Shielded against risk? European donor co‐ordination in Palestine

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. Development Policy Review published by John Wiley & Sons Ltd on behalf of Overseas Development Institute 1Associate Professor at Ghent University, Belgium 2Researcher at Ghent University, Belgium 3PhD candidate at University of Bristol, UK 4Assistant Professor at Ghent University, Belgium 5Policy Advisor at HERWiN, Belgium


ORBIE Et al.
an empirical one-analysing the successfulness of four hitherto undocumented cases of co-ordination based on semi-structured interviews with 74 stakeholders during field visits to Palestine in 2017 and 2019-we inductively arrive at theoretical building blocks that should be tested, refined and elaborated in further research. As we may expect that European development policy will become increasingly "politicized" (Hackenesch et al., 2021), engaging in such an exercise would appear all the more necessary for both academics and practitioners.
This inductive approach informs the structure of the article. First, the methodological section reflects on the implications of this research strategy, which allows for "practical" insights from the field to emerge, and issues of data collection. Second, the literature study section reveals the gaps in existing literature and the limited theoretical guidance for explaining the (dis)continued shield effect in European co-ordination. Third, the empirical section analyses the evolution of four co-ordination schemes that were created for risk-shielding purposes. Fourth, the discussion section discusses the theoretical insight on how and under which conditions cracks in the shield may or may not lead to co-ordination failure. The conclusions synthesize the main findings and contributions of this research, explaining the usefulness of the "circling the wagons" metaphor and the promise of a politicization perspective for further research. We also indicate suggestions for practitioners and possible implications of the COVID-19 context for donor co-ordination.

| METHODOLOGY
Methodologically, this study takes a pragmatic and inductive approach to social sciences. The pragmatic approach implies that research is driven by "real life" and "practical" problems (see Friedrichs, 2009;Hellmann, 2009). The question of successful European co-ordination is of high interest and relevance to European officials, diplomats and development practitioners. Preliminary findings were discussed with stakeholders at various stages of the research (May 2017, June 2017 and June 2019), which helped to validate the main findings. An inductive approach is justified because there is almost no empirical research into donor co-ordination in Palestine, and because we lack strong theoretical frameworks for analysing the sustainability of risk shielding as a co-ordination motive.
This research strategy has implications for the case selection. The four cases identified were those that emerged as the most relevant example of European donor co-ordination from interviews with stakeholders during the first phase in March 2017. Starting from such a "typical" and "descriptive" case study rationale (Gerring & Cojocaru, 2016, p. 396), we operationalized successful co-ordination by three criteria and six sub-criteria, which emerged from how stakeholders evaluate co-ordination mechanisms (see Table 1). All cases have been coded by two researchers independently of each other to guarantee intercoder reliability. Subsequently, findings were put together and discrepancies resolved through further discussion between the co-authors and after verification with privileged observers in June 2019.
When interviewing stakeholders, we accounted for variables that are described in the literature on (European) donor co-ordination. However, the semi-structured nature of interviews allowed for the emergence of new variables such as "risk shielding" as a key motive for co-ordination. Yet, several donors mentioned that domestic contestation (often called "attacks") challenges and perhaps even jeopardizes continued co-ordination. Coincidentally, we witnessed the collapse of one case study during our first phase of field research (the Secretariat, abbreviated SECR) and we observed how an initially successful case became seriously challenged two years later (Joint Financing Arrangement, JFA). This is how the research came to resemble an "exploratory, most similar" case study design (Gerring & Cojocaru, 2016, p  sustainable shielding allow for the exploration of potential causes. Indeed, although the intention was not to develop a positivist research design, the four cases turned out to have some factors in common, such as far-reaching co-ordination (pooled funding), pre-existing co-ordination and like-mindedness between participants (see below). Unsurprisingly, all also indicated shielding as a main co-ordination drive. Together with the coincidental variation of successfulness between and within cases, this made it possible to conduct a minimalist form of outcome-oriented process tracing analysis (Beach, 2017) that allows us to infer theoretical insights on: (a) what happens when the cracks appear in the shield; and (b) under which conditions this is more or less likely to happen. Our data mostly stem from semi-structured interviews with 74 stakeholders in Jerusalem and Ramallah (40 in 2017, 32 in 2019, sometimes the same people). These include European diplomats and development practitioners, Palestinian officials, experts and civil society organizations (CSOs). In the course of both visits in 2017 and 2019 we interviewed all donors involved in the co-ordination scheme as well as other relevant stakeholders when possible. Interviewees have been promised anonymity and have not been recorded on audio. Interviews took between 1-3 hours and were attended by one co-author who took written notes. Shortly after each interview, these notes were developed into transcripts that include key insights and an initial coding. Data from interviews were triangulated with primary sources, existing evaluations, verification meetings and secondary literature.

