WB. However, some of the policies introduced differed from the IOs' recommendations. The paper focuses only on the financing dimension in the [2000][2001][2002] period since this provides a clear case of contestation between domestic and international policy actors. The paper argues that the WB and IMF chose the approach of adjustment rather than trying to unilaterally shape Croatian healthcare financing. It contributes to the literature on healthcare financing policy changes and the interaction of the WB and the IMF with the national governments in CEE.
The paper will answer the following research questions: (1) To what extent and in which ways did the WB and the IMF influence Croatian healthcare financing policies? and (2) Why did the WB not use the leverage it had to compel Croatia to implement its policy prescriptions, despite it being in a weaker bargaining position at the time?
The paper reports on a case study that produced a rich account of the WB/IMF role in Croatia, a country which was facing major challenges posed by post-communist transition and 1991-1995 war. The case study is based on qualitative data, gained from combining in-depth semi-structured interviews with documentary analysis, which allows for a detailed insight into the reform process and actors involved. The data (including 32 media publications, six legal acts, minutes of 54 parliamentary sittings, three government strategies and 48 IO documents such as project and loan reports, evaluation reports, publications, letters of intent, loan agreements, and studies related to Croatia) were collected from online archives of the WB and the IMF, and both physical and online archives of the Croatian government and parliament, Croatian national library, Croatian national bank and media outlets. The data cover the period before, during and after the 2002 reform, from 1997 to 2006.
Fourteen interviews were conducted with experts knowledgeable of the subject and those who were directly involved in the reform process, such as ministers and their assistants, healthcare administration staff from the institutions relevant to policy (Ministry of Health, Croatian Institute for Health Insurance [CIHI] and Croatian Institute for Public Health), politicians involved in the healthcare system as well as academics and IO experts. Interviews were arranged through personal contacts and a snowball approach in which the respondents introduced the author to other potential respondents, up to the point where additional data revealed no new information. The data were triangulated using multiple sources to enhance the credibility of findings.
The data were analysed using qualitative content analysis and MAXQDA software. First, the entire data collection was carefully read. When documents included broader issues rather than healthcare alone, the keywords health and health financing were used to search for text segments relevant to the research topic. In the ensuing step the data were coded by focusing on: ideas and position of IOs and domestic policy actors about healthcare financing policies; role of IOs in the healthcare financing sector (advice, education, financing, imposition); position of domestic policy actors towards IOs (co-operation, contention).
The next section offers a literature review on the instruments the WB and IMF use to influence policymaking. The subsequent section analyses economic and political difficulties, as well as problems in healthcare system performance Croatia was facing at the time of healthcare financing reforms that is, its political and economic context and a short overview of Croatian healthcare financing policies. The fourth section focuses on the interaction between the IMF, the WB and the domestic actors and analyses how Croatia could introduce policies that were not completely aligned with the IOs' recommendations. Finally, a discussion and conclusions of the study are presented.

The World Bank and the IMF: coercive and non-coercive influence
Two different approaches to influencing policy in countries where the WB and the IMF offer aid can be deducted from the literature: a coercive and a non-coercive approach. Much of the literature stresses the importance of policy lending conditionalities, whereby the WB and the IMF act as coercive actors who try to unilaterally introduce policies they advocate (Cerami, 2006;Deacon, 2007;Deacon et al., 1997Paloni and Zanardi, 2006). Thus, coercion represents the first dimension of power (see Dahl, 1957;Lukes, 2005), forcing countries to do what they otherwise would not do. The IMF conditions focus on macroeconomic issues (Kaasch, 2015;Odling-Smee, 2006), while the WB also includes 'detailed microeconomic conditionality' with direct implications for other sectors such as healthcare (Deacon, 2007;Paloni and Zanardi, 2006: 2). Nevertheless, the IMF's policy prescriptions can indirectly influence healthcare policies or 'at least have implications for choices and decisions about health policy' (Kaasch, 2015: 56) since the healthcare sector constitutes a large part of government budget expenditure.
