Institutional Maximization and Path Dependency: The Delay of Implementation of the European Union Public Sector Information Directive in Sweden

European integration can be hindered in ways that have not been anticipated by EUlegislators. In this paper the Swedish implementation, or rather the non-implementation, of the European PSI-directive is used as an example of how administrative inertia and path dependence may affect the adaptation of EU-legislation. In the paper, it is explored how the administration managed to stall implementation of the directive and the rationales behind the administrative resistance. It is also shown that the choice – even if unintentional – on how to translate a single word in the directive might have had a huge impact on the outcome. In this case, the word chosen, “handling”, has embedded the PSI-issue in two centuries of legal precedent and created a process path dependency.


Introduction
Decision-making processes can take a long time. In this paper we explore how the interaction between self-interested civil servants, differences in legal traditions, and secondary effects on national legislation can contribute to delays of implementation of EU-directives.
The inquiry conducted in this paper is seeking to explain why the Swedish government first delayed, and later refused full implementation of the PSI-directive, using theory on institutional inertia and path dependence to understand the process. It is shown that bureaucratic self-interest and path dependence (Mahoney, 2000) due to the translation of a single word in the directiveintentionally or not -had a significant impact on the implementation process.
European integration is based on a system where legislation on EU-level is implemented on national level. In this context the case could be used to understand how local institutions influences which directives will be implemented in the manner intended by the EU, if at all. It could also be used to understand which decisions might face resistance.

Background
EU legislation can take two forms: regulations that work as national law when they are adopted by the EU and directives that first need to be implemented.
In 2003, the EU adopted the directive on the re-use of public sector information (PSI) (2003/98/EC) aimed at establishing a facilitating regime for commercialization of public sector data. Twelve years later, the directive was not, implemented in all EU member states, Sweden having abstained from implementation. This makes it a rare example as Sweden usually implements EU-directives rapidly and extensively. Besides Sweden, Poland was the only member state that did not implement the directive on schedule.
The rationale behind the PSI-directive -to increase transparency and provide opportunities for the IT-sector -makes the Swedish reluctance to implementation rather puzzling since both are areas where Sweden usually pride itself for being a leader. However, as we shall see, the Swedish administration had other reasons for stalling. The Swedish case here clearly illustrates the conflict between a political vision and the administration's inherited self-interest (Christensen 1997).
The research question of this paper is why it took so long for a country, in this case Sweden, which usually implements EU-directives swiftly and extensively, to implement the PSI-directive, and what role administrative self-interest in combination with institutional differences between EU and Sweden played here.
As the European Union's reach increases and EU-legislation become more important for the member states. How legislation is (or is not) implemented is an important aspect.
The case can be used to understand both how bureaucracies can capture decision-making processes with relatively simple means. The paper also shows that trivial matters, such as the differences in translation might have a huge impact on the implementation since words have meaning and since legal definitions usually exist.
The PSI issue is special since few elected politicians took an interest in the process. Despite the fact that the Government has refused implementation with reference to subsidiarity, the issue has only been subject to parliamentary debate once. The PSI-issue has also attracted limited interest from members of parliament; only three have filed motions regarding PSI. The puzzle here is how an issue that is regarded as less important by the parliament came to be of such importance for the administration. Refusal to implement EU-directives is not common, when Sweden has done so it has been with highly controversial directives were important national assets has been threatened.
Maybe the issue is considered important for civil servants despite the lack of interest among elected officials. This might also explain regulatory capture.

On the PSI-Directive
The rationale behind the directive was that the commercialization and re-use of data collected by public sector entities were assumed to provide economic benefits for Europe, and to catch up with the United States (European Commission, 2000;2003a;. The directive was initiated as a tool to make PSI available to commercial interests in order to provide public benefits. The market for reused PSI has been estimated to be EUR 32 billion (European Commission, 2011b;The Guardian, 2007). This estimate is considered to be conservative and based on assumptions that were in place in (European Commission, 2011b. In addition to indirect economics benefits, PSI can be used to make businesses more efficient, reduce waste in government, increase citizens' confidence in government, and limit environmental impact by proper resource usage through coordination . The information redistribution sector, including start-ups and individual entrepreneurs, were optimistic that a rapid implementation of the PSI-directive would enable these companies to build new information based business. New high-tech start-ups utilize data mining techniques and public information to create entrepreneurial business models and are dependent on the access to public service information (Lakomaa and Kallberg 2013).
However, for the national administrations, there are few incentives to disseminate PSI. The four ways of disseminating public information described by Piotrowski (2007), public meeting, leaks, voluntarily dissemination, and freedom of information requests, are driven by actors other than the administration. The dissemination described by Piotrowski (2007) sees the information sharing as a result of political processes. The challenge occurs when politicians restrain from regulating the dissemination of information in detail and the administration is given a wide mandate to structure the release of public information.

