Assessing intelligence oversight: the case of Sweden

ABSTRACT The study of intelligence oversight captures the inherently political nature of secret intelligence. However, many studies of intelligence oversight adopt rather instrumentalist views that omit important political aspects of the policy process. Typically, these studies focus on obstacles to effective oversight. This article discusses how the effectiveness of oversight can be assessed by applying broad evaluative categories that contain programmatic, process-related, political, and durability dimensions. Empirically, the study probes the case of Sweden as an illustration. Swedish oversight arrangements have on balance been successful in some dimensions, particularly the programmatic dimension, which arguably also contributed to its relative longevity.


Introduction
By default, the sub-field of intelligence oversight studies acknowledges the political nature of secret intelligence.The political contexts in the various policy arrangements are paramount in any understanding of intelligence as a phenomenon.This is arguably why intelligence oversight studies gained momentum after the Edward Snowden leaks. 1 The bulk surveillance programmes revealed by Snowden may have been more or less effective from an operational point of view, but they seem to have been politically self-defeating. 2The ultimate purpose of security and intelligence agencies in western societies is to uphold democracy and the rule of law, which begs the question of why governments, including their intelligence agencies, time and again attempt to push the borders of legitimacy.Most western democracies have put oversight arrangements in place to tame the power of secrecy.These play an important watchdog role as an intermediary between entrenched government agencies and an open society.However, can we be confident that such arrangements are working?This article discusses how the effectiveness of intelligence oversight can be assessed by probing a framework of broad evaluative categories that contain programmatic, process-related, political, and durability dimensions based on the case of Sweden.The framework originates in scholarship on policy success and failure, and includes a spectrum of outcomes from success to failure in each dimension.
Intelligence oversight arrangements are often studied in terms of their capacity as scrutinisers of the effectiveness, efficiency and propriety of intelligence agencies. 3Occasionally, the effectiveness of the arrangements as such are also studied.On the US Congress, Amy Zegart contends that 'weak intelligence oversight is a reasonable and useful point of departure'. 4 Some, like Peter Gill, list elements of what would qualify as effective oversight but put analytical weight on obstacles to effectiveness. 5 Njord Wegge and Thorsten Wetzling provide a percipient discussion of the increasingly legalistic nature of intelligence oversight and the challenges overseers face in enacting the laws, let alone those faced by the research community in gaining insight into their functioning: 'Researchers outside the ring of secrecy are often left with little choice but to draw inferences on CONTACT Dan Hansén dan.hansen@fhs.separtial and sometimes politicised information'. 6Scrutinising intelligence agencies, which is the role and function of overseers, is therefore difficult, while academically assessing the effectiveness of their efforts is just as daunting.
The intelligence studies research community is not unique in finding it challenging to assess the effectiveness of various policy arrangements.Policy scholars have been grappling with similar issues for decades, albeit seemingly without paying much attention to policies related to intelligence affairs.Conversely, references to policy scholars are few and far between in intelligence research.This is probably one reason for the predominantly instrumentalist analyses, which depart from given policy instruments and discuss the difficulties overseers have in using them.There is a certain negativity bias built into the research designs.
In order to discuss intelligence oversight effectiveness, this article takes stock of the scholarly debate on policy success and failure.This debate, in turn, developed out of policy evaluation scholarship dominated by a quest to improve policy through evidence-based analyses. 7Pioneered by Bovens and 't Hart's Understanding Policy Fiascoes, 8 the policy success/failure debate involves more dimensions of policy evaluation than the traditional instrumental understanding, and notably contains a political dimension.In essence, what might be a policy failure in terms of immediate service delivery could harbour political and strategic gains for the future, and vice versa.This article aspires to contribute to the flourishing body of intelligence oversight analysis by probing a multidimensional framework.As is further elaborated below, the analysis is explicitly sensitive to nuances of success.
Policymaking on Swedish intelligence oversight serves as an illustrative case study here since it is the case best known to the author.It is therefore worth mentioning that the point of departure is not that the Swedish case necessarily merits the label of policy success; in fact, the success level varies between dimensions, and dilemmas observed elsewhere are certainly also present to varying degrees in Sweden.There is a heuristic value in assessing policy arrangements with eyes sensitive to what works.The dimensions are therefore broad enough to capture variations in effectiveness.