Some studies argue that adjustment loans and the related conditionalities have not always been effective in securing desired policy change (Killick, 1998;Larmour, 2002;Mosley et al., 1995). Weyland (2006) argues that the WB was not particularly concerned with countries fulfilling conditionalities while Noy (2017) notes that conditionalities are 'often purposefully vague ' (p. 13). Moreover, non-compliance with conditionalities frequently went unpunished due to the systemic pressures within the WB to continue lending money (Killick, 1998;Mosley et al., 1995). Such pressure was internalised by setting country lending targets, staff promotion based on negotiating new loans, guaranteed repayment of loans by aid receiving countries and incentives to secure the future relationship with aid recipient countries (Killick, 1998;Larmour, 2002;Mosley et al., 1995). The WB itself recognised these problems and stated that conditionalities can be difficult to enforce (WB, 1998(WB, , 2005. Due to conditionality limitations and the inability to exercise formal veto power (Orenstein, 2008), the WB and the IMF often choose to act in a non-coercive way (Deacon, 2007;Paloni and Zanardi, 2006: 19;Yılmaz, 2017). Non-coercion entails technical assistance, survey missions, training institutes and policy advice disseminated through research publications and consultations with the governments (Bazbauers, 2018: 40). By persuading countries to follow their advice, IOs act as an epistemic community, which has been defined as a network of experts with an 'authoritative claim to policy-relevant knowledge' (Haas, 1992: 3). Positioned as 'The Knowledge Bank', the WB appears as the legitimate source of policy expertise which is based on accumulated cross-country evidence and its project experience in other countries (Bazbauers, 2018: 239;Paloni and Zanardi, 2006;Stone and Wright, 2007;Yılmaz, 2017: 121). Expertise frames policy discourse in recipient countries, moulds the knowledge of domestic actors and biases their choices (Deacon et al., 2007: 226). St. Clair (2006) argues that WB's framing of policy is influenced by 'hegemonic consensus' which is not independent from political and social set of values (see, for instance, Appel and Orenstein, 2018;Orenstein, 2008;Plehwe, 2007 which describe pervasiveness of neoliberal policy framing within the WB). Thus, non-coercion can heavily influence the agenda-setting process by including or excluding policy issues and solutions (Bazbauers, 2018;Larmour, 2002;Noy, 2018;Orenstein, 2008;Weyland, 2006). Such influence can be situated in the second and third dimension of power (see Lukes, 2005).
However, coercion and non-coercion can be regarded as an ideal type heuristic. In reality, the distinction is blurred as elements of non-coercion can be present in coercion. Conditionalities can have a partially non-coercive character. Such 'pro forma conditionalities' are consensual, mutually agreed between the lender and recipient and emerge as the 'outcome of often long periods of consultation and discussion' (Killick, 1998: 9). The point of the latter is to instil government identification with and ownership of the lending programme, and to increase the success of loan programmes (Bazbauers, 2018;Killick, 1998;WB, 1997cWB, , 1998. The WB has argued that this leaves a larger space for country-grown policies while following the donor's minimum standards (WB, 2005). In contrast to 'hard-core conditionalities', coercive influence is effectively reduced because 'pro forma conditionalities' are not 'made only at the insistence of the lender' (Killick, 1998: 11).
Regarding healthcare policy, scholars suggest the IMF and the WB promoted policies associated with the Washington consensus e.g. increasing individual responsibility, user charges for health services, marketisation, reduction of healthcare expenditures and healthcare benefits and the like (Cerami, 2006;Deacon, 2007;Laurell and Arellano, 1996;Stuckler and Basu, 2009: 771). Since the 1997 Health, Nutrition and Population Sector Strategy, the WB has advocated for social sensitivity in health, a greater role for the state, increased flexibility and avoidance of rigid policy prescriptions (Deacon, 2007;Kaasch, 2015;WB, 1997c). In general, the discourse shifted towards universal healthcare, expanded coverage, equity and increased access, especially for the poor (Kaasch, 2015;Noy, 2017). Unlike pension reform, for which the WB advocated the three-pillar pension system (Orenstein, 2008), the WB did not have a readily available blueprint for healthcare financing reforms (Radin, 2008). Rather, it tailored its assistance to different political, economic and institutional environments (Noy, 2017(Noy, , 2018Yılmaz, 2017: 124). Moreover, the policy prescriptions varied due to the different WB staff across countries and time (Noy 2017: 13).
Theoretically, the literature states that coercive or non-coercive influence is more powerful in countries which are less developed, where the state and the economy are weaker, or find themselves in a crisis (Deacon et al., 2007: 226;Noy, 2017;Woods, 2006). Such countries have less bargaining power vis a vis IOs (Batley, 2004: 55;Laurell and Arellano, 1996: 13;Woods, 2006: 72) and often had no choice but to accept conditionalities in order to access new sources of funding (Paloni and Zanardi, 2006: 3). Moreover, they sometimes lack the policy-relevant knowledge and rely on WB/IMF expertise (Paloni and Zanardi, 2006: 20;Weyland, 2006). Such expertise is not neutral and often translates into exercising non-coercive power (Bazbauers, 2018;St. Clair, 2006). However, even countries which faced economic or political difficulties were able to resist the pressures from powerful IOs. Noy (2017Noy ( : 173, 2018 argues that in Peru, Argentina and Costa Rica, the WB aligned its health strategy to the recipient countries and 'worked with governments to support their initiated reforms'. Weyland (2006) comes to very similar conclusions in other Latin American countries, such as El Salvador, Bolivia, Brazil or Costa Rica. Wireko and Béland (2017) show that the WB was willing to accommodate to the government policy position towards Social Health Insurance (SHI) in Ghana, and even promote it internationally. Although the influence of IOs can be expected to be greater in countries which find themselves in unfavourable position, the mentioned literature suggests this is not always the case. This paper provides a case study on Croatia which challenges the view of the IMF and the WB as rigid and hegemonic organisations. It shows that the IMF and particularly the WB did not unilaterally shape healthcare financing reforms in Croatia which was in a politically and economically challenging position and was still able to contest some of their policy prescriptions.

