Horizontal Accountability for SDG Implementation: A Comparative Cross-National Analysis of Emerging National Accountability Regimes

: Achieving the ambitious Sustainable Development Goals of the 2030 Agenda requires strong and accountable institutions. However, the focus on sustainability outcomes set forth in the Agenda’s 17 SDGs has side-lined the debate on accountability mechanisms to ensure effective policy implementation. Six years into the adoption of the 2030 Agenda, it is time to assess progress made in this regard. Drawing on theories of public interest and rational choice, this paper argues that horizontal accountability between different state organs will be key to the emergence of national SDG accountability regimes. We provide evidence on monitoring and evaluation mechanisms for SDG progress based on an original database (AccountSDG) built through a document analysis of 136 Voluntary National Reviews (VNR) submitted between 2016 and 2019. Our analysis provides the ﬁrst systematic cross-national empirical analysis of the role assigned to parliaments, Supreme Auditing Agencies (SAI), and National Human Rights Institutions (NHRI) in national processes of SDG implementation. We conclude that horizontal accountability, especially the role of SAI and NHRI, remains under-institutionalized in most countries. We conclude by discussing best practices and addressing the most important lacunae, as well as by pointing out limitations of our study and avenues for future research.


Introduction: Accountability as a Key Governance Challenge in Implementing the SDGs
In 2015, the UN member states adopted the 2030 Agenda for Sustainable Development, which sets out a 15-year plan to achieve 17 Sustainable Development Goals (SDGs) by 2030. Essentially, the SDGs constitute a universal call to end poverty, protect the planet, and improve the lives and livelihoods of everyone, everywhere. The SDGs replaced the Millennium Development Goals (MDGs). While the MDGs focused on a set of basic minimum living standards in developing countries and were mainly driven by aid donors [1,2], the SDGs are universal in scope and commit all nations in the Global North and South to contribute towards the effort for global sustainability. Commitments to improving governance were notably absent from the MDGs. In particular, the lack of systemic and thorough accountability mechanisms was seen as a major flaw of the MDG process and has been frequently cited as a key reason for the shortfalls in their achievement [3][4][5]. By contrast, thanks to SDG 16 on "peace, justice and strong institutions", elements of "good governance" (such as accountability, transparency, participatory decision-making, and inclusive institutions) have for the first time become part of a global development agenda [6]. Importantly, in the 2030 Agenda, governance is not only considered as a standalone goal, but recognized as an enabler for all other SDGs [7]. However, the SDGs are based on the principle of "governing through goals", and there are no formal sanctions for countries who do not comply with the agreement [1,8]. As Kanie [9] observes, it remains unclear how existing instruments, policies, and even institutions will adapt to goal-setting as a new global governance strategy. As the Agenda 2030 foresees no global compliance mechanisms such as regulations, monitoring, dispute resolution processes, and penalties for transgression, it will depend on member states' regulations and institutions to hold governments accountable for SDG implementation.
The neglect of accountability was already clearly reflected in the inter-governmental negotiations leading up to the adoption of the SDGs. Although the governance of SDG implementation was a concern of policymakers during the negotiations, critics remarked that many government representatives displayed a conveniently narrow concept of accountability, alleging that it could not be relevant because the SDGs would not be legally binding [10,11]. Consequently, the accountability vision offered by the 2030 Agenda remained relatively vague. Despite some dispersed references to accountability, the text of the declaration offers no clear guidance to member states on how to set up or adjust their national-level accountability frameworks. Instead, the chapter on follow-up and review merely points out the need for "effective, participatory, transparent and integrated" review processes (arts. 72 and 74.d), which should "build on existing platforms and processes, where these exist" (art. 74.f). Furthermore, article 79 encourages member states to organize their review processes in a way that includes a broad range of societal stakeholders and suggests that these processes could be supported by "parliaments as well as other institutions". Critical observers like Donald [11] disappointedly referred to these proposals as a rather tentative and under-ambitious negotiation outcome. Six years into the adoption of the 2030 Agenda, it is time to assess how governments have acted on these proposals. The present study thus constitutes a first exploratory step in the research to investigate national institutional arrangements to ensure SDG accountability.
Although reporting about SDG-specific governance structures has become a standard of almost all governments in their Voluntary National Reviews (VNRs) [12], accountability remains sidelined. In many places, civil society organizations have filled this void, aiming to strengthen accountability by highlighting best practices and gaps (e.g., [13][14][15]). While these initiatives constitute important examples of social accountability [16,17], much less is known about horizontal accountability mechanisms between state institutions. Few scholarly publications have commenced to explore the topic so far. Based on a literature review and through participation in policy-science dialogue events, Bowen et al. [18], for example, identify accountability as a major governance challenge for SDG implementation and conclude that it can be best pursued through systems for monitoring progress at the country level. A conceptual analysis by Karlsson-Vinkhuyzen, Dahl, and Persson [19], in turn, explores the question of whether and under which circumstances accountability is compatible with policy integration. Laberge and Touhiri [10] explore how availability of country-level data on SDG 16 can drive national accountability.
Our paper contributes to research and informs policymaking in the following way. Empirically, it provides the first cross-national analysis of national horizontal accountability mechanisms to ensure effective SDG implementation. On 31 May 2021, we conducted a search on the SCOPUS abstract and citation database of peer-reviewed literature with the following parameters: "Sustainable Development Goals" AND "horizontal accountability" are mentioned in the title, abstract, and/or keywords. The search returned 0 results. To broaden the search, we conducted an additional search with the following parameters: "Sustainable Development Goals" AND "accountability" are mentioned in the title, abstract, and/or keywords. This search returned 127 results, out of which only two publications [20,21] mentioned the relevant keywords "SDGs" and "accountability" in the title. Yamin and Mason [20] address the question of the relevance of universal health coverage for accountability and meeting the SDGs, and provide insights on the work of the WHO-ITU Commission on Information and Accountability in promoting accountability in global health governance. Pillai et al. [21] provide an analysis of how companies from five different countries integrate SDGs into their business activities. Having conducted a Sustainability 2021, 13, 7002 3 of 23 comprehensive review of the relevant literature, we can affirm that, to our knowledge, no such analyses have been conducted before. Specifically, we provide a systematic empirical analysis on the arrangements that states have adopted in order to enable parliaments, Supreme Auditing Institutions (SAI), and National Human Rights Institutions (NHRI) to oversee how their national governments perform in implementing their SDG commitments. We argue that such a stocktaking of existing national arrangements for horizontal accountability is both timely and necessary in order to obtain a baseline for future analyses testing hypothesis regarding the design and effectiveness of different institutional set-ups in achieving integrated and inclusive SDG implementation. Since evidence suggests that the SDGs will not be fully achieved by 2030 [22,23], it is even more important to establish a fundament for further assessments that goes beyond 2030. Conceptually, the paper offers an initial framework for the assessment of national SDG accountability regimes by mapping emergent mechanisms for SDG oversight, monitoring, and review against dimensions of accountability established by the political economy and rational choice strand of comparative politics.
The remainder of this paper is structured as follows: The next section elaborates on the conceptualization of horizontal accountability underlying this article and presents assumptions about the relation between horizontal accountability and effective SDG implementation. Special focus is put on enabling conditions for the engagement of parliaments, SAI, and NHRI in horizontal SDG accountability and on clarifying how different roles and functions assigned to these institutions in the SDG process contribute to different dimensions of accountability. In Section 3, we introduce the data and methods used in the empirical analysis. Section 4 renders a descriptive cross-national overview of the emerging horizontal accountability regimes for SDG implementation, discusses the most significant lacunae, and highlights some empirical examples. Section 5 concludes and discusses policy implications as well as limitations of the present study and avenues for future research.