Looking on the bright side of intelligence policymaking
The extent to which public policies deliver is a pivotal concern for policymakers and academics alike.Much scholarly attention has been paid to policies that have failed to live up to purported goals, such as in the classic works by Pressman and Wildavsky,and Hall. 9 In aspiring to improve the policymaking process, the academic field has developed a culture of seeking to understand and explain failure.More recently, concern that this type of scholarly endeavour has contributed to widespread public mistrust of government has created a counterwave of scholarship preoccupied with mapping and understanding policy success. 10This is to some extent a new deal for policy scholars, at least when success parameters become selection criteria for case analysis.In this case success is not a selection criterion, but the analytical framework at least dampens the effects of potential negativity bias.
Success and failure are polar opposites on a continuum when assessing policy effectiveness.What constitutes success or failure is, of course, inherently contestable.Even if consensus can be reached on a good match between purported policy goals and the ensuing results of the policy intervention, those in a position to have formulated the goals will be more likely to proclaim success. 11While admitting a slight exaggeration in paraphrasing Clausewitz, Bovens et al. assert that evaluating policy 'is nothing but the continuation of politics by other means'. 12The analyst must therefore proceed with caution in inventing categories and criteria for evaluation.However, rather than resolving political disagreements on evaluative criteria, which must anyway be a task of the political process, the analyst, according to Majone, would be better 'facilitat[ing] a wide-ranging dialogue among advocates of different criteria'. 13o this end, McConnell has made an influential contribution by categorising relatively open dimensions of policymaking that can succeed or fail, or something in between. 14For each dimension, the framework tries to establish what would qualify as success, and delineates a spectrum of decreasing levels of success until failure, which is then the mirror image of success.The grey areas inbetween are resilient, conflicted and precarious success. 15The success spectrum certainly reflects a nominal scale intended to facilitate analyses based on judgment, where the nuances serve the purpose of avoiding assessing anything short of a clear success as constituting a failure.This undoubtedly allows for more nuanced analysis and discussion of the effectiveness of public policymaking.The dimensions in question have been refined over the past decade, and this article sets out from the dimensions put forth by Compton and ´t Hart, such as programmatic, process, political and endurance dimensions. 16hese dimensions were constructed to capture policy success and to relate empirical findings to their fit with success criteria.These dimensions are further explored below in the realm of secret intelligence policymaking, or more precisely the policy subfield of intelligence oversight.Here it is however relevant to point out that the general idea with the dimensions is to enable assessment of different areas, certainly stressing that the political is one of them, and thereby showcasing that policy interventions typically succeed to varying degrees simultaneously, depending on the dimension.Moreover, the relatively broad categories lend themselves more easily to comparison between polities, compared for example to the evaluation of singular policy instruments (the functioning of which may pertain to contextual features in a polity).Within the study of intelligence oversight, the observation that 'comparative research must amount to more than the juxtaposition of a range of single-country studies', 17 may be indicative of the absence of an appropriate framework.The proof of the pudding lies in the extent to which the case study presented in this article can be seen as an implicit comparison. 18s a corollary to public administration scholarship's interest in policy failure, the academic field of intelligence studies has developed a strong current around the theme of intelligence failures, which seeks to understand, explain and learn from them.The dearth of cross-referencing between the two research areas could indicate that there are particularities with secret intelligence (as a policy subsystem) that call for particular consideration when analysing effectiveness.However, the heuristic value of keeping an eye on what works is arguably also valid in the intelligence studies subfield of oversight, as it slants towards a focus on what does not. 19he secretive nature of intelligence work and politics is arguably what sets it apart from other issue areas.Policy analysts typically engage in high visibility policymaking, such as on healthcare, education or environmental policies -areas that large parts of the population, researchers included, experience first-hand.The abundance of accessible empirics has provided much material for theorising.The toolbox for assessing policy effectiveness owes a lot to policies that are public through and through.For researchers interested in analysing the effectiveness of policies on secret intelligence, the scope of possible cases is more limited.Peter Gill depicts the intelligence governance process as comprising control (direction), prior authorisation (supervising) and oversight (scrutinising), arguing that scholars from outside (of the intelligence community) have their best shot at accessing empirics in the latter oversight phase of the process. 20From a governance perspective, the scrutinising of policy arrangements that include oversight bodies is important in its own right.Governments bestow their secret intelligence organisations with extraordinary mandates, essentially to carry out tasks that would in other circumstances qualify as unlawful.Oversight is hence a crucial policy arrangement that scrutinises the deeds of intelligence organisations at the behest of open society.