Political and economic context
The early 2000s marked a major point of discontinuity in Croatian politics (Bičanić and Franičević, 2003). The strongest party since Croatian independence, the Croatian Democratic Union (CDU), was severely weakened when the most influential figure in the party and the Croatian president, Franjo Tuđman, passed away in 1999 (Stubbs and Zrinščak, 2007). The elections organised in 2000 were won by a coalition of six parties spearheaded by the Social Democratic Party (SDP) and the Croatian Social Liberal Party (CSLP) (Bičanić and Franičević, 2003: 24). The weakening of the CDU and the new SDP/CSLP government's support for change from a semi-presidential to a parliamentary system resulted in more open and deliberative policy processes. The new reform-oriented government and 'its explicit stance of openness to all forms of international cooperation, was quickly seized upon by the World Bank and others' (Stubbs, 2008: 370). To further increase its openness to the West, the new government started negotiations to join the EU and North Atlantic Treaty Organization (NATO), joined the World Trade Organization (WTO) and Central European Free Trade Agreement (CEFTA) and cooperated with the International Criminal Tribunal for the former Yugoslavia (Bičanić and Franičević, 2003).
However, the coalition between ideologically diverse parties suffered from 'internal struggles, blackmailing, and deliberate politicization of particular reform policies and moves' (Bičanić and Franičević, 2003: 24). In addition, the government was heavily criticised by the media and public opinion for pushing 'neoliberal reforms' under the tutelage of the IMF and the WB (Bičanić and Franičević, 2003) and for willingness to cooperate with the International Criminal Tribunal for the former Yugoslavia (Kasapović, 2005). All of this led to the destabilisation of the government, frequent changes of ministers (including the Minister of Health) and eventually the exit of two parties (most importantly CSLP) from the coalition government in 2002 (Kasapović, 2005). Policymaking under such conditions was 'difficult, slow, and often without an easily recognizable pattern, but with a serious negative public relations effect' (Bičanić and Franičević, 2003: 24).
Moreover, Croatia was still recovering from the devastating 1991-1995 war which disrupted trade flows, tourism and foreign investments (WB, 2001a), and slowed the development of the economy. 1 For instance, gross domestic product (GDP) only reached pre-war 1990 growth levels in 2003 (Bebek and Santini, 2013). Also, the war and postcommunist transition influenced social policymaking, including healthcare. Disempowered groups such as war veterans and to an extent pensioners successfully claimed social benefits, justified by the unfairness of privatisation processes (Puljiz et al., 2008: 39;Stubbs and Zrinščak, 2007: 91). Although the economy grew in the 1995-1998 period (WB, 2001a), Croatia still had to tackle excessive trade deficits, rising unemployment, insolvencies of banks and state-owned enterprises, and excessive public sector employment (Bičanić and Franičević, 2003: 18;International Monetary Fund (IMF), 2001: 4;WB, 2001d). At the same time, social policy reforms continued, including healthcare, pensions, social protection and assistance among others (Stubbs, 2008;Stubbs and Zrinščak, 2007). To stabilise the macroeconomic situation, curb large fiscal deficits, rising public debt, and reform its social sector, the Croatian government turned to the World Bank and the IMF.

The Croatian healthcare system in the 1990s
Croatia inherited a heavily decentralised system that had been established when it was part of Yugoslavia. It produced unfavourable outcomes and could neither control expenditure nor collect adequate revenue to finance healthcare services (Chen and Mastilica, 1998;Džakula et al., 2012;Rodwin and Šarić, 1993). The 1993 healthcare reforms centralised healthcare financing. Payroll taxes were pooled into one national health fund, the CIHI. Co-payments for selected services and two forms of private insurance were introduced. One form was full private insurance, which persons above a certain income threshold could opt for instead of the SHI offered by the CIHI. The second form was supplementary insurance, which offered better access and quality of healthcare. In addition to SHI, the government's budget financed care for vulnerable groups and bailed out CIHI deficits when necessary. The reforms curbed healthcare expenditure and pooled additional funds in the healthcare system (European Observatory on Health Care Systems, Regional Office for Europe, World Health Organization, 1999: 46).