| LITERATURE STUDY
As explained above, the shield effect and its sustainability came up inductively as from the empirical research and was originally not conceptualized. The aims of this section are, first, to illustrate the gaps in existing literature, and second, to search for theoretical handles that may help to understand the "shield effect." Emerging scholarship on European donor co-ordination mostly addresses the question of why EU member states (do not) engage in co-ordination. After all, development is "only" a shared competence of the EU and member states jealously guard their national prerogatives (Carbone, 2017;Lightfoot & Szent-Iványi, 2014;Orbie & Lightfoot, 2017). Explanations for (un)successful co-ordination can be classified in different ways. Rationalist accounts focus on incentive structures of donors' development institutions (e.g. Klingebiel et al., 2017;Koch et al., 2016) whereas constructivists highlight donors' shared convictions about the need to collaborate (e.g. Delputte, 2013). Explanations could also be categorized in terms of the three i's: interests (e.g. member states' national interests, see Bodenstein et al., 2016), ideas (e.g. like-mindedness, see de Felice, 2016;Delputte & Söderbaum, 2012;Saltnes, 2018) and institutions (O'Riordan et al., 2011). Authors have also often used the pragmatic distinction between "enabling" and "constraining" factors (e.g. Carbone, 2017;Delputte, 2013).
However, literature has been surprisingly silent on one motivation that appears key in our cases: development co-ordination as a deliberate strategy by donors for "risk shielding" when faced with domestic contestation. A recent special issue on "collective action" in EU development policy lists various obstacles against co-ordination and strategies to overcome these, but apart from a brief reference to the potential "political fall-out of member state elections" (Bodenstein et al., 2016, p. 444), the publication does not elaborate on the implications of domestic contestation. Similarly, an article on the political economy of EU aid co-ordination extensively surveys the advantages and disadvantages for member states and recipients (Klingebiel et al., 2017, p. 152), without considering the issue of risk-shielding against critiques back home. When risks are mentioned, this is from an entirely different perspective, namely the partner countries' interest in downplaying co-ordination in order to keep some independence from donors (Klingebiel et al., 2017, p. 148). In the same vein, risk shielding as a driver for co-ordination is entirely absent from Carbone's (2017) comprehensive analysis of European co-ordination "on the ground", and from Lundsgaarde and Keijzer's (2019) elaborate overview of obstacles against co-ordination and (creative) alternative co-ordination schemes. While Delputte (2013, p. 296) identified spreading the risk of potential development failures as a potentially enabling factor and domestic politics as a mainly constraining factor for aid co-ordination (Delputte, 2013, p. 304), she does not elaborate on their mutual relation.
Policy-oriented work has been somewhat more attentive to the "risk" factor in (European) development co-ordination. However, and in line with general critiques against the aid effectiveness agenda (see Wilks, 2010), such risks have usually been approached in a technocratic way (in terms of "fiduciary risks") and without elaboration of political dynamics. A wide-ranging study for the European Parliament frames the issue in terms of the "cost" of non-European co-ordination. Aversion to risk is considered an obstacle against co-ordination rather than an incentive for donors to work together (Wilks,p. 28), or as something that concerns partner countries who have an interest in aid fragmentation (Wilks,p. 35). Similarly, a study for the European Commission on delegated co-operation merely mentions risks sharing in terms of managing "politically sensitive situations" within the partner country (De Groot et al., 2016, p. 9). An OECD (2014) study that is entirely devoted to "risk" in development co-operation does mention "political and reputational pressures" in home countries and donors' risk aversion stemming from "domestic scrutiny of aid programmes through the media, parliament and public voice" (OECD, 2014, p. 39), and the "institutional risk" involving inter alia "political damage in home country" (OECD, p. 21). Again, however, this is not elaborated, and donor co-ordination is merely presented as one of the many ways to manage such risks (see also DFID, 2010).
On the other hand, domestic contestation of European aid has recently received some more attention. Koch et al. (2016) point to domestic politics in order to explain the waning popularity of the budget support modality. They also describe how domestic politics played out differently in different European headquarters and therefore hindered co-ordinated action on budget support (Koch et al.,p. 469). Hence, they do not consider such risks as an incentive for co-ordination in terms of risk shielding, but instead as something that fails to happen because of diverse domestic preferences. Similarly, an entire special issue of the Journal of Common Market Studies is devoted to the politicization of European aid (Hackenesch et al., 2021). Again, however, domestic contestation is not theorized in terms of stimulating European development co-ordination. Instead, most contributions implicitly or explicitly conceive this as an obstacle for European co-ordination. This resonates with a wider tendency in recent literature to highlight donor fatigue with the aid effectiveness agenda which includes donor co-ordination as one of its core pillars (Aldasoro et al., 2010;Brown, 2020;Delputte, 2013;Fejerskov, 2013;Keijzer et al., 2019;Lundsgaarde, 2012).
Literature on EU foreign policy co-operation only offers slightly more insights. Interestingly, Weiler mentions a "shield effect" in an early contribution on EU-Israel relations (Weiler, 1988, p. 250). Without much elaboration, he sees the shield effect as a government strategy to explain positions that face parliamentary or popular opposition as being in line with "a common European line" (Weiler,p. 250). The shield effect is defined more precisely in Smith's work on EU human rights promotion, namely "hiding behind the collective decision, disclaiming responsibility and citing the exigencies of going along with everyone else" (Smith, 2001, p. 196). However, the authors merely present this as a side note within an exposé that instead stresses the EU's difficulties to reach common positions (Smith, 2006, p. 116). In a chapter on Danish and Irish foreign policies, Tonra also mentions the existence of a "European shield … used to deflect domestic political pressure on foreign policy issues" (Tonra, 2000, pp. 233-4). On Spanish foreign policy, Barbé notices that the EU has been used as "a 'shield' to justify decisions that are likely to be badly received by public opinion" (Barbé, 2000, pp. 47-48). Interestingly, her example concerns the González government defending diplomatic recognition of Israel to the Spanish public (Barbé, 2000). Müller (2011) also refers to a "shield effect"-in relation to the Israeli-Palestinian conflict, but here it involves Germany's support for Palestinian rights under the European umbrella. The German government has, to some extent, used the "legitimizing function of Europe's common foreign policy" (Müller, 2011, p. 399) faced with "the costs of pursuing a controversial policy" (Müller, 2011, p. 387). All these accounts have in common, first, that they emphasize the limits of this the shield effect when national preferences are too strong, and second, that they lack any theorization of this phenomenon.
More theoretical guidance can be found in the literature on depoliticization. European donor coordination could indeed be conceptualized as a depoliticization strategy that intends to shield against contestation. Specifically, it would concern "institutional depoliticization," defined as the "delegation" of issues by politicians to "arm's-length bodies, judicial structures or technocratic rule-based systems that limit discretion" (Wood & Flinders, 2014, p. 165). Governments pursue such delegation for two related objectives. First, non-majoritarian institutions are less prone to short-term interests and more likely to pursue long-term goals (Flinders & Buller, 2006). Translated to (European) development, co-ordination schemes are indeed expected to follow the aid effectiveness principles without political interference. Second, delegation can be adopted to avoid blame. In this regard, Schimmelfennig mentions the "circling the wagons" strategy whereby politicians design institutions in such a way that blame can be "diffused" to several participants (Schimmelfennig, 2020, p. 348). Weaver describes "circling the wagons" as follows: "no one has to stick their neck out: everyone provides political cover for everyone else" (Weaver, 1986, p. 389). Applied to European co-ordination in development, this does indeed resemble the shielding strategy.
Relevant for the remainder of this article is that Weaver considers circling the wagons to be a "risky" blame avoidance strategy when there is no unanimity among the participants (Weaver, 1986, p. 389). This cautious note reflects the assessment of Weiler (1988), Smith (2001Smith ( , 2006, Tonra (2000), Barbé (2000) and Müller (2011) about the limits of the shield effect in European foreign policy. However, none of these authors elaborate on these predictions. Moreover, we lack theoretical tools that account for the sustainability of such depoliticization strategies. While literature has elaborated on the motivations behind institutional depoliticization and delegation to non-majoritarian institutions (e.g. Buller, 2006;Wood & Flinders, 2014), we have to the best of our knowledge no insights on the conditions under which the risk-shielding strategies continue to be effective in the long run. Similarly, scholarship on European donor co-ordination has not considered why some initially successful coordination schemes continue to exist for years whereas others eventually collapse.
Against this backdrop, the present research aims not only to provide an empirical contribution but also to gain analytical insights into the sustainability of donor co-ordination schemes that are designed as depoliticization devices. Inductively derived from the analysis of four cases, we will elaborate a script of how domestic contestation may erode the co-ordination shield and identify three conditions for sustainable co-ordination.