Oversight: programmatic, process, political and endurance effectiveness
Any attempt to delineate what intelligence oversight is and what it is not will inevitably also be interspersed with contextual features, most notably regarding the nature of the political system.Just as the US notion relies on the Framers' ideas on checks and balances, 21 the Swedish notion falls back on control and supervision ('oversight' does not translate easily into Swedish). 22Such cultural aspects are integral to expectations of what oversight ought to achieve.To exemplify, a fairly common view of intelligence oversight is that, '[i]t suggests something looser than control in the sense of day-to-day management of the operations of the agencies'. 23This seems to hold for systems where the legislature is the premier overseer, but arguably less so where expert bodies within the executive branch perform the role, at least spot-checking on day-to-day activities.Not surprisingly, therefore, the more oversight arrangements are subject to comparison, the broader the scope of oversight tends to be.Oversight can hence be understood as something that happens before (i.e., authorising), during, or after an intelligence operation. 24It can encompass scrutiny by the executive and/ or the legislature, as well as the actual legislating. 25Emphasis can be put on effectiveness, efficiency (cost-benefits), or propriety (the quality of the laws and lawfulness of the agencies).This is what Gill refers to as the trilemma of scrutiny, since it is difficult to fulfil more than two of the above at a time without renouncing the third. 26This is not an exhaustive account of definitional variations.The point here is that the catch-all nature of the concept is useful for the purpose of this study.The many ways in which intelligence oversight have been constructed informs the policy dimensions of what could qualify as more or less successful.

Programmatic effectiveness
This dimension straightforwardly examines the relation between policy goals, the theory of change that underpins them, and the policy instruments selected to achieve the goals. 27Policy arrangements can by design have more or less auspicious prospects of success.In the realm of secret intelligence oversight, the institutional set-up is frequently implied to be pivotal to achieving efficiency and effectiveness. 28olicy goals are often vague, and intelligence oversight is no exception.Across polities they revolve around the general idea of safeguarding democratic scrutiny of agencies with extraordinary mandates.The variation is rather to be found in institutional design and the selected policy instruments.Since these are contextually constructed, the task is to analyse their viability in any given polity.To be successful, there must be tangible evidence that what was once problematic is now considerably less so (see Table 1).Governments have a range of policy instruments to choose from when designing and redesigning oversight arrangements.In the mid −1970s, the introduction of oversight arrangements proliferated, especially in western democracies, 29 which indicates that governments are impressionable and learn from each other in this policy domain.Intelligence oversight can be kept together in a designated legislative committee or in a committee external to the legislature, or be dispersed in various bodies with varying relations to the legislature and the executive.Expertise in intelligence work can be mixed with expertise in law and technology, and with lay persons' perspectives.Mandates can be specific and narrow or wide-ranging and all-encompassing.Do intelligence agencies need to be better at abiding by the law, or is the legitimacy of their actions being questioned?Arguably, the question somehow insinuates a false dichotomy, since consistently ignoring the law decreases legitimacy.However, keeping the oversight capacity close to the legislature is often seen as enhancing legitimacy, 30 whereas oversight by the executive, be it a ministry or an external authority, restrains the illegality of agencies. 31Intentionally or not, different designs convey different problem representations.

Process effectiveness
The effectiveness of a policy programme or arrangement must take account of the design and procedural organisation of decision-making and implementation.To be successful an oversight arrangement must confer legitimacy on the policy.The design process needs to carefully take into consideration the mix of policy instruments that would be contextually appropriate, to be perceived as correct and fair (see Table 2).Different stakeholders should ideally be given the opportunity to influence the policymaking process before choices are made.Process effectiveness also implies striking a balance between acceptable costs and adequate levels of funding, and agreeing on realistic timelines. 33arliamentary ownership has in a comparative study by Born & Wetzling emerged 'as a decisive condition for the overall success of intelligence control'. 34The argument rests on the premise of independence of the legislature vis-à-vis the executive (including the agencies).It also presupposes an engagement and commitment by legislators, which is harder to achieve (see next section).Rather than independence, which in parliamentary systems are characterised by shades of grey, the nature of the relation between overseers and overseen may be telling as to the process effectiveness.According to Wegge, security service staff in north-western Europe caution that overseers, especially if they are parliamentarians aligned with the incumbent executive, sometimes use secret information for their own political purposes rather than to safeguard national security. 35In general, there is an information asymmetry between the secret intelligence agencies and their overseers, since the latter rely on the former to obtain information.As long as the attitude among the secret services is one of mistrust of the overseers' precautions taken with secret information, they are likely to veer on the minimalistic side of cooperation.Undoubtedly, such a relationship seriously hampers the effectiveness of the oversight arrangement.Oversight arrangements have become commonplace in recent decades, and part of the normal in secret intelligence agencies in democracies.The two parties have developed different relations depending on the context.Some intelligence agencies have learned that keeping the overseers happy can turn them into patrons, which is particularly useful if the overseers have any influence over budgets. 36From an effectiveness perspective, there is obviously a risk that the overseers might be co-opted by the agencies in 'the subtle process by which watchdogs are transformed into lapdogs'. 37n order to be effective, overseers obviously need resources in terms of both investigative powers and mandates.For a mandate to qualify as successful, it must allow the overseer to dig as deeply as it finds pertinent.At the very minimum, overseers must not be able to complain about weak mandates.However, a strong mandate does not get very far if its owners do not know where to look or what to ask.Some level of expertise is therefore necessary for oversight bodies to be effective, but also sufficient time to absorb the culture and probe relevant factors. 38In some countries oversight is chiefly conducted by the legislature, in the case of the US Congress by subcommittees in the Senate and the House of Representatives.Amy Zegart has reported on the difficulties in attaining an acceptable level of expertise, especially in the House of Representatives which has biennial elections. 39A key to developing the expertise required to ask relevant questions of the right people and for the right documents is a relatively slow turnover of staff and organised handovers.Perhaps the major drawback with having inexperienced overseers is that the overseen agencies can more easily keep up a lip-serving facade, behind which nothing really changes, or what Rittberger and Goetz call an organised hypocrisy and Hijzen refers to as a ritual dance. 40