However, in the late 1990s, problems of rising expenditure and lack of funding started to re-emerge. High expenditure was related to inadequate primary care, overreliance on hospital and specialist care, generous benefits and extensive exemptions from co-payments which generated moral hazard (Vončina et al., 2007;WB, 2004). Payroll taxes did not provide adequate funding due to the ever-growing older population (about 20% of the population was older than 60 years in 2003), increased unemployment and a large informal sector (roughly 7% of GDP) (Vončina et al., 2007). All of this created financial deficits in the CIHI from 1998 until 2001. In 2000, total healthcare expenditure amounted to 10.2% of GDP (Croatian Parliament, 2006) and the CIHI debts were HRK 4.2 billion (about EUR 0.55 billion) (Zrinščak, 2007). To sustain its operations, the CIHI required sizable financial infusions from the government budget, which was already under severe strain (WB, 2001b). Taken together, these problems created an urgency for reform.

Reforms introduced in the early 2000s
In the early 2000s, full private insurance was abolished, which resulted in pooling more insurers into the SHI. The CIHI was incorporated into the government budget, losing its extra-budgetary status and the limited autonomy it previously had. The idea was to achieve more control over expenditures, debt collection and management (Vončina et al., 2007: 150). To reduce reliance on payroll taxes, reforms focused on diversifying revenue collection, which included lowering the payroll contribution rate from 18% to 15.5% (WB, 2004: 40), increasing the scope and rates of co-payments (Vončina et al., 2007) and reducing exemptions from co-payments from 80% to 50% of the population (WB, 2004). Moreover, voluntary complementary health insurance (CHI) was introduced, which covered the above-mentioned co-payments. The premium was set as a lump sum to be paid monthly, HRK 80 and HRK 50 for retired persons (about EUR 10.5 and EUR 6.5).
However, the government in principle subsidised CHI premiums for large parts of the population such as children under 18, war veterans or people with disabilities (Vončina et al., 2007). The CIHI was given the exclusive right to initially offer CHI while private insurers could only enter the market after 2 years.
Throughout the reform process, the Croatian government was co-operating with the WB in the healthcare sector (and to a lesser extent with the IMF). However, insufficient co-payment exemptions and the introduction of CHI clashed with the WB recommendations. Although the WB was initially not in favour of such policies, its adjustment and the prescription of pro forma conditionalities enabled Croatia significant room for manoeuvre in ignoring the WB's recommendations. The next section analyses the involvement of the WB and IMF in Croatian healthcare policy, including their advice, conditionalities, interaction with the Croatian government and the reasons for the WB's adjustment towards the Croatian government.

Non-coercion: advancing the WB agenda
Croatia's co-operation with the IMF and the WB started during the 1990s, after Croatia joined the IMF in 1992 and the WB in 1993. Both organisations offered economic stabilisation programmes, loans, technical assistance and expertise (Croatian National Bank, 2021; WB, 2021 Interview 1, 2020 2 ). The IMF was focused on macroeconomic policies to reduce inflation and stabilise the currency while the WB focused on sectoral policies, particularly pensions and health (Interview 1, 2020; Stubbs and Zrinščak, 2007). This engagement enabled the WB and the IMF to occupy a seat at the government table and gain access to the most important policy actors and institutions to promote their policy ideas.
The co-operation continued when the new SDP/CSLP government came to power in 2000. The government requested support from the WB and the IMF to resolve the economic crisis and emerging issues in social policy, including healthcare (Stubbs and Zrinščak, 2007). However, the IMF was not particularly involved in the healthcare discussions even though it negotiated a Stand-by Arrangement (SBA) with the government in 2001. The IMF SBA was focused on reducing public sector employment and government deficits, privatising state-owned enterprises, and establishing price and currency stability (Croatian Government 2001Government , 2002IMF, 2002b: 21-22). IMF Staff Country Reports (IMF, 2000(IMF, , 2002a stated that healthcare benefits were generous and that expenditures were unsustainable and should be reduced. However, the IMF did not promote specific healthcare policies. Instead, the WB took the lead in healthcare policy dialogue with the government (IMF, 2002a: 38). The WB experts involved in the process stated that: As far as health, pension and social policies are concerned, we [the WB] would mostly lead a dialogue . . . and the IMF in principle then takes over our reform proposals and has to align with us . . . (Interview 2, 2020  The WB had already been engaged in the healthcare sector during the tenure of the CDU government through consultations, producing healthcare studies and the conclusion of the Health Project in 1999 (Zrinščak, 2007). The WB continued the same practice with the SDP/CSLP government. The WB experts participated in government meetings (Interview 1, 2020) and positioned themselves as part of the epistemic community (Haas, 1992) which exerted non-coercive influence on domestic policymakers. Compared to Croatian pension reform, where the WB promoted a Chilean model and eventually settled for Argentinean one (Stubbs and Zrinščak, 2007), the WB did not rely on any specific reform model regarding healthcare financing. Rather, it tailored its assistance to the country-specific context. Policy ideas were based on WB studies which contained evidence-based analyses of the Croatian healthcare system, and also on the WB's long-term experience of working in other countries (Interview 1, 2020; Interview 2, 2020; Interview 3, 4 2020).