| Four cases of co-ordination as risk shielding
We examined four cases of European donor co-ordination: the Joint Financing Arrangement (JFA), a pooled fund in the education sector, Transfer Agreements to the EU for social infrastructure in Area C, a pooled fund managed by the Food and Agricultural Organization (FAO) in agricultural development, and a Secretariat (SECR) channelling funds to human rights civil society organizations. Although each of them has enough interesting dimensions that would warrant single and in-depth case studies, the four have a number of common features that allow for a comparative analysis into when and how co-ordination practices continue to function as a shield against domestic contestation. First, they were all (initially) considered to be successful cases of donor co-ordination. Second, co-ordination between donors has been longstanding (JFA, AREA C and SECR) or building on pre-existing co-operation (FAO). Third, they are all far-reaching forms of co-ordination (or "delegation") in the form of pooled funding or delegated budgets (De Maesschalck et al., 2014). Fourth, they are (or were) all characterized by a significant degree of like-mindedness between the participants.
Last but not least, all donors involved pointed at the importance of the "shield" effect, namely the ability to stand together against pressure from domestic constituencies and to hide behind each other when they are "attacked." Area C infrastructure investment is perceived to entail risks, much more than humanitarian aid. Reflecting on the shield motive, one participant noticed that "it was not brave … but it was more convenient" (2017/7). Another stated: "If Israel would come and demolish the schools, kindergartens, etc., it is not going to be France or Denmark: it is going to be 'EU'" (2017/8); and "on risk sharing and hiding this project is definitely a big plus" (2019/7). Similarly, the JFPs strongly emphasize the importance of shielding through co-ordination. Already in 2017, the education sector was considered to be sensitive because Palestinian handbooks may include incitement against Israel or display historically doubtful maps, because schools may be named after martyrs or terrorists, and because investments risk being demolished by the Israeli authorities. JFPs appreciate the joint consortia for being less visible as an individual donor and acting jointly when a school risks being demolished (2017/17; 2017/13; 2017/18; 2017/23). This was occasionally expressed more cynically: "We are all cowards; we don't want to have our flag there-if it goes to the JFA it is the Palestinians doing it" (2017/13). A donor of the SECR stated that "if alone we would be even more attacked … you would go for the safe options, and not take the political risks" (2017/1; 2017/25). Consortium members co-ordinated responses to "attacks" and in correspondence to headquarters shielded with reference to their European partners (2017/1). While shielding is somewhat less highlighted by FAO donors, they also indicate that they do not envisage a high-profile role because of political sensitivities (2017/14; 2019/31) and that co-ordination makes it easier to "share risks," especially for the Netherlands and Denmark (2017/1; 2017/2; 2019/1; 2019/2). "Standing stronger together" would become all the more important if they were to decide to implement also in Area C (2019/31; 2019/15; 2019/4; 2019/6).