Political effectiveness
For a policy arrangement to be politically successful, it needs to garner a broad coalition of supporters among policymakers and representatives in authority.It needs to ramp up the reputation of its initiators, or at least to be largely unchallenged (see Table 3). 42ustering legislative commitment to secret intelligence work presents challenges as incentives are largely absent.The political payoff is too small and the issue area complex and time-consuming to penetrate. 43A political reluctance to engage with secret intelligence affairs clearly detracts from the effectiveness of the policy area, not least in terms of its legitimacy.Politicians, Loch Johnson observes, however tend to devote much attention to accountability and turn to firefighting mode when intelligence failures break open or when scandals erupt, only to gradually fall back to a lowintensity patrolling once the intelligence shock has lost topicality. 44Success in this dimension would entail visibility on the political agenda and the political will to engage with the complexities of secret intelligence, thereby heightening the political status of secret intelligence as a tool in the service of democracy.

Endurance effectiveness
The longevity of a policy arrangement is an essential but insufficient indicator of success.As social and politico-administrative environments gradually evolve, arrangements must also adapt and learn in order to deliver on the initially purported core objective: to excel at dynamic conservatism. 46There is nothing about this policy dimension that is particular to secret intelligence.All organisations and institutions periodically face surprises and critical events, and at times must endure heavy criticism for not having been vigilant enough.To be successful in this dimension, actors within the oversight arrangement will need to perform pre-emptive adaptation in order to absorb new circumstances and to protect the core institutions (see Table 4).If the core objective of the arrangement is to increase the legitimacy of secret intelligence, adaptations should ideally enhance legitimacy.A pertinent question, however, is how long a rope is in this context; what would qualify as policy longevity?Typically, a full policy cycle, including agenda-setting, decision-making, implementation, and evaluation based on feed-back loops, takes about a decade (or more). 47Success criteria can therefore relate to the policy cycle.By the end of it, even initial core institutions may be in need of reassessment.

Intelligence oversight in Sweden: a success story?
In the case of Sweden, secret intelligence encompasses both defence and law enforcement agencies, with foreign as well as domestic remits.The foreign/defence side comprises four intelligence agencies, two of which -the Swedish Armed Forces Intelligence and Security Agency (MUST) and the National Defence Radio Establishment (NDRE) -actively collect intelligence.The other two, the Swedish Defence Research Agency and the Defence Materiel, are passive consumers of intelligence, and therefore of lesser interest to overseers and in terms of this paper.The defence intelligence agencies are overseen by the Foreign Intelligence Inspectorate (FII).On the domestic/law enforcement side, the Security Service, the National Police, the Public Prosecutor, Swedish Customs and the Swedish Economic Crime Authority are overseen by the Swedish Commission on Security and Integrity Protection.The Swedish National Audit Organisation scrutinises all authorities, including the defence and security sector, with a focus on efficiency.The Swedish Authority for Privacy Protection supervises authorities' adherence to rules on personal data.The focus of this article is on oversight arrangements in relation to foreign/defence intelligence.

Analysis of the programmatic dimension
By the turn of the millennium, when the Cold War seemed far away and international terrorism and other non-military menaces had largely replaced military threats, at least from the Swedish horizon, the time was ripe for a review of the tasks of the intelligence agencies and their ways of working.The wider security agenda that had been generally embraced called for a shift in the centre of gravity from traditional military warning of an armed invasion, and from comprehensive tactical and operative military forewarning, to strategic, non-military intelligence, as well as more selective monitoring of military operations in the near abroad and a greater focus on broader and longer term security policy analysis. 49Intelligence agencies were therefore supposed to be alert to international terrorism and international organised crime, as well as the potential risks related to migration flows and climate change.To this end, the Government proposed replacing the concept of military intelligence with the broader concept of defence intelligence.Essentially, all foreign threats should be assessed and analysed by the intelligence agencies in support of Swedish foreign, security and defence policies, and in order to identify external military threats to the country.Activities should also assist Sweden's participation in international security cooperation. 50This transition happened without much fanfare, arguably because it made sense to make use of available resources more effectively and adapt to the new realities.