In 'The Reform of Health Care in Croatia' document, the WB stated that healthcare financing policy was inadequate and that 'changes in revenue sources and revenue collection efficiency are needed' (Croatian Ministry of Health and World Bank, 2000: 1). Moreover, in 'Public Sector Financing, Health Care Reform and Pension Reform in Croatia' it stated that public expenditures were too high, and argued in favour of introducing cost-containment policies and policies that would moderate demand (WB, 1997b). These WB studies, alongside others such as 'Croatia Beyond Stabilization' (WB, 1997a), 'Country Assistance Strategy' (WB, 1999), 'Croatia: A Policy Agenda for Reform and Growth' (WB, 2000) and 'Regaining Fiscal Sustainability' (WB, 2001b) recommended diversifying revenue collection and decreasing the payroll tax, enhancing debt and payroll contribution collection through increased government control over the health fund, increasing the rate and scope of co-payments for a range of services, exempting only vulnerable groups such as poor or chronic patients from co-payments, explicitly defining a publicly-financed basket of health services, reducing the overly generous sick pay and maternity leave and, finally, supporting the development of a private insurance market.
The WB's presentation of ideas and its direct access to the government and relevant policy actors proved to be very influential in educating the domestic policymakers and shaping the way in which they thought about healthcare policy. The WB was seen as a legitimate source of expertise and at the same time Croatia lacked healthcare policy experts in the fields of health economics and health system management (Interview 4, 5 2020; Interview 5, 6 2019; Interview 6, 7 2019; Croatian Ministry of Health and World Bank, 2000). The following quotes confirm the latter.
[The WB] actually had an extremely beneficial impact because they brought in expertise, consultants, who steered the healthcare policy in a direction which, at that time, was considered to be the right one. (Interview 7, 8 2020) . . . We [the WB] could master a critical body of researchers, top experts, experienced people to go through this process and create a report which we discussed with our counterparts. This process itself is very influential . . . (Interview 1, 2020) Through such analytical studies, governments simply get the opportunity to look at the same system through a different focus, we usually do it with comparisons of good practices in the world, which then gives them [the domestic policymakers] a different way of thinking and sometimes the studies alone are enough to incentivise the government to make policy changes. (Interview 2, 2020) At this point, the government did not contest the WB advice and relied on the WB expertise to decide which were the best policies to consider and implement. Arguably, this reflects the notion of the third dimension of power (Lukes, 2005) as the WB was able to influence the government's agenda and control issue attention in healthcare without contestation. The government's healthcare strategies adopted in 2000 and 2002 largely reflected WB ideas, emphasising the need to reduce healthcare expenditure through stricter financial discipline, introduce a standard basket of services covered by SHI, increase the use of co-payments, reduce exemptions to co-payments and support the development of a private insurance market (Croatian Government, 2000: 30;Croatian Ministry of Health, 2003: 36;Croatian Ministry of Health and World Bank, 2000;Croatian Office for Development Strategy, 2002).

Coercive influence: a deliberately blunt knife
To support the government agenda, the WB provided financing and included healthcare policy components into the 2001 Structural Adjustment Loan which focused on structural reforms, supporting economic growth, market competition, enhancing flexibility in the labour market, reducing investment barriers, strengthening social protection and improving the health and pension systems (WB, 2001d). The WB's non-coercive influence was complemented by coerciveness when the WB demanded prior fulfilment of policy actions in order to release the loan. 9 To disburse the loan, the WB insisted on increasing the scope and rate of co-payments, exempting only vulnerable groups such as poor or chronic patients from co-payments, and levying contributions on income from non-wage labour.