Definition and Forms of Accountability
Accountability refers to a rule-based system that stimulates or constrains behaviour by holding actors responsible to their actions [24]. It entails three elements, namely information, answerability, and sanction [25][26][27]. The informational dimension "involves the right to receive information and the corresponding obligation to release all necessary details" [25]. Answerability "implies the right to receive an explanation and the corresponding duty to justify one's conduct" [25]. Sanction in the narrow sense is defined as the idea that accounting actors do not just call into question but also eventually reward good behaviour and punish improper behaviour [25]. Accordingly, accountable actors bear the consequences of their actions, including eventual negative sanctions. Consequences of sanctions do not necessarily imply punishment. A few authors do not include "sanction" as a constitutive element of accountability [16]. Instead, they argue that a critical reflection and positive assessment of the appropriateness of policies and the personal qualities of political actors might also facilitate policy change through learning [19,25]. However, in more general terms, the possibility of sanctioning proves to be decisive for effective accountability relations [28].
Furthermore, accountability has a temporal element, which is relevant for policymaking processes. Scholars distinguish between retrospective mechanisms (i.e., punitive, "post-factum" mechanisms such as trials, investigations, and consequent sanctions) and prospective mechanisms (i.e., preventive, "pre-factum" tools, ranging from routine audits and inspections to performance measures and reporting). Prospective accountability mechanisms are forward-looking and used to ensure appropriate behaviour and performance of bureaucrats and other public officials and complement retrospective, sanction-based mechanisms, which are backward-looking and focus on harms that have already occurred [29][30][31].
Given that accountability depicts the relationship between different actors, it is central to ask who is accountable to whom. Strategic actor theory literature has distinguished three main types of accountability. O'Donnell [17] characterizes accountability relations according to the type of actors involved and their hierarchical position to each other and uses spatial adjectives to describe these relations [29]. According to O'Donnell vertical accountability can be understood as a principal-agent relation, in which voters (as principals) hold governments (as their agents) to account through elections. Horizontal accountability, in turn, is exercised by the different branches of power who engage in mutual control and by the network of independent state institutions that are specialized in oversight, such as courts, parliaments, or audit institutions [32]. Furthermore, accountability over public officials can also be exercised by civil society organizations (CSO) and independent media. This type of accountability has been termed social accountability [16,17] and, in essence, also describes a vertical hierarchical relationship.
Among the three types of accountability described above, the one that is most decisive for effective implementation of the 2030 Agenda is horizontal accountability. For the time being, predominantly horizontal accountability mechanisms create sufficient pressure on governments to prioritize SDG implementation in the short run. In the following, we briefly outline the reasoning of this assessment.
Vertical accountability usually refers to voting by citizens, whereby salient valence issues such as the economy, crime, or education represent key signals, by which voters judge an incumbent's suitability to continue in office. Failure to deliver on these issues may thus lead voters to sanction incumbents by voting them out of office [33,34]. Although there has been a growing public debate about sustainable development since the publication of the Brundtland Report in 1987 (The so called "Brundtland Report" (in recognition of former Norwegian Prime Minister Gro Harlem Brundtland) was published in October 1987 by the United Nations through Oxford University Press under the title "Our Common Future". The report dealt with multilateralism and interdependence of nations in the search for a sustainable development path), success or failure to deliver on the SDGs as an integrated framework for action does not figure among the issues that receive sufficient attention from national publics to determine the electoral fate of national governments. Rather than that, as is the case with most UN processes, attention to the 2030 Agenda has largely remained confined to the political agendas of the UN member states and academic debates. This can partly be attributed to the fact that it is almost impossible for an individual citizen to marshal the necessary information that would be required for a close monitoring of government activities to implement such a highly complex target system as the 2030 Agenda [19,35] (In contrast, we acknowledge that the content of individual SDGs become relevant in elections under certain circumstances, for instance climate adaptation in OECD countries or food security in the least developed countries).
Social accountability, in turn, is performed by non-state actors like CSOs who voice demands regarding the lawfulness of state actions and engage in respective monitoring activities, as well as by independent media who expose and document governmental wrongdoing or performance deficits (e.g., [36]). In fact, the 2030 Agenda stresses "people" or "citizens", mostly in the form of national or transnational CSOs, as the main principals in the accountability regime for the SDGs (see [37], paragraphs 47, 73, 79, and 84). However, in many countries, the ability of CSO to perform their accountability functions is limited by the lack of funding for long-term monitoring and advocacy work, shrinking civic spaces, and governments' unwillingness to be held to account by them [19]. When it comes to accountability for SDG implementation, another limitation is imposed by the fact that CSOs often engage in advocacy on a specific set of priority issue. This clashes with the holistic vision of the 2030 Agenda as a network of indivisible and highly interdependent goals and targets.
In sum, vertical and social accountability are relevant for implementing the 2030 Agenda. However, a lack of systematic sanctions and limited capacities to monitor governments' actions towards achieving the SDGs make vertical accountability less effective than horizontal accountability. The latter is, thus, more likely to initiate domestic accountability systems for SDG implementation.