Controversial adaptations
The intelligence oversight arrangement remained unchanged until the second half of the first decade of the new century, when a new government proposed changes to the intelligence legislation.In June 2007, the government proposed deleting the word military from section 2 of the Swedish Law (2000:130) on defence intelligence operations so that intelligence agencies would now identify external threats to the country as opposed to external military threats, which parliament found unproblematic.The government also drafted a new Signals Intelligence Act, empowering the National Defence Radio Establishment (NDRE) to intercept data traffic in fibre optic cables.Previously, they could only listen to signals in the ether, which made their job less and less effective since so much traffic had moved to cables. 51In sharp contrast to the enlargement of the remit of intelligence agencies, considerable controversy surrounded extending the means by which the signals intelligence agency could gather information.Wire-tapping cables obviously raised the possibility of mass surveillance of data traffic, such as e-mail, which caused considerable public discontent.Even though the government had a parliamentary majority, parliament put the proposed Signals Intelligence Act on hold for twelve months.When a slightly amended version of the Act was tabled for parliamentary discussion and a vote in June 2008 it passed; however parliament stated that the law should be further amended.The government was asked to specify how the law would act to protect the privacy of individuals, and to confirm that the authorising body for signals intelligence operations would be organisationally separate from the oversight body and have similar procedures to a court. 52The Signals Intelligence Act became law on 1 January 2009.The tapping of cables was supposed to start on 1 October 2009 but was delayed until 1 December due to the 'extensive and meticulous treatment of the Parliament's suggested amendment to the Signals Intelligence Act'. 53Even though the basic structures of the intelligence oversight body were more or less settled by the initial 2007 Government proposition, the ensuing political process emphasized the importance of its free-standing role and strong powers.

Historical reasons for controversy
The Foreign Intelligence Inspectorate replaced its predecessor, the Intelligence Committee (Underrättelsenämnden), an organisation within the Swedish Armed Forces which had been established in 1976 following a scandal that emerged in 1973. 54Two journalists had published a story about the military intelligence organisation's special operations branch, the IB (Information Bureau).Part of their story revolved around the domestic activities of IB, and that part became highly politicised.The article suggested that military intelligence had infiltrated political parties on the farleft and other lawful organisations, and had kept files on 20,000 individuals. 55In particular, the purported links between the IB and the Social Democratic Party, which by 1973 had ruled Sweden for 41 years, fuelled its politicisation.The so-called IB affair gave rise to an investigation by the Parliament Defence Committee, a government investigation and an investigation by the Justice Ombudsman. 56The Intelligence Committee was thus a product of these investigations.The Justice Ombudsman confirmed that the IB had spied on people living in Sweden, often by infiltrating political organisations, but found that this was not directed by the Social Democratic Party. 57owever, the IB affair lingered in Swedish political life, and official reports were published in 1976,  1979, 1980, 1998 and 2002.The affair became an issue in the general election campaigns of 1973,  1988, 1991 and 1998, and has been the subject of numerous books and documentaries. 58A 2002 report established that the IB and the Social Democratic Party were to some extent entangled, and that the party benefited from this entanglement. 59It is therefore important to keep in mind that the 1973 IB affair was not just a historical event when the revamped signals intelligence legislation became politicised at the end of the 2000s.The process of amendment clarified the objectives of signals intelligence -as a tool for the acquisition of foreign intelligence -and specified the procedure for authorisation of operations and the destruction of information. 60ence, the current oversight arrangement with regard to defence intelligence stems from December 2009, when the Signals Intelligence Act took practical effect.The arrangement consists of an authority, the Foreign Intelligence Inspectorate (FII), 61 but also of a privacy protection representative (integritetsskyddsombud) based at the Foreign Intelligence Court (Försvarsunderrättelsedomstolen), an authorising body, and a Privacy Protection Council located at the Signals Intelligence Agency (NDRE), appointed by the government with a mandate to observe the measures taken to ensure the integrity of individuals involved in signals intelligence operations.Thus, apart from the external authority, there are also elements of what Born & Wetzel call embedded human rights in intelligence affairs. 62

Programmatic assessment
The fact that FII was made an independent authority outside the armed forces is undeniably a step in the right direction for obtaining independent scrutiny.The creation of an authorising Foreign Intelligence Court, and the division of labour between it and the FII are also important for avoiding compromising the integrity of the scrutiniser.The iteration process leading up to the amended legal framework at the end of the decade almost exclusively considered safeguarding the protection of individual privacy from potentially intrusive intelligence agencies, implying that the theory of change was based on legitimacy.Most of the legitimisation however relates to the legal framework, which in a way boxes in what oversight should mean.On balance, however, given the period of amendments, Swedish intelligence oversight arrangements qualify as a resilient success on this dimension.