However, such a co-payment policy would have greatly eroded the generous benefits people were already entitled to. It was heavily contested by the opposition in parliament, trade unions, media and the public which criticised the proposed reforms as being unfair, shifting the burden of financing from the state to the citizens and blindly following the 'neoliberal' logic of the IMF and the WB (Đuretek et al., 2001;Rebić, 2002;Interview 1, 2020;Večernji list, 2001Večernji list, , 2003. At this point, the interests of the WB and the government started to diverge. Although the government was initially in favour of the WB policies as outlined in its agenda, once the backlash from the opposition started, it gave in to this pressure. Already unpopular and suffering from internal disputes, the government adjusted its standpoint in order to appease the public, prevent social backlash and increase its prospects in future elections. To mitigate the domestic pressure and satisfy the WB at the same time, the government remained open to increasing the rate and scope of co-payments. However, it proposed much broader co-payment exemption policy compared to the one favoured by the WB. Among others, groups eligible for co-payment exemptions included war veterans, unemployed, disabled, frequent blood donors etc. (WB, 2004). Moreover, the government proposed voluntary complementary health insurance (CHI), which would be offered exclusively by the CIHI for the duration of 2 years. CHI premiums would be set as a fixed lump sum with an additional discount for pensioners (Croatian Parliament, 2002;Vončina et al., 2007). The idea was 'to reduce the adverse effects on financial protection' (Vončina et al., 2010: 228) and to increase the CIHI's revenue (Interview 8, 10 2019;Interview 9, 11 2019;WB, 2004). A member of the government who worked in the Ministry of Health at the time stated, IOs asked for the introduction of co-payments, we [the government] then introduced the reform in which the citizens themselves would not directly pay out of pocket, but would rather be insured through the CHI, thus exempting them from co-payments. (Interview 10, 2019) The WB (2004) argued that the role of co-payments was negligible in moderating demand for health services and that broader co-payment exemptions and CHI would reintroduce the moral hazard problem (p. xi). Moreover, the WB (2004) stated that CHI would introduce the problem of adverse selection because 'it is likely to be purchased by beneficiaries with highest medical cost' (p. xi). The WB experts involved in the reform claimed, We [the WB] focused on increasing co-payments . . . but then they [the government] exempted the unemployed, veterans, children, voluntary blood donors, everyone who is below a certain income threshold, and then the question arose as to whether such a policy made any sense. (Interview 2, 2020) Co-payment has a corrective factor to prevent overuse of health services . . . For this reason, Akiko Maeda [WB consultant] objected to the introduction of complementary insurance because she wanted to instil . . . the cognitive feeling that there is economic value behind the individual health service. (Interview 11, 12 2019) However, the WB acknowledged that many of the proposed reforms which 'aim to dismantle Croatia's generous welfare state' would be very difficult to implement as they 'threaten significant vested interests and could engender a social backlash, particularly in an environment of already high unemployment' (WB, 2001d: 34). Moreover, the WB (2004) was satisfied with the government's progress in reducing the payroll tax, increasing control over the health fund by incorporating it into the general government budget, reducing sick pay and maternity leave, and broadening the contribution base to non-wage labour incomes. In turn, the WB softened its demands and adjusted its standpoint towards co-payment exemptions and CHI. It approved the Structural Adjustment Loan and concluded that Croatia fulfilled the necessary conditions concerning the increased price, scope and reduced exemptions from co-payments, and the broadened contribution base (WB, 2001c: 8).
Clearly, the WB did not have a firm stance on the conditionalities and was open to discussing them with the government. The coercive influence was limited to pro forma conditionalities which are open for negotiation and mutually agreed between the lender and the recipient (Killick, 1998). The WB experts indicated that The Bank has its desires, what it considers to do, the right sort of conditionality for a specific project and then it is a matter of negotiations with the country . . . they [the WB] send you a list and say this is what it is going to be, we can talk about it but this is our starting point. (Interview 3, 2020) First of all, there has to be volunteerism in cooperation . . . We made and published policy studies where we analysed healthcare financing and what the problems were, and then we defined policies based on that. But then of course we have negotiations with the government where we determine what is possible and what is not possible to introduce . . . The government says: ok within such a political economy we can or can't do this, we can introduce it in this way, in two steps in three steps etc. (Interview 2, 2020) Rather than forcing explicit and rigid conditionalities beforehand, the WB took into account the position of the Croatian government in determining the loan disbursement conditions, as long as it considered it viable to reach the loan objectives (e.g. financial sustainability of healthcare) (Interview 1, 2020; Interview 3 2020; WB, 2001d). The main goal of the WB was to provide different instruments for supporting the country's development, and create balanced priorities in open and deliberative negotiations with the government (Interview 1, 2020; Interview 2 2020; Interview 3, 2020; Interview 12,13 2019; Interview 13, 14 2019). Such an approach reflected the changing discourse within the WB (1997c) itself, stemming from the 1997 Health, Nutrition and Population Strategy which highlighted the adaptation of 'lending policies and procedures to client needs' and 'participatory approaches' that encourage government ownership (p. 16). Similar notions have been elaborated in its 1998 Assessing Aid publication (WB, 1998).