Horizontal Accountability for SDG Implementation
This paper analyses horizontal accountability, which Schedler [25] defines as a means of "subjecting power to the threat of sanctions, obliging it to be exercised in transparent ways and forcing it to justify its acts." As Ocampo and Gómez Arteaga [38] have pointed out, this definition implies that certain preconditions must be met for accountability to function properly: First, the action to be subjected to review must be clearly delimited. Second, the responsibility to perform the action must be clearly assigned to a person or specific set of actors. Third, an instance or mechanism must exist that makes it possible to monitor the execution action. Consequently, in order to characterize an accountability regime, three questions need to be addressed: Who is responsible? Over What? and How? [19,38]. Applying these preconditions and questions one-on-one to the subject of the 2030 Agenda is not an easy endeavour for a number of reasons.
First, implementing the 17 SDGs with their 169 sub-targets can hardly be described as a clearly delineated task. As Le Blanc [39] pointedly remarks "the novelty of the SDGs compared with their predecessors is that they aim to cover the whole sustainable development universe, which includes basically all areas of the human enterprise on Earth". Hence, the question "what action is to be accounted for?" is not applicable in this simplistic form when it comes to the SDGs.
Second, and directly related to this, problems arise when it comes to answering the question "who is responsible for the execution of this action (i.e., who should be held to account)?" At the global level, the SDG accountability system is led by the UN High Level Political Forum on Sustainable Development (HLPF). Established at the Rio Conference in 2012, the HLPF convenes annually under the auspices of the UN Economic and Social Council (ECOSOC) for eight days, including a three-day ministerial segment during which national governments (as account givers) report on their countries' SDG efforts and progress in the form of Voluntary National Reviews (VNR). Although governments are responsible for achieving the SDGs, they cannot necessarily be held accountable for a holistic implementation of the 2030 Agenda. In line with rational choice theory, governments are neither unified actors that can make cohesive decisions, nor can their decisions always be implemented as intended [40][41][42]. Instead, they consist of different ministries that are subject to pressure by different interest groups. In addition, the 2030 Agenda contains no specific indications as to which government actors should be entrusted with its implementation at the national level. Instead, the declaration acknowledges "different national realities and capacities" and pledges to "respect the national policy space" ([37], paragraph 21). As a result, national governments have considerable leeway in assigning responsibility for SDG implementation both horizontally between different ministries and vertically between different levels of government.
Third, referring to the "how" question, there is no single account holder (i.e., an individual entity that has the mandate to monitor the "action" of SDG implementation and to sanction failure to do so). Some countries have aligned their previously existing Sustainable Development Strategies with the SDGs, while others have set up national strategies or "roadmaps" for their implementation. However, these strategy documents rarely enjoy legal status. Rather than that, they constitute non-binding declarations of intent that are designed to provide a framework for the orientation of government action beyond individual legislative periods. Their non-binding character forecloses the possibility of judicial oversight and legal punitive sanction for failure to deliver on them.
In view of these problems, recent research-and to a greater extent grey policy literature-has started to address the related questions "through what processes can SDG accountability be assured at the national level?", "by which standards can SDG action be assessed?", and "with what effect can governments be held to account for their SDG-related commitments?" [19]. This emerging literature has identified parliaments and independent oversight agencies, such as Supreme Auditing Institutions (SAI) and National Human Rights Institutions (NHRI), as essential cornerstones for national SDG accountability regimes.
In the following, we will provide a brief overview of (a) the suggested ways in which these institutions could fulfil their role as national SDG account holders, and (b) constraining factors to the fulfilment of that role. In doing so, we depict these institutions' role in horizontal accountability as ideal types. Nevertheless, we recognize that, empirically, the accountability role of these institutions' accountability may be further constrained under certain structural conditions. In particular, autocratic regimes, per their definition, concentrate political power in the executive office, and developing countries have limited capacities to staff and maintain accountability institutions [43]. While accountability mechanisms may also exist in autocratic and low-state-capacity contexts, they might be more informal and personalized, and might be top-down rather than bottom-up. An emerging literature about accountability in autocratic regimes sheds light on these forms of accountability [44,45]. However, analysing the impact of these contextual factors and related alternative forms of accountability is beyond the scope of this analysis.

The Role of Parliaments in SDG Accountability
As Tsakatika [46] argues, in liberal democracies, parliamentary accountability is probably the "[ . . . ] most comprehensive in scope, insofar as the governing agent taken as a whole is accountable to the representatives of a demos for all policy areas and the linkages between them [ . . . ]" ( [46], pp. 556-557). Given the importance of parliaments, one could assume that they should be a key player in internationally agreed development agendas. However, this has not necessarily been the case in the past. Parliaments were not consulted on the adoption of the MDGs in 2000, and many of them were not even aware of the existence of the framework [47]. Drawing on this negative experience, very early on in the intergovernmental negotiations for the SDGs, the Inter-Parliamentary Union (IPU) began to organize a parliamentary input to the UN to ensure that the SDGs would include the views of parliamentarians. Furthermore, in collaboration with other partners, the IPU also lobbied for the adoption of SDG 16, which among other things calls for accountable and inclusive institutions at all levels [47].
Essentially, five main functions of parliaments in promoting accountability for the SDGs have been highlighted:

1.
Making or amending laws in a way that ensures consistency with the 2030 Agenda.
In 2015, the IPU Assembly adopted the Hanoi Declaration [48], in which member parliaments committed to translate the SDGs into enforceable domestic laws and regulations. In order to make laws that are consistent with the SDGs, it is necessary that parliamentarians are sufficiently informed about the scope and content of the goals in the first place. This requires the organization of training sessions and awareness-raising activities for parliamentarians and parliamentarian staff [49][50][51].

2.
Monitoring the actions of governments and their agencies in implementing the SDGs.
In order to exercise their oversight functions effectively, parliaments need to clearly assign responsibility for the issue to be monitored within their own structures. In the context of the 2030 Agenda, one possibility of doing so is through the establishment of a dedicated parliamentary SDG committee. Committee oversight is one of the strongest mechanisms of retrospective accountability through which parliaments can hold governments accountable for SDG implementation [49,50]. Committees enable parliamentarians to undertake in-depth examinations of government action by allocating more time to specific issues. Parliamentary committees typically have the authority to ask government agencies for information and documents, to interrogate government officials, and to conduct hearings and assessments. Therefore, committees enable parliamentarians to examine, in retrospect, whether policies, regulations, and programs were effectively implemented in support of the SDGs, and if not, to formulate recommendations on how to strengthen future SDG compliance. While the establishment of a new, dedicated SDG committee offers an opportunity for parliaments to make a strong public statement on the importance of the SDGs, parliaments may also decide to use existing structures for SDG oversight. The necessary subject expertise may already exist within committees that are well-established and have funding and supporting staff members at their disposal. Another, though less formalized, way for parliaments to join the forces of different stakeholders for discussing and overseeing SDG action is the establishment of cross-party parliamentary networks, working groups, or caucuses on the SDGs [13]. Such networks do not possess formal mechanisms to demand government explanations. Instead, their work typically aims at creating opportunities for cross-party exchange on possibilities for mainstreaming the SDGs into legislation.

3.
At the global level, the principal platform for monitoring SDG progress is the United Nations High-level Political Forum on Sustainable Development (HLPF). The HLPF convenes annually to assess progress towards the achievement of the SDGs based on the VNR. The process of conducting a VNR thus presents an opportunity to hold governments accountable for their efforts (or lack thereof) towards meeting the SDGs during the preceding reporting period. While VNR are normally government-led, a broad range of stakeholders, including parliaments, should be involved in the review process in order to provide a balanced account of national progress in SDG achievement [52]. However, conducting a VNR is a complex process and is therefore usually not carried out annually (According to the UN Sustainable Development Knowledge Platform, between 2016 and 2020, 129 countries presented one VNR at the HLPF, 26 countries presented two VNRs, and only two countries (Benin and Togo) presented three VNRs; https://sustainabledevelopment.un.org/vnrs/ (accessed on 17 June 2021)). In order to exercise oversight effectively, parliaments should therefore ask governments to present more frequent periodic progress reports on the SDGs beyond the VNR [13,49]. The tabling of such reports in parliament enables legislators to demand explanation on government SDG action.

4.
Budget evaluation that ensures an adequate allocation of financial resources to achieve the SDGs. Although the development of the national budget is formally the responsibility of the executive branch of government, parliaments are responsible for adopting national budgets and overseeing the use of public funds. Parliamentarians also play a crucial role in advocating for budget priorities in advance of the budget preparation process. In some countries, they can even use their law-making and oversight powers to propose formal modifications to the budget once it has been presented to parliament [53,54]. Budgetary control powers thus also equip parliaments with the ability to impose sanctions on the executive branch by not granting budgets for a certain policy [55,56]. Their vital role in the budgeting process helps them to influence government expenditure. Therefore, it has been postulated that parliaments should use the SDGs as an evaluation framework for budget proposals and make sure that government funds are being adequately allocated to nationally defined SDGs [13,49,57].