| Varying successfulness between and within cases
Although all cases were (initially) relatively successful, there was also significant variation between and within cases. We operationalized successful co-ordination by three criteria and six sub-criteria, which emerged from how stakeholders themselves evaluate co-ordination mechanisms. First, the membership of the donor group, more specifically whether the membership is stable over time (no withdrawals) and new donors are attracted, and whether the donor group has a critical mass in terms of budget, expertise and strength of donors. Second, institutional functioning relates to both the effectiveness delegation (autonomy of the lead donor) and practical functioning (day-to-day co-operation processes). Third, the perception of stakeholders (donors involved and externals) of co-ordination practices and whether they see it as a model for other sectors in the country or for the same sector in other countries.
Despite intercoder reliability and consultation with stakeholders, this coding is not an exact measurement. The results should rather be seen as indicating relative positions between and within coordination practices. This is described in the remainder of this section.

| JFA: From model to headache
The JFA is a pooled fund supporting the implementation of the Education Development Strategic Plan (EDSP) of the Palestinian Authority (PA). In 2017 the JFA was still widely considered as an example of best practice in donor co-ordination in Palestine. Stakeholders on donor and Palestinian sides described the JFA as a "model" and "example" of co-ordination for other sectors such as health and agriculture. They had even been consulted on how the JFA could be extended to other ministries (e.g. 2017/21; 2017/39; 2017/34). A report from the Ministry called it a "unique example in Palestine" (MoEHE, 2014, p. 146)  Meanwhile, debates in Germany concern alleged biases in Palestinian schoolbooks that it (co-)financed. Some anticipate that also Germany may have to leave the JFA (2019/27; 2019/21). Norway also experiences contestation on the domestic front. Although the institutional setting of the JFA remains unchanged, its practical functioning has suffered because of the time and energy that JFPs are investing in responding to the "attacks" from Israeli lobbies and domestic constituencies (2019/20; 2019/21; 2019/23). In this new context, the JFA is no longer mentioned as a co-ordination model. Whereas its budgets, expertise and institutions still constitute key strengths, the co-ordination scheme faces existential challenges and has been put on the defensive.