Besides the changing approach within the WB, another factor for the WB's adjustment to the Croatian government's position follows from the perverse incentives within the WB. After all, if the WB were to cancel the loan and leave the country, it would lose its influence and the potential to lend money in the future. As Mosley et al. (1995) argue, the system for promoting WB staff based on country lending targets creates perverse incentives to approve the loan despite deviations from the originally planned policies. The interviewed WB experts who worked in Croatia confirmed this: We [the WB] want to lend so we work with the governments, that is our mandate . . . The point is that we make our living as an institution by lending, so yes, if anything, we often end up being not too tough, but too lenient . . . I think the real concern with us is that probably we should be tougher. (Interview 1, 2020) Money out the door is what counts, that is what gets you promoted and there are perverse incentives in the Bank and you will not get promoted for denying the funds because the programme in the country was bad, ok . . . You failed, there is no money out the door. And that is a huge problem, that is absolutely a huge problem. (Interview 3, 2020) Similarly, one CIHI board member at the time noted that despite policy slippages, the WB wanted the reform processes to continue and that it was willing to adjust 'either by changing goals or changing a plan, etc. So, let's say there has never been such a firm stance [of the WB] . . . It should not be forgotten that they make money on it' (Interview 14, 15 2019). When a new government led by the CDU came to power in 2003, the WB (2004) encouraged it 'to explore alternative solutions' which would avoid 'the problems associated with adverse selection and moral hazard' (p. xi). The CDU government was open to restructuring CHI and co-payment exemptions, and in turn the WB offered policy lending and support by including the above-mentioned components in the future Programmatic Adjustment Loan (Independent Evaluation Group, 2006: 3). Clearly, the WB has a long-term interest to stay in the reform game and continue influencing the policymaking process in the future. To achieve this aim, the WB adjusted to the government's position regarding co-payment policy and eventually offered a new loan. Staying in the game allows the WB to act as a constant policy advisor shaping the government agenda and trying to correct previous policy setbacks while pressing demands for further changes.

Discussion and conclusions
The study explored the influence of the IMF and the WB in Croatian healthcare financing reforms in the early 2000s. During this time, Croatia was in an unfavourable political and economic position and its healthcare system was inefficient and experienced a sharp rise in expenditure. To initiate reforms in healthcare it turned to the IMF and the WB, who provided expertise and financial support. As in some other countries, the IMF delegated the healthcare issues to the WB due to their own lack of expertise (Odling-Smee, 2006: 182). In comparison to the IMF, the WB was heavily involved and mostly relied on its non-coercive influence such as producing healthcare financing studies, providing technical assistance, consulting and persuading the government.
The WB acted as an epistemic community (Haas, 1992), helping domestic policymakers to evaluate healthcare policies, and to define issues and solutions based on evidence. Its expertise and policy advice were tailored to the country-specific context and based on evaluation of the Croatian healthcare system and evidence from other countries. Similar observations have been made in other countries (see Druga, 2022;Noy, 2017Noy, , 2018Weyland, 2006;Yılmaz, 2017). In this way, the WB was able to control the government agenda and 'derive power through classifying the world, ordering information so that it is known and interpreted in a certain way' (Heneghan and Orenstein, 2019: 68; see also Lukes, 2005 for a discussion of the third dimension of power). It can be argued that the WB was mostly successful in this endeavour as Croatia indeed implemented most of its policy recommendations.
Although the WB formally imposed conditionalities, it is hard to categorise the latter as a form of true coercion. Instead of 'hard-core conditionalities' that are made only at the insistence of the lender, conditionalities had a non-coercive and voluntary character as they were negotiated and mutually agreed with the government (Killick, 1998). By opting for 'pro forma conditionalities' (Killick, 1998: 9), the WB provided the SDP/ CLSP government with significant room for manoeuvre to introduce policies which were not recommended by the WB, for example, CHI and broader co-payment exemptions. The question is why did the WB not use the leverage it had to impose its policies on a country whose political and economic position weakened its bargaining power?
The first argument stems from the WB's benevolent interest in supporting Croatia in its development. Instead of acting as a top-down institution prescribing rigid policies, the WB was flexible and took the Croatian political and economic context into account as well as already achieved reform progress and preferences of domestic policy actors. The WB adjusted its standpoint and allowed Croatia to introduce policies which would lessen the resistance from the opposition and prevent social backlash. This reflects the changing approach within the WB as outlined in the 1997 Health, Nutrition and Population Strategy and in its 1998 Assessing Aid publication. The new approach highlighted that relying on conditionalities without domestic support and government ownership is not effective in securing policy change (WB, 1997c(WB, , 1998. Hence, the desire to shift to pro forma conditionalities (Killick, 1998) and adapt to client needs (WB, 1997c). Other scholars have come to similar conclusions. Kaminska et al. (2021) show that in Albania and Poland, the WB did not unilaterally shape healthcare financing policies and that the role of domestic policymakers should not be understated. Also in Peru, Costa Rica, and Ghana, the WB adjusted to respective government priorities and supported their policy ideas (see Noy, 2017Noy, , 2018Weyland, 2006;Wireko and Béland, 2017).