5.
Representing the interests of their constituents in the process of SDG implementation. The role of parliamentarians in representing the interests of their constituents is crucial for SDG accountability, particularly for vulnerable and marginalized groups who have few opportunities to voice their grievances [13]. The way and degree to which parliament and the government work together on the SDGs is relevant both for the due representation of citizen interests as well as for the integrated and coherent implementation of the SDGs. Governments and parliaments comprise a broad variety of institutions responsible for a wide range of subject areas. A national body responsible for coordinating the SDG efforts of these various institutions is therefore key to ensuring integrated implementation of the SDGs, whereas lack of coordination can lead to inconsistent laws and regulations that result in policy incoherence [12,[58][59][60]; consequently, most countries have established some form of taskforce or steering committee to oversee and guide national SDG implementation. While these bodies are usually set up by the executive branch of government, there is broad consensus that it is good practice to provide for their inclusive membership.
However, whether or not parliament should be represented in these bodies is a matter of debate. According to the IPU "in many countries, there are still questions about whether parliament should even be part of the SDG coordination body or whether its role should be to monitor the work of these bodies and hold them to account" [47]. From the latter point of view, it might rather be desirable that parliamentarians participate in working groups or technical committees that collaborate with and consult the executive-level national SDG coordination body. The representative function of parliament should also come to bear in the process of national planning as a future-oriented exercise. Since the adoption of the SDGs in 2015, many countries have started to revise their national development plans in order to align them with the SDGs [1] or have adopted specific SDG implementation plans or "road maps". So far, parliaments seem to be only weakly associated with these processes [47]. However, as representatives of the people, it is the responsibility of parliaments to ensure that government plans and programs meet the needs of their constituencies on the ground. For the adequate and context-sensitive localization of the SDGs, it would thus be desirable that parliamentarians receive the opportunity to feed their constituents' priorities into these processes. For prospective horizontal accountability to function effectively, parliaments should thus contribute to the development or revision of such plans through wide-ranging, public consultation processes, and only adopt them after a comprehensive review and formal debate [49]. Another potential avenue for parliaments to contribute towards localization of the 2030 Agenda is their involvement in the development of national priority goals and targets and related indicators for progress monitoring. Through adoption of such indicators, governments subject their SDG action to monitoring and make their performance measurable, including the requirement to provide explanations for underperformance.

The Role of Supreme Auditing Institutions in SDG Accountability
A country's SAI is an independent body with the mandate to safeguard the rational spending of public money [61][62][63]. To this purpose, SAI engage in auditing to evaluate governments' management of public funds, credibility of reported financial data, and policy conduct.
The demand for governmental auditing institutions results from agency problems that are inherent to principal-agent relationships in the legal-political system of a state [64]. Rational-choice literature identifies two main sources of agency problems. First, the principal and agent may have different preferences, and agents, as rational actors, seek to maximize their own utility rather than that of their principals. Second, an information asymmetry exists between the two parties whereby the agent has access to information that is not observed by the principal. This may lead to so called "moral hazard problems", whereby agents use their informational advantage to engage in actions that are not in the best interests of their principals [64,65]. In legislature-government relations, the governments' authority is constrained by the fact that they are financially accountable to the legislature, who must approve government spending. Furthermore, the government is responsible for the efficient implementation of approved programs. However, legislators cannot directly monitor government members' actions in order to ensure economically efficient public administration [64]. SAI can attenuate these agency problems through financial audits to ensure that financial transactions have been correctly accounted, efficiency audits aimed at revealing wasteful expenditure, and effectiveness audits to assess whether government programs or projects have met their intended policy goals.
Over time, SAIs have evolved from exercising purely financial oversight and have increasingly taken a broader, more comprehensive view on reliability, effectiveness, efficiency, and economy of policies and programmes [13,61]. It has thus been argued that they are also well-equipped to audit the compliance of their respective governments to international obligations and commitments if given the relevant mandate [19,[66][67][68]. Accordingly, the UN General Assembly recognized the conducive role of SAIs to the achievement of internationally agreed upon development goals [69] and encouraged its member states to give due consideration to the potential of SAIs to improve public accounting systems in the context of the 2030 Agenda [70]. Against this background, the International Organization of Supreme Audit Institutions (INTOSAI) has stressed the "natural mandate and prominent role" of its member organizations in the follow-up and review of implementation of SDGs [67] and has launched numerous initiatives to support them in contributing to the success of the 2030 Agenda.
A temporal distinction of audits became more relevant as SAIs evolved. Audits are either retrospective or prospective. Regarding the retrospective oversight of government action, SAIs can contribute to the follow-up and review process by verifying that the SDGs are adequately reflected in the national budget and by undertaking performance audits of governments' SDG implementation efforts [13].
Although not yet linked to the SDGs in the past, SAIs have, for example, reported on aspects of national environmental performance [43]. In addition, audits are being increasingly coordinated across borders. One example is a parallel audit to evaluate the implementation of indicators and targets of SDGs 1, 2, and 5 in ten Latin-American countries coordinated by the INTOSAI Development Initiative (IDI) and the Latin American and Caribbean Organization of Supreme Audit Institutions (OLACEFS). Furthermore, close involvement of SAIs in elaboration of their countries' VNRs will add extra credibility to the national review processes. Regarding the prospective oversight of government action, SAIs can also contribute to the implementation process in more proactive ways. One notable example is conducting audits of government preparedness for SDG enactment in the early implementation phase, which has thus far taken place in 45 countries. In conducting these audits, SAIs sought "to urge national governments into action, if there wasn't any and provided constructive recommendations at an early stage" [71]. Depending on whether such audits arrive at a positive or negative assessment and whether the corresponding report is made available to the public, SAI auditing can lead to reputational gain or loss, and thus represents a form of soft sanctioning.
It is evident that the closer SAIs cooperate with the government in the SDG implementation process, the better they can fulfil their proactive role of providing constructive criticism. It therefore seems advisable that SAIs should be represented in the national SDG coordination body or at least in one of the working groups or technical committees supporting this body. Furthermore, they are particularly qualified to contribute to the localisation of the 2030 Agenda by participating in the development of national indicator frameworks, and should be given the opportunity to feed their expertise in the development of national SDG implementation plans or processes of aligning national development plans to the 2030 Agenda.