| Area C: A model of delegation
Via Transfer Agreements with Denmark, France/AFD and the UK/DFID, the EU manages the contributions to the social infrastructure component of a wider EU programme in Area C territory. Area C encompasses 60% of the West Bank that remains under Israeli control. As Israeli authorities almost never approve building permits as provided by the "Master Plans for Development" of the village councils, this programme takes a different approach. It considers the Master Plans to be tacitly approved by Israel if no major objections are raised within 18 months of their submission (Asseburg, 2019, p. 48). Considered a successful case of co-ordination in 2017, it continues to be highly rated for a number of reasons. Institutionally, the Transfer Agreements leave much discretion to the EU, which manages the contributions. This limits transaction costs and helps to avoid duplication and stimulate complementarity (2017/8; 2017/7). Furthermore, implementation of the social infrastructure projects is delegated to the Municipal Development Lending Fund (MDLF), a strong and well-established semi-governmental institution. Calling themselves the "four musketeers," the EU, Denmark, France and the UK meet informally to discuss the technical dimensions of the programme (2017/8; 2019/8; 2019/26; 2019/29). The Steering Committee between the donors and the MDLF is seen to be beneficial for information exchange (2017/25). In 2019 the UK decided to discontinue its participation but, unlike the Belgian exit from the JFA, this is unrelated to political contestation (instead, it related to bureaucratic restructurings between DFID and the Ministry of Foreign Affairs). Denmark decided to continue its support, AFD decided to double its funding from one million euros to two million, and the EU hopes to also increase its contribution depending on the MFF outcome (2019/24; 2019/8). All donors involved emphasize the EU leadership and MDLF efficiency as key strengths (2019/7; 2019/26; 2019/30; 2017/28; 2017/7; 2017/25). Stakeholders' perceptions have become even more positive over the past year. Interviewees emphasize that the "experiment" seems to have worked as there have been no demolitions and only three stop-orders out of 42 projects since 2014. The possibility to do longterm development in Area C is seen by European and Palestinian stakeholders as at least a symbolic breakthrough that might inspire more donors (2019/26; 2019/29; 2019/9). There are plans to upgrade the programme by focusing on Master Plans for clusters of villages instead of individual villages.

| FAO: (continuing) teething problems
The pooled fund managed by the FAO finances a programme in support of Palestinian agribusiness development through an integrated value-chain approach that favours market-oriented, competitive and profitable agribusiness. Its most successful element is critical mass in terms of budget and expertise. As well as the FAO, most other participants had worked in agriculture in Palestine. Only Denmark had less experience, but it invested in an expert on agriculture who co-ordinates the donors' positions vis-à-vis the FAO (2019/2; 2019/4; 2019/31). Membership has remained stable. However, the programme has suffered from institutional shortcomings even before its establishment. The twoyear negotiations were referred to by interviewees as "nightmare," "mess," "headache," and "hell" (2017/3; 2017/6; 2017/15; 2017/16). In 2017 the donors expressed their hope that the "high transaction cost" in negotiation would pay off in smoother implementation (2017/4; also 2017/25). However, two years later, the programme still suffered from these ills. Participants complain about "micromanagement" and extensive meetings (2019/2; 2019/4; 2019/15; 2019/17). The FAO continues to lament the diversity of requirements by the European donors. While the work of the Danish co-ordinator is praised, interviewees confirm that there is no division of labour. Hence it may not be surprising that participants are modest about the successfulness of the FAO scheme. In 2017 the common conclusion was that "it is too early to say" whether it will work (2017/15; 2017/4). In 2019 the donors repeat this, as the effective implementation of the programme still has to start.

| Secretariat: A death foretold
Despite its relatively large budget and longstanding expertise in funding human rights organizations, the SECR collapsed in mid-2018. Norway left in 2016 and Denmark and Sweden pulled the plug in 2018. In 2017 interviewees emphasized the institutional faults of this co-operation scheme. "High transaction costs" (2017/14) affected every detail, such as wording of contracts, costs, overheads percentage, terms of reference for audits, financial reporting standards or procurement, all needed to be discussed extensively and iteratively among the partners (2017/5). There was no clear division of labour. The institutional setting with NIRAS in Sweden and Birzeit in Palestine was also sub-optimal (2017/1; 2017/5). The former Norwegian member stated: "we ended up administering the secretariat instead of dialogue with the organizations." Nor was less contact with CSOs the price for more | 11 ORBIE Et al. efficiency, as this was not achieved. A critical mid-term evaluation already indicated a "deterioration" of co-ordination (Sønderskov & Maali, 2016, p. 39). Following the SECR collapse, European donors in Palestine have been exploring if and how co-ordination on funding for human rights organizations could be continued (2019/19; 2019/24; 2019/22).
In conclusion, the Area C and SECR cases emerge as respectively the most and least successful co-ordination practices. The FAO scheme is somewhere in the middle. The most striking finding concerns the JFA scheme which despite years of relatively successful experiences has recently found itself in heavy weather. This variation within and between the four cases makes it possible to reflect on explanations for more or less successful co-ordination, as will be done in the next section.