The intelligence oversight arrangement
Signals intelligence gathering in particular became more regulated under the new legislation and new authorities, so the oversight arrangement also needed to be revamped, with a core objective to ensure that the actions of the defence intelligence agencies were lawful.However, their mandate also included ensuring that defence intelligence was conducted in accordance with directives, to scrutinise signals and human intelligence work carried out by special means, to scrutinise other means and methods for collecting information, and to review the principles for recruiting and training personnel.The FII must also oversee MUST's and the NDRE's treatment of personal data, respond to allegations from the general public that intelligence agencies have unduly intercepted their messages (i.e., ex post facto reviews), and control the agencies' usage of qualified protected identities.In addition, FII has the exclusive right to give the NDRE access to signals bearers when authorised by the Foreign Intelligence Court and a mandate to discontinue information gathering that is not authorised by this court. 63he FII comprises up to seven committee members, of whom the president and vice president are experienced judges and the rest are nominated by the parliamentary party groupings, but appointed by the government for a term of at least four years.(Most commonly, they are former parliamentarians.)The authority also has a secretariat of five to six people working full time, while the committee convenes when called on by its chair or at the request of one of the defence intelligence agencies.The FII conducts some 20 to 30 investigations per year, and is free to decide for itself what to investigate.The Swedish intelligence oversight arrangement is more ambitious than the one in place before the 2008 regulations and the 2009 amendments.In addition to increased resources intended to reflect the increased mandate bestowed on the Signals Intelligence Agency, it was also given a more independent role in relation to the agencies it is supposed to oversee and enhanced investigative powers.
The FII is not an isolated entity.The ordinances governing its functioning state that it can call on experts in its deliberations when needed. 64In its annual reports, it regularly chronicles having paid study visits to peer organisations abroad and in Sweden, especially to peers overseeing the domestic arena. 65

Domestic reassurance
The Swedish National Audit Office (NAO) investigated the FII in 2015 and, from an effectiveness perspective, its report was rather comforting.Among other things, it observed that the FII has been well funded from the start.In fact, the FII had not spent its entire budget in a single year since its inception 66 ; and that holds up until fiscal year (FY) 2021. 67The Swedish NAO also found that the FII had been fulfilling its obligations according to the law and ordinances. 68he Swedish NAO believes that the agencies under the FII's scrutiny are attentive to the latter's opinions and recommendations. 69In the period 2010-21, the FII conducted 303 examinations, 29 of which led to opinions and recommendations (16 to the NDRE and 13 to MUST).In its annual reports, the FII writes of having a good rapport with the agencies it scrutinises, based on dialogue and a will to guide them on legally difficult balancing acts. 70To the extent that trends can be identified with such low numbers, the most recent six-year period has seen a decrease in opinions, as about 6 per cent of all examinations led to the issuance of opinions compared to about 11 per cent in the first six years. 71This might be indicative of intelligence agencies learning from scrutiny, and in that sense also becoming more effective in their work.

An international reality check
The NGO Centrum för rättvisa (Center for Justice) first reported the Signals Intelligence Act to the European Court of Human Rights in 2008.It took until May 2021 for the Grand Chamber of the ECHR to deliver its verdict, which found the state of Sweden and the Signals Intelligence Act in breach of Article 8 of the European Convention: the right to respect for private and family life, home and correspondence.In large part, the ECHR found the legislation to be of high quality, but the arrangement nonetheless suffered from three shortcomings: the absence of a clear rule on destroying intercepted material which [does] not contain personal data; the absence of a requirement in the Signals Intelligence Act or other relevant legislation that, when deciding to transmit intelligence material to foreign partners, consideration [is] given to the privacy interests of individuals; and the absence of an effective ex post facto review. 72e last point constitutes unadorned disapproval of Swedish intelligence oversight arrangements as far as ex post facto reviews are concerned.This is what the FII does at the request of individuals who want to know whether their communications have been intercepted by the NDRE.In the period 2010-21 such a request occurred on 171 occasions, and in none of them did the FII find any undue interception by the NDRE. 73The problem that the ECHR pointed out is that it cannot be ruled out that the FII might have incentives not to find undue interceptions, since this would mean that their oversight of the NDRE had at some point been at fault.The dual roles are in principle incompatible. 74he Government communicated an action plan to the ECHR in May 2022, according to which any shortcomings would be addressed in a forthcoming government inquiry, which would be a fundamental review of the Signals Intelligence Act.The same went for the absence of an effective ex post facto review, even though the government referred to the independence enjoyed by the Swedish authorities which meant that it was up to the FII to interpret laws and ordinances.The government announced the remit of the inquiry in June 2022.It will make recommendations on the issues addressed by the ECHR in April 2023 and complete its work in April 2024. 75he FII does not mention the ECHR ruling in its 2021 annual report.It does, however, discuss the difficulties of measuring the effectiveness of secret intelligence oversight.It maintained that it is not only through its opinions and recommendations that it exerts influence -the fact that the intelligence agencies know that they are being regularly scrutinised also has an impact. 76arliament discusses signals intelligence in early March each year, based on the report that the FII submits to the MoD in February, which is then passed on to the Parliamentary Defence Committee.In 2022, this discussion took place on 2 March, only days after the Russian military invaded Ukraine.Only one representative of the Moderate Party (opposition, right wing) and the Social Democratic Party (minority government) respectively took part in the debate, and the discussion revolved around the upcoming government inquiry into the Signals Intelligence Act.The Moderate Party member lamented that it had not started years ago, since he regarded it as paramount to increase the capacity of signals intelligence.He also made a case for more transparency when it comes to international cooperation among intelligence agencies, thereby touching on the second point of criticism by the ECHR.The Social Democrat parliamentarian, a former president of the FII, said that the inquiry would start in due course.Both professed the importance of strong and capable intelligence agencies, and praised the functioning of the highly important oversight arrangements.77