The Role of National Human Rights Institutions in SDG Accountability
The 2030 Agenda explicitly references human rights. In article 4 of the declaration's preamble, the signatory states pledge that "no one will be left behind", recognize that "the dignity of the human person is fundamental", and commit to endeavour "to reach the furthest behind first". Furthermore, the 17 SDGs and their related 169 targets directly or indirectly reflect human rights standards. It has thus been stressed that, given their unique mandate and role, NHRI can play a key role in the implementation and follow-up of the Agenda, and "are at the core of the SDG 'web of accountability'" [72].
NHRI are state bodies with a legal mandate to hold their respective governments accountable to their human rights obligations through horizontal procedures. Hence, the main task of NHRI is to prevent, to discourage, and to promote the sanctioning of unlawful actions or omissions of other state agencies [32,73,74]. Furthermore, the mandate of NHRI include the empowerment of vulnerable people vis-à-vis state and non-state duty-bearers. They exercise these tasks by providing oversight and carrying out investigations. While oversight here refers to preventing poor performance or misconduct ex ante, investigation refers to revealing the extent of such misconduct ex post [74]. In their work, NHRI apply a human-rights-based approach, which is a conceptual framework directed towards promoting and protecting human rights, based on international human rights standards [75]. In 1993, the UN General assembly adopted the Paris Principles [76], which lay down the international standards that all NHRI must meet if they are to be legitimate and effective in promoting and protecting human rights (A plethora of CSOs with a focus on human rights exist both at the international and national levels. For reasons of feasibility, our analysis is restricted to NHRI that have been accredited by the Global Alliance of National Human Rights Institutions (GANHRI) with A or B status in compliance with the Paris Principles).
In 2015, the International Coordinating Committee of NHRI (now Global Alliance of NHRI, GANHRI) adopted the Mérida Declaration, which outlines the functions and activities that NHRIs can undertake in order to promote a human-rights-compliant, transparent, participatory, and inclusive implementation of the SDGs [13,77]. These include: • Provide advice to governments on the promotion of a human-rights-based approach to implementing the 2030 Agenda by assessing the impact of laws, policies, programs, national development plans, administrative practices, and budgets; • Engage with all stakeholders to raise awareness, build trust, and promote dialogue; • Assist in shaping national indicators and data collection systems; • Monitor progress at all levels to identify inequality and discrimination, through innovative and participatory approaches to data-collection and monitoring; • Engage with and hold governments to account for poor or uneven progress in implementation, by reporting on SDG progress and obstacles to parliaments, the general public, and national, regional, and international mechanisms;

conduct inquiries into, and investigate allegations of rights violations in the context of development and SDG implementation
It is important to note, though, that NHRIs are peculiar institutions with regard to their ability to impose sanctions for inappropriate behaviour, which is often perceived as an intrinsic part of retrospective accountability [19,25,78]. While they can receive public complaints and conduct investigations on human rights violations, they do not have the right to sanction misconduct directly. Rather than that, they can apply reputational sanctions through revealing misconduct in their reporting. In return, this also implies that a positive judgement of the conduct of the actor under review can lead to praise and reputational gain [19]. In order to stress the mutually reinforcing relationship between the SDGs and human rights and to align SDG monitoring with human rights monitoring, NHRI may also choose to map their existing work programs with the SDGs [72] in order to routinely include the assessment of human-rights-compliant implementation of SDGs in their reporting.
Furthermore, like SAIs, NHRIs should have the opportunity to include their expertise in the development of national SDG implementation plans or processes of aligning national development plans to the 2030 Agenda. Their chances of doing so will be better if they closely cooperate with the government and are represented either in the national SDG coordination body, or in working groups and technical committees collaborating with this body. Finally, their involvement in the elaboration of their countries' VNRs will add credibility to the national review processes.

Methodology and Data
The descriptive analysis of the paper focuses on horizontal accountability for SDG implementation; in particular, it focuses on the role of the three above-mentioned institutions (parliaments, SAIs, NHRIs). From a temporal perspective, we assume that these institutions engage in holding governments accountable either retrospectively or prospectively. In order to capture whether these institutions take an accountability role in SDG implementation, we analyse the governance mechanisms, which UN member states, which constitute our universe of cases, established after the adoption of the 2030 Agenda in 2015 [12,79].
The aim of this study is to provide the most comprehensive possible cross-national overview of the phenomenon under study. Since quantitative data on the phenomenon of "national, horizontal SDG accountability" are not yet available, its investigation necessarily involves a qualitative research approach. However, the application of typical qualitative research methods such as expert interviews, focus groups, or direct observation would not have allowed us to cover a universe of cases as large as the UN member states (However, this analysis is part of a research project with a mixed methods approach. Past and planned case studies, which apply interviews and social network analysis, will allow us to refine the dataset over time).
In order to collect the necessary data, we therefore opted for qualitative, directed content analysis of the Voluntary National Reviews (VNR) presented to the UN High-level Political Forum on Sustainable Development (HLPF) between 2016 and 2019 as our data collection method. Directed content analysis is an established method to investigate a phenomenon on which prior research is incomplete or would benefit from further description [80]. We compiled the obtained information on horizontal SDG accountability mechanisms in an original dataset (AccountSDGs) that constitutes the basis of our study.
VNRs are both a statement of intent of governments and a point of departure for an ex post assessment of achievements. However, it is pertinent to note that the VNRs present some limitations as an evidence base. Most importantly, based on document analysis alone, it is not possible to assess the degree to which government statements made in VNRs reflect political practice. Nevertheless, at this point they constitute the sole internationally standardized and comparable documentation of SDG implementation efforts and the related national institutional arrangements.
Following the recommendations of UN-DESA's "Handbook for the preparations of Voluntary National Reviews" [81], these documents typically contain a dedicated section on "institutional mechanisms" that provides information on how the country has adapted its institutional framework in light of the 2030 Agenda. This chapter typically includes information on how the views of different ministries and state agencies are considered, and how they engage in the coordination, follow-up, and review of SDG implementation. In some cases, however, this information may also be dispersed throughout the text or contained in a technical annex.
In a first step, based on the arguments and recommendations put forth the literature reviewed in the previous section, we distilled a list of 34 variables concerning the way in which parliaments, SAIs, and NHRIs can exercise horizontal accountability in the SDG process, and accordingly, how they can be mentioned in the VNR (Figure 1). Each of the functions assigned to these institutions refers to one or more dimensions of accountability and can take effect retrospectively and/or prospectively. Classifying functions in this way allows us to assess the strength and primary nature of national accountability mechanisms as described in the VNRs.
In a second step, we transformed these functions into coding items and performed a directed content analysis (Hsieh & Shannon, 2005) of the VNRs submitted between 2016 and 2019 using the qualitative data analysis software Atlas.ti. For those countries that already submitted multiple VNRs, we only retained the latest VNR as basis for our coding. The total number of countries covered by our original dataset is 136. In the interests of fidelity and accuracy, content analysis was performed, wherever possible, on the original-language version of the VNR (English, French, Portuguese, or Spanish). Each VNR was screened for mentions of the functions listed in Figure 1. In case of occurrence, the respective function was coded 1, and otherwise, it was 0 for this country. Throughout the coding process, the coding scheme and instructions were subject to an iterative process of refinement, whereby the initial guidelines were revised multiple times in order to ensure that they were straightforward to understand and implement (The authors will provide the code book and coding instructions upon request). The VNRs were coded by at least two persons separately and checked afterwards for consistency by a third person. When coding discrepancies were observed, a fourth person (who in each case was one of the authors of this study) took an executive decision on the final coding of the data point concerned. In light of the nature of the documents coded, the authors had to take executive decisions regarding ambiguous coding in about 6 per cent of all data points. This is due to the fact that the wording of the VNRs is often rather vague and open to interpretation. Yet the fact that the data was coded by multiple coders along with our own expertise regarding the characteristics of the data makes us confident about the quality of the database ("AccountSDG", which is available as a supplementary file to this article). In a second step, we transformed these functions into coding items and performed a directed content analysis (Hsieh & Shannon, 2005) of the VNRs submitted between 2016 and 2019 using the qualitative data analysis software Atlas.ti. For those countries that already submitted multiple VNRs, we only retained the latest VNR as basis for our coding.