| DISCUSSION: WHEN THE SHIELD BEGINS TO CRACK
This section discusses some analytical insights that can be derived from the within-and cross-case analysis, aiming to better conceptualize the sustainability of institutional depoliticization or "circling the wagons" strategies. First, we discern a script whereby domestic contestation with one participant eventually entails failed donor co-ordination. Second, we identify three conditions under which this script seems less likely to be followed and hence co-ordination may be more sustainable. Each time we substantiate insights with empirical evidence and indicate suggestions for further research.

| A script of co-ordination failure
Analysing commonalities between the SECR and JFA cases, we can discern a script consisting of five phases whereby domestic contestation eventually seriously challenges co-ordination. While the phases partly overlap in practice, they can be distinguished analytically.
In the first phase, contestation emerges in one or several participants' home countries. The SECR was already under attack around 2015. The centre-right government that came into power in Denmark in 2015 (and the Danish foreign minister's friendship with Netanyahu, according to one source) and the many questions in the Swiss and Dutch parliaments were cited as proof of growing domestic contestation (2019/3; 2019/28; 2017/1; 2017/2). In 2017 some JFPs mentioned pressures from their domestic media, parliaments and governments (2019/20; 2019/21; 2019/22). It was already clear that their support of the Palestinian Ministry of Education was a sensitive issue.
In the second phase, this entails the withdrawal of one participant of the co-ordination scheme. When Norway joined the SECR in 2016, the government was "heavily attacked." It subsequently announced that funding the SECR was a mistake and withdrew (2019/20; 2019/21; 2019/22). Belgium left the JFA in September 2018, following domestic contestation on the naming of a school funded with Belgian aid money.
In the third phase, the remaining donors display more risk-averse behaviour and frustration with co-operation increases. In the consortium of the SECR, those donors that were most intensely scrutinized domestically, namely Denmark, the Netherlands and Switzerland, became more cautious about the SECR financing of human rights organizations. Fearing that funded organizations were too radical or indirectly associated to terrorism, they started to engage in extensive micro-management. This provoked intensive discussions about even the smallest details. The situation was all the more frustrating for Sweden, which still feels supported domestically. Similarly, attacks against the JFA have taken much time and energy from JFPs, which triggered risk-aversive behaviour from those participants that faced domestic contestation. Germany started to put more emphasis on infrastructure (building of schools) and became less eager to co-operate with the Palestinian Ministry of Education (2019/27; 2019/23; 2019/11). This new context frustrated Ireland and Finland, which would like to move ahead with their collaboration with the Palestinian government.
In the fourth phase, like-mindedness between participants erodes as underlying differences are magnified. Within the SECR, different country approaches to the Israel-Palestine conflict surfaced when the donor group was attacked (Sønderskov & Maali, 2016, p. 7). The fact that Sweden was the only participant to recognize the Palestinian state and that it could fund high-profile organizations (2017/1) became a divisive element in the group. Fault lines also emerged in the JFA. The withdrawal of Belgium distorted the balance within the group and sharpened divergences (2019/27; 2019/11). A debate emerged between JFPs on alignment with the government-a key pillar of the aid effectiveness agenda-whereby some advocated a move towards sector budget and others insisted on more donor control (2019/27; 2019/19; see also Wolsey & de Groot, 2018, pp. 35-37).
In the fifth phase, the shielding benefits have dissolved to such an extent that the institution breaks down. This happened with the SECR. Early in 2019 the German government nearly decided to end Germany's participation in the JFA (2019/10; 2019/11). The donors managed to "buy some time" by commissioning an independent study on schoolbooks in Palestine (2019/21; 2019/23). As one senior diplomat formulated it in relation to the SECR: "Together we can handle bigger storms than when we are alone, but if a storm gets too big we all go down" (2019/24). On the JFA, a similar dynamic was described: "Sometimes you can hide and stand strong as a group. But, on the other hand, if one falls there is a risk for the others" (2019/10).