Process assessment
The process leading up to the legislative framework and institutional set-up seems to have conferred legitimacy on the arrangement.This was arguably paramount in order to satisfy a sense of fairness and of the correctness of the arrangement, given the nature of the debate at the time.The 2015 (domestic) reality check by the Swedish NAO consolidated intentions of a well-funded and authoritative FII that actually works.The mandates are wide and the expertise is developing.In light of the continuously underutilised budget, one however wonders why they have not assumed a larger assignment.Is there nothing more to scrutinise?The financial suit was adjusted after the Swedish NAO's investigation, but is still oversized, which allows for the interpretation that a balance has been struck between acceptable costs and appropriate funding, including some growth potential.When looking only at the domestic arena, the process dimension is a clear success.Obviously, the verdict from the ECHR does not speak in favour of a successful assessment of the legislative framework, even though the ruling also said that for the most part it was sound.From the perspective of the international herald of guiding principles on legal security, the verdict is arguably somewhere between resilient and conflicted success.

Political engagement and complacency
Compared to many other western democracies, the Swedish Parliament plays a fairly unobtrusive role as an overseer of secret intelligence.Oversight authorities are responsible to the government, and for defence intelligence to the Ministry of Defence.This means that oversight is an arrangement within the executive branch.Naturally, the government is responsible to parliament so the Ministry of Defence submits a report to parliament on an annual basis on the oversight of defence intelligence, which is debated first in the Defence Committee and then in a plenary session.When the Signals Intelligence Act was amended in 2008-2009, however, parliament asked the government for annual reports only on oversight of the Signals Intelligence Act.For this reason, the annual reports and the parliamentary discussion only concern that legislation; not how the intelligence agencies operate more generally.Of the FII's 303 investigations in the period 2010-21, 82 had a bearing on the Signals Intelligence Act, yielding six opinions and recommendations. 78Of the rest of the FII's activities, the 221 examinations that led to 23 opinions have by arrangement not been of visible interest to parliament.In fact, not even the oversight of signals intelligence led to much discussion in parliament.Every year since 2010, the Defence Committee has endorsed the report from the Ministry of Defence and rejected the few motions raised, and this has been endorsed in the plenary.Motions have been issued by the Sweden Democrats, which argues that the fact that not all parliamentary parties are represented in the FII is a democratic deficiency.The Left Party laments the NDRE's close ties with US intelligence agencies, especially in the light of the Snowden revelations.The Moderates, Christian Democrats and Liberals have generally pushed for a wider mandate for the NDRE.The overall impression however is that the political establishment as a whole, including the government of the day and parliament, is highly complacent with regard to the intelligence agencies and the oversight arrangement set up to scrutinise them.
There is no intelligence committee in parliament so defence intelligence is discussed in the Parliamentary Defence Committee.On the one hand, the broader scope of the committee reduces the risk of disincentivised parliamentarians, as discussed by Zegart, 79 but consideration of intelligence affairs might also be diluted in the sea of defence-related issues.

Political assessment
The Swedish political establishment regularly reiterates the importance of a vigorous oversight arrangement for preserving public trust in the intelligence agencies, not least given the secrecy surrounding the matters they are working on.The Swedish arrangement gives the impression of having the necessary checks and balances in place and there is little or no evidence of dissatisfaction among stakeholders.
Within the Swedish bubble, there is a strong narrative that the intelligence oversight arrangements are a success story, as the legal framework is rigorous, it champions the rule of law, and it encourages the intelligence agencies to be effective within legal parameters.This self-confident selfimage does not seem to have been particularly harmed by the 2021 ruling by the ECHR.It is as if we are all excellent when we mark our own homework.On the domestic arena, there is however nothing much that detracts from success criteria, even though the disinterest shown by Parliament in intelligence matters beyond the working of the Signals Intelligence Act rather testifies to a complacent indifference.