Results: Towards Horizontal Accountability in SDG Implementation
The degree to which parliaments, SAIs, and NHRIs are explicitly recognized as account holders in VNRs differs considerably. As shown in Figure 2, the institution mentioned most frequently in the 136 VNRs that constitute the empirical base of this paper are parliaments (74%), whereas SAIs are only mentioned in 36% of the VNRs, and NHRIs only in 13%. At first glance, these results already confirm the criticism of earlier studies that admonish the insufficient consideration of accountability in the 2030 Agenda and in policies for its implementation.
In light of the nature of the documents coded, the authors had to take executive decisions regarding ambiguous coding in about 6 per cent of all data points. This is due to the fact that the wording of the VNRs is often rather vague and open to interpretation. Yet the fact that the data was coded by multiple coders along with our own expertise regarding the characteristics of the data makes us confident about the quality of the database "(Ac-countSDG", which is available as a supplementary file to this article.

Results: Towards Horizontal Accountability in SDG Implementation
The degree to which parliaments, SAIs, and NHRIs are explicitly recognized as account holders in VNRs differs considerably. As shown in Figure 2, the institution mentioned most frequently in the 136 VNRs that constitute the empirical base of this paper are parliaments (74%), whereas SAIs are only mentioned in 36% of the VNRs, and NHRIs only in 13%. At first glance, these results already confirm the criticism of earlier studies that admonish the insufficient consideration of accountability in the 2030 Agenda and in policies for its implementation. On second glance, the picture is even more sobering. Very often, VNR references to parliaments, SAIs, and NHRIs are vague and superficial-that is, a generic statement is made about them being relevant stakeholders for SDG implementation, but their role or task in the SDG process is not further specified. At other times, their mention refers to a punctual, time-limited activity, but no permanent role or recurring task is assigned. Typical examples of this are instances in which members of the institution in question are On second glance, the picture is even more sobering. Very often, VNR references to parliaments, SAIs, and NHRIs are vague and superficial-that is, a generic statement is made about them being relevant stakeholders for SDG implementation, but their role or task in the SDG process is not further specified. At other times, their mention refers to a punctual, time-limited activity, but no permanent role or recurring task is assigned. Typical examples of this are instances in which members of the institution in question are mentioned as having participated or organized SDG-related awareness-raising, capacitybuilding, or dialogue events.
In the following presentation of findings, we therefore concentrate on mentions of these institutions that exceed such vague, generic references or one-time activities, and thus indicate more specific and permanent accountability arrangements. To provide a rough overview of these arrangements across our sample, Figure 3 lists those VNRs that mentioned either all of the three institutions, different combinations of two institutions, or only one institution in a specified capacity.
As can be seen from Figure 3, only six of the analysed VNRs refer to all three accountability institutions investigated in this paper, thus making it the smallest group in our sample (4.4%). The largest group consists of those VNRs that mention only one institution (64 countries, 47.1%), followed by those who mention two institutions (35; 25.7%). All the while, the number of countries that do not specify a specific role or task in the SDG process for any of the three institutions is relatively high with 31 (22.8%).
In the following, we present our main findings concerning the foreseen exercise of horizontal accountability by parliaments, SAIs, and NHRIs as per their contribution to the information, answerability, and sanction dimension of accountability. The more dimensions a certain instrument or mechanism contributes to, the stronger its accountability effect. Sanctions are the strongest accountability mechanism. We thus structure our presentation of selected findings from stronger to weaker instruments and mechanisms. or only one institution in a specified capacity.
As can be seen from Figure 3, only six of the analysed VNRs refer to all three accountability institutions investigated in this paper, thus making it the smallest group in our sample (4.4%). The largest group consists of those VNRs that mention only one institution (64 countries, 47.1%), followed by those who mention two institutions (35; 25.7%). All the while, the number of countries that do not specify a specific role or task in the SDG process for any of the three institutions is relatively high with 31 (22.8%).
. The strongest accountability mechanism of parliaments in the SDG process is their power to either adopt the budget proposed by the executive or to demand modifications, which provides them with an opportunity to ensure that budget allocations adequately reflect national SDG policy priorities. However, as of yet, only a minority of national parliaments in our sample (19%) seem to routinely use this opportunity. Overall, 19% of the VNRs in our sample contain passages suggesting the national parliament's intention to use the SDGs as an evaluation framework in the budget control process (see Figure 4). The Serbian VNR, for example, lists "providing additional funds for implementation of the SDGs through the procedure for adopting the Budget Law" among the main goals that the National Assembly has set for its activities on implementation of the 2030 Agenda [82]. In a similar vein, the Japanese VNR calls for "active support by parliamentarians [ . . . ] to incorporate the perspective of the SDGs in the national budget" [83]. According to the Sudanese VNR, the National Assembly "undertakes budget reviews and [ . . . ] evaluates the compatibility with allocations and system requirements of the economy taking into account the SDGs principles" [84]. parliaments in our sample (19%) seem to routinely use this opportunity. Overall, 19% of the VNRs in our sample contain passages suggesting the national parliament's intention to use the SDGs as an evaluation framework in the budget control process (see Figure 4). The Serbian VNR, for example, lists "providing additional funds for implementation of the SDGs through the procedure for adopting the Budget Law" among the main goals that the National Assembly has set for its activities on implementation of the 2030 Agenda [82]. In a similar vein, the Japanese VNR calls for "active support by parliamentarians […] to incorporate the perspective of the SDGs in the national budget" [83]. According to the Sudanese VNR, the National Assembly "undertakes budget reviews and […] evaluates the compatibility with allocations and system requirements of the economy taking into account the SDGs principles" [84]. The strongest SDG accountability instrument at the disposal of SAIs are SDG-related audits, which have thus far been conducted in 31% of countries in our sample (see Figure  4). While some of these audits were aimed at evaluating progress on the indicators and targets of specific SDGs, and thus rather retrospective in nature, the majority were audits of government preparedness for SDG implementation conducted with the intention to prospectively provide constructive recommendations at an early stage of SDG implementation. One example is the "Audit of the national implementation of the United Nations Sustainable Development Goals-Agenda 2030" conducted by the German Federal Court of Auditors [85] in 2018. In its concluding message to the Federal Chancellery, which is publicly available on the Internet, the BRH criticised the German Sustainability Strategy as incomplete and not being implemented consistently enough, and formulated concrete advice to the German government concerning the improvement of policy coherence, the development of indicators for progress monitoring, and the establishment of a public sustainability communication strategy.
By aligning their work program with the SDGs, NHRIs can potentially contribute to all three dimensions of accountability, including soft forms of sanctioning by means of negative or positive reporting on retrospective government performance. However, the The strongest SDG accountability instrument at the disposal of SAIs are SDG-related audits, which have thus far been conducted in 31% of countries in our sample (see Figure 4). While some of these audits were aimed at evaluating progress on the indicators and targets of specific SDGs, and thus rather retrospective in nature, the majority were audits of government preparedness for SDG implementation conducted with the intention to prospectively provide constructive recommendations at an early stage of SDG implementation. One example is the "Audit of the national implementation of the United Nations Sustainable Development Goals-Agenda 2030" conducted by the German Federal Court of Auditors [85] in 2018. In its concluding message to the Federal Chancellery, which is publicly available on the Internet, the BRH criticised the German Sustainability Strategy as incomplete and not being implemented consistently enough, and formulated concrete advice to the German government concerning the improvement of policy coherence, the development of indicators for progress monitoring, and the establishment of a public sustainability communication strategy.