| Three conditions for sustainable co-ordination
Analysing evolutions within and between the four cases, we can identify three conditions that explain why some schemes have remained more shielded than others.
First, the donor composition matters. Involvement of partners where aid to Palestine is more prone to domestic contestation, such as Germany (JFA), the Netherlands and Denmark (both SECR), is riskier in triggering the first stage of the script. In contrast, co-ordination schemes with donors where such aid is less contested, such as Spain (FAO) and France (AREAC), have more chances of success. More research is required into the causal chain between aid contestation "back home" and development policy abroad. Studies of the foreign policies of European countries towards the Israeli-Palestinian conflict (Asseburg & Goren, 2019; The occupation of Palestinian territories since 1967: An analysis of Europe's role [Special Issue], 2018;Martins, 2015) and on domestic contestation in this regard (Bicchi, 2018b;Voltolini, 2020) do not elaborate on aid co-ordination and how this feeds back into domestic contestation. In turn, politicization scholars have barely researched when and how a traditionally technocratic domain as development aid witnesses a higher involvement of politicians (politicization as "bureaucratization"; de Wilde, 2011) and an increase in the scope of conflict with the wider public (politicization as increasing "salience," "polarization" and "mobilization"; de Wilde et al., 2016;Schimmelfennig, 2020).
Inclusion of donors that are more vulnerable to domestic contestation does not, however, sufficiently explain variations between and within cases. Danish participation in FAO and AREAC coordination, as well as Dutch participation in the FAO, have not been challenged. Interaction with two other conditions needs to be considered.
Second, inclusion of international organizations can make a difference. The relatively successful co-ordination schemes involve the EU (AREAC) and the FAO as the lead donor. The latter also includes the EU as a participant. In contrast, international organizations are not included in the SECR | 13 ORBIE Et al. and JFA. The issue seems to be not so much the degree of delegation in institutional terms, which is high in the four cases. Rather, based on the four cases, these international organizations provide an additional cushion to the shield because they have a more neutral profile and because "domestic" contestation is more difficult to organize. AREA C stakeholders often stressed the importance of EU backing through Council Conclusions on the Middle East process (FAC, 2012). One senior diplomat explained that the EU has been relatively "immune" to politicization because the Israeli and Palestinian lobby balance themselves in the European Parliament, contrary to the parliaments of some member states. Similarly, FAO participants regularly highlighted the good relations of the FAO with the Israeli and Palestinian authorities as an advantage (2017/25; 2017/1; 2017/2; 2019/2; 2019/4; 2019/31).
When and how international organizations are effectively "double shielded" requires further investigation. For instance, there may be a spurious causation between the involvement of international organizations and successful co-ordination, as the former may typically promote programmes that are less sensitive-which brings us to the third condition.
Third, the political relevance of the aid programme influences contestation. In this context, political relevance concerns the extent to which aid to Palestine limits the power of the Israeli authorities. While such analysis is beyond the scope of this article, insights from our cases suggest that higher political impact increases the likelihood of donors being targeted domestically. Indeed, the more "successful" co-ordination cases are also the most modest ones in terms of changing power relations. The FAO approach is relatively uncontroversial: one participant noted that "even a right-wing party would support this" (2019/31); another source stated that the Danish government switched to agriculture in Palestine after the right-wing government came into power in 2015 because it is less sensitive (2017/1). The FAO programme complies with what Wildeman (2018) criticizes as a (neo)liberal and technocratic aid model that further dismantles the state. Indeed, the purpose is agricultural trade and not strengthening of the Palestinian state. Notably, one participant explained that private sector support in Palestine has become so popular because of "fatigue on the state-building agenda" (2019/24). Stakeholders also admitted that its impact could easily be curbed if Israeli authorities decide to impose export restrictions (2019/31; 2019/15; 2019/2). Moreover, participants have not yet decided if they are going to work in Area C which would increase the political relevance of the project (2019/4; 2019/15; 2019/31). As farmers are co-financing, they are also more risk averse (2019/31). The AREA C programme concerns very small budgets (see Table 1) for interventions on clinics (not hospitals) or kindergartens (not schools). While stressing their "symbolic" importance, interviewees confirm that "it is not huge, it is a start" (2019/29) and that wider and regional investments are needed to make a real impact (2019/8; 2019/19; 2019/26). Moreover, donors concede (and fear) that AREA C constructions could easily be demolished by Israeli authorities as has happened with other programmes (2019/29; 2019/7).
Conversely, the SECR and JFA are potentially more relevant. Even the staunchest critics acknowledge that the SECR core funding has been essential for a number of relevant human rights organizations in Palestine (and Israel). Beneficiaries of the SECR also greatly appreciated the funding and deplored its collapse (2019/28; 2019/12). However, many of the supported CSOs, such as Breaking the Silence and B'Tselem, were a thorn in the eye of the Israeli government. Similarly, the JFA impact as strengthening of the Ministry is widely recognized. Stakeholders and reviews stress that the JFA has improved the governance of the Ministry (2017/21; Bird et al., 2013) and even "acted as a main catalyst for several institutional and administrative improvements across the Ministry" (Ministry Of Education and Higher Education, 2014, p. 24;PESPP, 2014, p. 2).
The (negative) causality between political relevance and successfulness might fit with previous analyses suggesting that the political impact of aid in Palestine is non-existent and that there is even a risk of financing the occupation (see, for example, Le More, 2005). Again, further research should elaborate this empirically and theoretically. This should be done together with the previous suggestion on studying feedback loops between the political impact of aid on the ground and domestic contestation (and politicization) through various channels.