Other arguments which explain the WB's leniency include systemic pressures within the WB to lend money (Killick, 1998;Mosley et al., 1995). The WB had an interest to stay in the reform game and provide financing and policy advice in the future. However, it is not only about future prospects for contracting loans, it is also about prolonging influence. Aware of conditionality limitations (Killick, 1998;Larmour, 2002;Mosley et al., 1995;Orenstein, 2008;WB, 1998WB, , 2005, the WB focuses on non-coercive influence which is more likely to take hold by ensuring long-term access to the most important policymaking actors and institutions. To this end, the WB is receptive to positions held by domestic policymakers, it is willing to bargain, make trade-offs and offer new loans. As Larmour (2002) argues, IOs 'do not like to seem to be pushing governments around. They seek influence, rather than control' (p. 259).
This was evident in Croatia as the WB did not only disburse the money for the 2001 loan, but also offered a new loan in 2005 which encompassed similar healthcare financing components. Similar processes can be observed in other countries. In Albania, Druga (2022) explains the WB's 'attempts to stay in the reform game even though the reform is not in line with its preferences, and after it failed to convince the Government of its preferred choice' (p. 3). In the Pacific, Larmour (2002) notes that the WB wanted a 'longterm relationship: the loan buys them a seat at the table, and they are usually happy to roll over another one to stay in the game' (p. 259).
To conclude, the study shows that the WB's main mode of influence in Croatia was non-coercive. Croatian policymakers were influenced by the WB's expertise and in the most part, followed its advice. However, some of the policies introduced differed from the policies initially advocated by the WB. This shows that coerciveness played a limited role. Indeed, the WB did not impose strict conditionalities and force Croatia to introduce all of its policy prescriptions. Moreover, it has been shown that the use of pro forma conditionalities blurs the line between coercion and non-coercion and that the WB prefers a less coercive approach which avoids unilateral policy prescriptions. The latter findings support more recent literature which suggests a move away from coerciveness or imposition towards the WB's readiness for collaborative interaction and accommodation (Noy, 2017(Noy, , 2018Wireko and Béland, 2017); In contrast to the literature which states that countries which find themselves in unfavourable position have little or no leverage against the WB, this case study supports the literature which provides evidence that the WB does not act as a hegemon and that even such countries can steer the direction of policy changes in their favour. It can be argued that the WB's influence is determined by the interplay of endogenous political and economic factors, interests of the recipient country and interests of the WB itself, which is, under certain circumstances, willing to soften its demands in return for the prospect of a long-term relationship. Finally, it is important to note that the findings of the paper are limited to a case study of Croatia at one period in time. Future research should offer insights on this topic by focusing on longer time frames and different countries and regions. Nevertheless, the study offers valuable insight on the interaction of IOs and aid recipient countries and different ways IOs can exert influence. At the same time, it shows that a country which was facing many challenges was able to bargain and make tradeoffs with powerful IO.

Funding
The author disclosed receipt of the following financial support for the research, authorship, and/ or publication of this article: "Deutsche Forschungsgemeinschaft (DFG,German Research Foundation), Grant/AwardNumber: Projektnummer 374666841 -SFB1342".

ORCID iD
Ante Malinar https://orcid.org/0000-0002-9793-3098 Notes 1. The study by Praljak (2006) estimates direct war damages at $56 billion and indirect war damages at $86 billion in the period from 1991-2004. Other studies are more modest and do not make the distinction between direct and indirect war damages e.g. $27.5 billion (WB, 2001a) 2. WB expert involved in the reform process. 3. WB expert involved in the reform process. 4. External consultant of the Croatian Ministry of Finance managing the World Bank social sector loan. 5. Croatian healthcare expert, external observer. 6. Croatian healthcare expert, external observer. 7. Croatian healthcare expert, external observer. 8. WB consultant, Croatian healthcare expert and assistant Minister of Health (2012-2014), external observer. 9. Besides healthcare financing, they included amendments to the pension law, market competition law, reducing barriers to foreign direct investment, reducing direct subsidies in the economy, and so on (WB, 2001c). 10. Croatian healthcare expert involved in Coalition of Health Associations in Croatia, external observer.