By aligning their work program with the SDGs, NHRIs can potentially contribute to all three dimensions of accountability, including soft forms of sanctioning by means of negative or positive reporting on retrospective government performance. However, the intention of doing so was only mentioned by four VNRs (2.9%) in our sample (see Figure 4). The VNR of Australia [86], for example, states that the Australian Human Rights Commission (AHRC) has mapped its work program against the SDGs. In 2018, the AHRC conducted an "Inquiry into the United Nations Sustainable Development Goals" [87]. The related report, which was submitted to the Senate's Foreign Affairs, Defence and Trade References Committee and is publicly available on the Internet, admonished a low level of public awareness about the SDGs and the fact that Australia was only on track to meet approximately 30% of the SDG indicators. Actions recommended to the Australian government included the establishment of a community engagement campaign on the SDGs, the identification of interlinkages between the SDGs and Australia's human rights obligations, as well as the utilization of United Nations human rights reporting mechanisms to commit to, prioritize, and report progress in achieving the SDGs. However, there is evidence that NHRIs are, in reality, much more active in the SDG process than their number of mentions in VNRs would suggest.
The three instruments described above are among the strongest accountability mechanisms, as they potentially contribute all three dimensions of accountability. In the following, we present mechanisms that can be considered as medium strength, as they mainly contribute to the information and answerability dimension of accountability.
As pointed out in Section 2, in order to exercise their oversight functions regarding the SDGs effectively, parliaments need to clearly assign this responsibility within their own structures. As illustrated by Figure 5, in 28 cases, national parliaments have created a new, sometimes even several formal dedicated bodies (such as a standing committee, subcommittee, or special committee) for this purpose. One example is Bangladesh, where "the Speaker of the parliament has decided to form goal specific parliamentary sub-committees to offer regular guidance to the government in implementing SDGs and also oversee the progress" [88]. In eight cases, the mandate of a previously existing committee has been expanded to cover the SDGs. In Japan, for example, the Special Committee on Official Development Assistance and Related Matters has assumed responsibility for the SDGs [83].
anisms, as they potentially contribute all three dimensions of accountability. In the following, we present mechanisms that can be considered as medium strength, as they mainly contribute to the information and answerability dimension of accountability.
As pointed out in Section 2, in order to exercise their oversight functions regarding the SDGs effectively, parliaments need to clearly assign this responsibility within their own structures. As illustrated by Figure 5, in 28 cases, national parliaments have created a new, sometimes even several formal dedicated bodies (such as a standing committee, sub-committee, or special committee) for this purpose. One example is Bangladesh, where "the Speaker of the parliament has decided to form goal specific parliamentary sub-committees to offer regular guidance to the government in implementing SDGs and also oversee the progress" [88]. In eight cases, the mandate of a previously existing committee has been expanded to cover the SDGs. In Japan, for example, the Special Committee on Official Development Assistance and Related Matters has assumed responsibility for the SDGs [83]. While the number of cases that have not established a formal parliamentary SDG committee is quite high (n = 100), in thirteen cases, less formalized groups within parliament-such as a parliamentary SDG network, caucus, or system of focal points-have While the number of cases that have not established a formal parliamentary SDG committee is quite high (n = 100), in thirteen cases, less formalized groups within parliamentsuch as a parliamentary SDG network, caucus, or system of focal points-have formed to promote parliamentary action on the SDGs. In Kenya, for example, a caucus on SDGs and business [89] was formed [90].
When it comes to localizing the 2030 Agenda, the ability of parliamentarians to represent the interests of their constituents and the ability of SAIs and NHRIs to support the process with their expertise are critical for prospective SDG accountability. Evidently, this ability increases the closer these institutions collaborate with the government, get information about SDG policy planning, and receive the opportunity to demand explanation on "why" certain policies are adopted and "how" they shall contribute to successful domestication of the SDGs. Yet, as shown in Figure 6, parliaments, SAIs, and NHRIs are rarely represented in national SDG coordination bodies or councils, which are usually located in the executive branch of power. They also only infrequently participate in permanent working groups or technical committees that collaborate with these coordination bodies. In Kuwait, for example, the General Secretariat of the National Parliament is represented in the National Sustainable Development Committee created for the purpose of structuring and steering the implementation of the 2030 Agenda [91]. Similarly, in Madagascar, a representative of the National Assembly sits on the SDG Steering and Monitoring Committee (Comité d'Orientation et de Suivi, COS), which is presided over by the Prime Minister [92]. Compared to parliaments, SAIs and NHRIs are still seriously underrepresented in these bodies: we only found one VNR that reports the SAI as being represented in the national coordination body [93] or in a technical committee [94]. In only a single case [95] is an NHRI reported as a member of the national SDG coordination body.
(and other governmental and non-governmental actors) on its draft strategy of the National Sustainable Development Strategy [96], although at a rather late stage of the process. Compared to parliaments, NHRIs and SAIs have been involved in these processes to a far lesser extent. Likewise, while twenty VNRs in our sample indicate that parliaments have been involved in the development of national indicator frameworks to measure SDG progress or in the process of national SDG priority setting, this has only been the case for two SAIs and three NHRIs ( Figure 6).  As representatives of the people, another important prospective accountability function of parliaments in the SDG process is to ensure that "road maps" or specific plans for national SDG implementation meet the needs of their constituents. Similarly, the participation of NHRIs in the elaboration of such strategic documents could ensure a human-rights-based approach to SDG implementation. This makes it necessary to seek information and answerability of governments in order to assess whether strategies are in line with peoples' and human rights' interests. However, our findings suggest that, thus far, parliaments are not yet routinely involved in the elaboration of these strategic documents. In Germany, for example, the government consulted representatives of the Bundestag (and other governmental and non-governmental actors) on its draft strategy of the National Sustainable Development Strategy [96], although at a rather late stage of the process. Compared to parliaments, NHRIs and SAIs have been involved in these processes to a far lesser extent. Likewise, while twenty VNRs in our sample indicate that parliaments have been involved in the development of national indicator frameworks to measure SDG progress or in the process of national SDG priority setting, this has only been the case for two SAIs and three NHRIs ( Figure 6).
We close our discussion of findings by presenting the results on two kinds of references to parliaments, SAIs, and NHRIs that can be found considerably more often than references to the functions and instruments described above. At the same time, however, they only reflect a relatively low degree of government willingness to be formally held to account by these institutions in the SDG process.
The exercise of oversight describes the close but still retrospective monitoring of organizational conduct which, in turn, requires the disclosure of information regarding that conduct on behalf of those to be monitored. However, unless organizations or agencies who are supposed to oversee government conduct are equipped with formal mechanisms or instruments to do so, their mere recognition will remain a bland weapon. To fulfil their oversight function, they will largely depend on governments' willingness to release the necessary information but will find it difficult to enforce explanation on the "whys" and "hows" of governmental conduct. Furthermore, evidently, in order to exercise oversight in a meaningful manner, parliamentarians and the staff of SAIs and NHRIs need to be sufficiently informed about the aims of the 2030 Agenda in the first place. Their participation in capacity-building and awareness-raising events thus constitutes an important milestone towards achieving horizontal SDG accountability. As can be seen from Figure 7, only thirteen VNRs (9.6%) ascribe these functions to SAIs, and only five (3.7%) to NHRIs. By contrast, it is predominantly parliaments (45 cases; 33.1%) that are mentioned as the key institutions in charge of exercising oversight and scrutiny in the SDG process. This tendency also reflected the number of mentions of capacity-building and awareness-raising events which appear to have primarily organized in the realm of parliaments. milestone towards achieving horizontal SDG accountability. As can be seen from Figure  7, only thirteen VNRs (9.6%) ascribe these functions to SAIs, and only five (3.7%) to NHRIs. By contrast, it is predominantly parliaments (45 cases; 33.1%) that are mentioned as the key institutions in charge of exercising oversight and scrutiny in the SDG process. This tendency also reflected the number of mentions of capacity-building and awarenessraising events which appear to have primarily organized in the realm of parliaments.