| CONCLUSION
The main aim and starting point of this research on European donor co-ordination in Palestine was to provide an empirical contribution to the literature by examining a particular country context and four hitherto unstudied cases. While we consider the thick description of how these co-ordination schemes have evolved as an added value in itself, our inductive research strategy and between/across case study analysis also raised two observations that have been neglected in the existing scholarship and that begged for more investigation: first, donors may co-ordinate in order to "shield" against domestic contestation; and second, this strategy may not always be successful. These findings then entailed additional analysis into how and when donor co-ordination schemes established for risk-shielding purposes may fail, eventually resulting in theoretical insights on: (a) what happens when the cracks appear in the shield; and (b) under which conditions this is more or less likely to happen. In doing so, we hope to have developed theoretical handrails that can be used for further research.
Reverting back to the literature study, our research therefore not only confirms that caution is warranted when assessing the "shield effect" and the "circling the wagons strategy," as briefly suggested by some authors; it also lays the first bricks of further theorization of this phenomenon and its sustainability. Perhaps more imaginatively, our contribution can be synthesized by extending the circling the wagons metaphor. The script's phases then analyse what happens (a) when one soldier protecting the wagon comes under fire and (b) perishes in the conflict, making it (c) an uphill battle for remaining soldiers and (d) causing frustration within the group (e) until they lose the fight. In turn, the conditions which make this more likely relate to (a) which soldiers constitute the weakest link of the circle; (b) whether additional defensive belts are built around the wagon; and (c) how important the content of the wagon is perceived by those attacking.
Clearly, more theoretical analysis and empirical investigation on the complex causalities between these phases and conditions is needed. For instance, it would be interesting to study the apparently negative feedback loop between political relevance and domestic contestation. While our study focused on European donor co-ordination for shielding on the ground, this would require a more indepth examination of domestic contestation and politicization of aid to Palestine, with attention paid to the relative impact of Palestinian, Israeli and other lobby groups (see Voltolini, 2020) as well as the complexities of the historical, economic, ideological and cultural context against which this takes place. The emerging literature on politicization in Europe (see Schimmelfennig, 2020) and European foreign and trade policy (Barbé & Morillas, 2019;Costa, 2019;Gheyle & De Ville, 2019) could provide useful guidance for this endeavour, not only for identifying different types of contestation and politicization across European donors but also for comparative analysis of the (f)actors that lead to more or less contestation and politicization. While we know that politicization of European aid varies significantly across EU member states, research into what form this takes and how it develops is still in its infancy (Hackenesch et al., 2021).
In practical terms, our research contributes to a better understanding for donors working in the field. One provisional lesson would be that they should consider going it alone if they are confident about domestic support, rather than taking the risk of being contaminated by partners that are more vulnerable to domestic contestation. In our study, this was particularly relevant for Ireland and Finland on education and Sweden on human rights. As one diplomat put it, donors can "shine in isolation"

| 15
ORBIE Et al. when risks are low (2019/24). Inversely, donors wanting to engage in sensitive programmes under a co-ordination shield are advised to integrate their aid into multilateral programmes, as was done by Denmark and the Netherlands on the FAO case and by Denmark and the UK on the Area C case. While Palestine is admittedly a unique context, these insights may well be relevant for development policy in other countries. Following other policy areas that used to be rather technocratic, such as trade and finance, the granting of aid is indeed also likely to become more scrutinized in Europe. Our research shows that, once the genie is out of the bottle, it is challenging to put it back in, and practitioners involved in development co-operation should prepare strategies on how to deal with this new situation.
Finally, scholars and practitioners will undoubtedly need to reflect on the impact of COVID-19 on these issues. In this regard, we can already indicate two tentative evolutions that were suggested by a privileged observer of aid in Palestine (2020/1) that are relevant for our story. First, it seems that the crisis has already stimulated some European donors into providing more short-term, humanitarian and health-related aid. While understandable, this development is not entirely guileless as it contributes to moving away from politically more relevant investments and eventually to financing the Israeli occupation. Second, strong limitations on physical meetings within Palestine and the return of some practitioners to working from headquarters have made it more challenging than ever for donors to meet and discuss matters in depth. As internal co-ordination within each donor's office has already become more difficult, one can imagine that the building and sustaining of shields could become all the more challenging.