Changing requirements
One of the 2008-2009 amendments involved a specification as to which agencies should be able to direct NDRE operations.The original 2008 law left this up to the government to decide but the amendment allows only the government, the Government Offices and the armed forces to direct NDRE operations.This has consequences for law enforcement authorities, which also need signals intelligence on foreign targets.An official government inquiry and a ministerial inquiry investigated policing needs and their conclusions led to a government proposition that was accepted by parliament, which means that of 1 January 2013 the Security Services and the National Police can direct the NDRE's signals intelligence. 80As a result of the 7 April 2017 terrorist attack in Stockholm, a crosscoalition agreement between six of the eight parliamentary parties (excluding the Sweden Democrats and the Left party) paved the way for the police and the Security Services to direct the NDRE during ongoing criminal investigations, 81 which had been explicitly ruled out in the 2009 amendment. 82In 2021, the principle that signals intelligence operations could be allowed only where at least one party (sender or receiver) was located abroad was nibbled around the edges, since it was made inapplicable if the signals pertain to foreign military personnel, foreign state ships, foreign state aircraft or foreign military vehicles. 83In November 2021, parliament also decided that signals intelligence could under certain circumstances be gathered at the behest of international parties, even in the absence of a threat to Sweden or Swedish interests. 84Slowly, the reservations that once set the tone for the 2009 amendments have given way to empowering secret intelligence agencies.

Endurance assessment
Once up and running, the Swedish intelligence oversight arrangements have remained intact in terms of structure, mandate and resources since 1 December 2009.At the same time, the legislative framework that governs the intelligence agencies has undergone incremental change, to some extent rolling back the 2009 amendments.From a strict controlling-the-legality perspective, it is possible to argue that enhancing the mandates of the agencies made it easier for them to stay within the law, thereby lowering the burden on the oversight arrangement.In a way, the arrangement may be seen to have protected the agencies from scrutiny.Indeed, the number of opinions from the FII has decreased over the years, although this could also be attributable to learning and adaptation processes in the FII.At least by its own account, the FII has improved its methods over the years and developed expertise in how to control and encourage the agencies, which speaks to a developing effectiveness not only of the FII, but also of the intelligence community.Worth mentioning is also that the Snowden revelations, a potential pusher for overhaul of intelligence-related policies everywhere, did not leave any traces on Swedish intelligence oversight.In terms of durability, the arrangements have been successful.

Concluding remarks
It is one thing to compare intelligence oversight policy instruments with the conditions on their use by overseers and be disappointed, but this was never the objective of this study.Rather, by applying broader dimensions of policy evaluation that in and of themselves maintain a distance from politicised assessments, it has sought to disentangle evaluative categories and allow a more nuanced view of success.Overall, however, it is difficult to avoid the assessment that Swedish intelligence oversight arrangements are in a better state today than they were before 2009.That is not to say that there is no room for further improvement, but the direction of travel is promising.
In a nutshell, the relative longevity of Swedish intelligence oversight arrangements is probably attributable to the legitimacy given to them by the iterative process that led to their inception, that is, the theory of change.Most domestic observers seem to find the policy instruments appropriate and purposeful.Whether the political enthusiasm is genuine or merely reflects a lack of interest, however, is hard to tell.The fact that the signals intelligence legislation, including the oversight arrangements, is up for review after 12 years could be testament to a dynamic conservatism or to a hands-off approach until internal and external circumstances call for change.
This type of analysis deliberately omits some of the difficulties that overseers might run into when implementing policy instruments, and instead focuses on policy arrangements.It adds value because it complements analyses of a more instrumentalist nature, and perhaps inspires a focus on the bright side of secret intelligence policymaking.It is, after all, not certain that the relative success of Swedish intelligence oversight arrangements had been identified at all within more traditional analyses.That being said, adopting the analytical framework has also implied some challenges, which may provide fruit for thought in future refinements.In particular, the issue of from whose vantage point assessment should depart has proved somewhat problematic.Intelligence oversight is domestically oriented, which made it difficult to accommodate the outsider's view, in this case the verdict from the ECHR.The secret nature of intelligence also made it problematic to assess political effectiveness, at least in the absence of scandals or intelligence failures that could have distilled engagement from complacency.However, the lack of intelligence shocks may testify to the effectiveness of the oversight arrangement.

Table 1 .
32telligence policy as program -the spectrum from success to failure.32

Table 2 .
Intelligence policy as process -the spectrum from success to failure.41

Table 3 .
45telligence policy as politics -the spectrum from success to failure.45

Table 4 .
48durance of intelligence policy -the spectrum from success to failure.48