Conclusions
This analysis set out to assess if and how national governments acted on the call to establish national horizontal accountability mechanisms for SDG implementation following the adoption of the 2030 Agenda. Such a stocktaking of emerging national accountability arrangements is both timely and necessary in order to obtain a baseline for future hypothesis-testing analyses regarding the effectiveness of different institutional set-ups in achieving integrated and inclusive SDG implementation. In order to create such a baseline, we conducted a systematic analysis of the arrangements that states have adopted in order to enable parliaments, SAIs, and NHRIs (a) to oversee the national government's efforts to implement the SDGs; (b) to assess performance in achieving the SDGs; and (c) to facilitate learning and adaption of SDG policies. Empirically, our analysis was based on the content-directed coding of information contained in the 2016-2019 VNRs, the results of which we compiled in an original dataset on horizontal SDG accountability (Ac-countSDGs). (The dataset is available from the authors upon request.) Our results can be used for generating hypotheses for future analyses to build on.

Conclusions
This analysis set out to assess if and how national governments acted on the call to establish national horizontal accountability mechanisms for SDG implementation following the adoption of the 2030 Agenda. Such a stocktaking of emerging national accountability arrangements is both timely and necessary in order to obtain a baseline for future hypothesis-testing analyses regarding the effectiveness of different institutional set-ups in achieving integrated and inclusive SDG implementation. In order to create such a baseline, we conducted a systematic analysis of the arrangements that states have adopted in order to enable parliaments, SAIs, and NHRIs (a) to oversee the national government's efforts to implement the SDGs; (b) to assess performance in achieving the SDGs; and (c) to facilitate learning and adaption of SDG policies. Empirically, our analysis was based on the contentdirected coding of information contained in the 2016-2019 VNRs, the results of which we compiled in an original dataset on horizontal SDG accountability (AccountSDGs) (The dataset is available from the authors upon request). Our results can be used for generating hypotheses for future analyses to build on.
In sum, the descriptive analysis revealed a fair but still very limited degree of horizontal accountability in national governance mechanisms for SDG implementation. Parliaments, SAIs, and NHRIs engage in both retrospective and prospective accountability-taking, but to different degrees. While the role of NHRI has remained very limited, the role of SAI is better recognized but largely confined to their original task of conducting either retrospective-oriented performance audits or prospective-oriented SDG preparedness audits. By comparison, parliaments stand out in the sense that the number of VNRs who recognizes their role is considerably bigger and the variety of functions ascribed to them is much broader. Such functions range from participation in national SDG coordination bodies or technical working groups, and ensuring the alignment of the national budget with the SDGs, to committee oversight through dedicated parliamentary SDG committees.
Our results support the earlier criticism concerning the lack of strong accountability in the SDG process at the national level. We have shown that serious formal commitment to accountability in SDGs implementation has clearly been a choice of individual governments rather than a standard in national SDG implementation across countries. Furthermore, among the VNRs that recognize parliaments, SAIs, and NHRIs as account holders, references to weaker accountability mechanisms that merely contribute to the information dimension of accountability are far more frequently found than references to stronger mechanisms that also contribute to the answerability and sanction dimension of accountability. In particular, the "heavier weapons [of accountability]" [78] in the form of formal sanctions are notoriously absent. Mechanisms that could potentially contribute towards the sanction dimension of accountability rather entail soft sanctions in the form of reputational gain or loss. These findings resonate with the argumentation of scholars who see facilitative accountability arrangements, rather than punitive accountability, as more appropriate in the context of the 2030 Agenda. Karlsson-Vinkhuyzen et al. [19], for example, argue that facilitative accountability arrangements hold the potential to contribute to the development of a cooperative, reflective, and trust-based relationship between account holders and account givers through learning, critical reflection, and the identification of lessons that can lead to policy change. At the same time, one must not underestimate that there are areas which require "heavier weapons"-for instance, interests of veto-players like business communities or emerging powers which oppose fossil free growth might hinder SDG implementation.
However, it should not be concluded from the above that existing mechanisms for horizontal SDG accountability are toothless. Specifically, parliaments can potentially sanction governments' SDG policies through budget (re-)allocation. SAI audits and NHRI reports, in turn, can have important government advisory functions. However, the ability of these mechanisms to generate the desired accountability effects is highly dependent on the country context. For example, a parliament might be granted a prominent role in SDG budget oversight on the paper of a VNR in a specific country but, de facto, be a weak institution, which has very limited competences in national policy processes. This is a likely scenario in autocratic states. For another example, SAI audits and NHRI reports can provide a strong basis for social accountability where civil society is strong and non-state actors use their results to blame and shame the government. This, in turn, would be a less likely scenario in the context of an autocratic regime.
Concluding, it is important to note that, being based on a document analysis, our study is subject to a number of limitations that provide opportunities for future research. First and foremost, we cannot assess the degree to which statements made in the VNRs reflect political practice. On the one hand, a mechanism or instrument that is mentioned in a VNR may not be properly functioning in reality-for example, a parliamentary SDG committee has been formally established but rarely ever convenes in practice. On the other hand, the institutions under investigation in our paper may be actively engaged in the SDG process in practice even though their activities are not mentioned in the VNR. This would signal a limited willingness of the government to officially recognize these institutions as account holders. Case studies including field research and expert interviews will be needed to fill these research gaps concerning practical implementation of SDG accountability. Furthermore, while contextualizing emerging SDG accountability arrangements with political-institutional factors such as regime type or state capacity is beyond the scope of this study, the AccountSDGs dataset facilitates the conduction of such analyses. In particular, a hypothesis-testing analysis could inquire why governments opt for more or less strict accountability provisions for SDG implementation. On the one hand, it is not necessarily a "bad sign" when governments do not create strong accountability mechanisms for SDG implementation. Instead, they might-as is the case for strong democracies-already have accountability processes in place. Building additional accountability mechanisms would be an extra cost for them. On the other hand, governments of states with low capacities might not hesitate to promise strong accountability mechanisms [97], which will control their record in SDG implementation. They have nothing to fear because it is unlikely to achieve strong horizontal accountability where state capacity is low. In the longer term, the dataset can serve as a baseline for research exploring the relationship between different institutional set-ups and integrated SDG achievement.