Previous Research
Existing research on PSI is extensive but predominantly related to e-government inquiries addressing aspects of democratic theory, voter participation, democratic deliberation, and open government in a broader context (Amichai-Hamburger, McKenna and Tal, 2008;Bertot, Jaeger and McClure, 2008;Blackstone, Boganno and Hakim, 2005;Bertot and Jaeger, 2006;Chadwick, 2008;Carter and Weerakkody, 2008;Garson, 2003;Hernon, Cullen and Relyea, 2006;Ganapati, 2010;Khosrow-Pour, 2009). The value of PSI is often seen as strengthening democratic institutions and empowering citizens instead of commercialization. The existing relevant research is derived from democratic theory (Dahl, 1971;1998;Habermas, 1971;Estlund, 1997;Kallberg, 2011). The core set of reasoning and inquiry addresses citizen participation, the legitimacy and authority of the state, and the level of citizen confidence in the state's execution of power. Public access to PSI is a vehicle for transparency in the democratic context (Bertot, Jaeger and McGrimes, 2010;Franzel and Coursey, 2003;Nixon and Koutrakou, 2007;Nixon, Koutrakou and Rawal, 2010) or a method to address corruption and waste (Wong and Welch, 2004;Kallberg, 2011;Bertot, Jaeger and McGrimes, 2010).
The future value of commercialization, despite being the rationale behind the PSI-directive, has not attracted the same interest, the exception is the use of geo-GIS data, (Janssen 2011, Janssen, Charalabidis et al. 2012) -for other uses see e.g., Kallberg 2013).
The inherent conflict of interest between PSI owners and potential PSI-utilizing commercial entities is mentioned, but has not been investigated thoroughly in the existing literature. Blakesmore and Craglia (2006) state that the conflict between PSI-owners and potential PSIutilizing commercial entities can be seen in four different ways. These four perspectives are derived from previous research.
The first perspective is built on the assumption that the PSI-aggregating agencies, which in practice are governmental entities, are used to deliver services to the public. The PSI-aggregating agencies are influenced by institutional reasoning, such as possessing domain knowledge, and therefore regard themselves as superior in the delivery of PSI to the citizens (Lash, 2002). According to Lash, these agencies intend to manage and administrate all public aspects of the issues within their realm of influence. Therefore, they are seeking to obtain the eventual commercialization of PSI-information under their control.
The second perspective is that PSI is like any other commodity. In this market, the buyer and the seller have equal power and understanding of the market (Harvey, 2001). This perspective is indirectly founded on the assumption that the agencies are unbiased and lack institutional selfinterest, an assumption that is challenged by other research (Rothstein and Torell, 2008).
The third perspective focuses on the power struggle between Nation States and the EU. According to Richardson and Jensen (2000) there is a conflict between the European Commission, the executive arm of the European Union, and the member states. PSI is, according to the member states, considered to be a national issue and the control over PSI-information resides with the member states. Meanwhile the European Commission has exerted their authority on the PSI issue as an integrated part of their mission to increase European integration and cross-border cooperation.
The third perspective is interesting from a Nordic and Swedish perspective. It can explain why Sweden only implemented the PSI-directive twelve years after the directive was decided by the EU, and only after the EU had threatened Sweden with legal proceedings to enforce EUlegislation. According to the third perspective, the PSI itself is not the issue; it becomes a proxy for a larger power struggle between the over-arching European Union and the member states. This view could have explanatory power in the Swedish case as the Swedish administration has collected data extensively for over 50 years, which directly makes the data a core asset of the agencies. The Swedish administration is also in international perspective highly automated and relies on the data integrity of their databases.
The notion to surrender these data to a data dissemination based on a European Commission directive could, according to the third perspective, be seen as a significant financial loss for the Swedish government and a decline in national sovereignty.
The fourth perspective is purely economic and focuses on the notion that since data gathering has been funded by the tax payers, the data should also be freely accessible by the tax payers. Citizens are accustomed to free access over the Internet to unprocessed information, even if processed and refined information may cost money to access (Longhorn and Blackmore, 2004).
International research on access to information and pricing has mainly addressed topographical and geo-spatial information (Longhorn and Blackmore, 2004;Ganapati, 2010;Harvey, 2001;Norris, 2007;Richardson and Jensen, 2000;Pubellier, 2005) or national statistics (Cook, 2000). Limited research has been conducted in the field of inter-municipal cost sharing in the context of coordination and provision of jointly generated services (Tomkinson, 2007). Cerrillo-i-Martinez (2011) reports the Spanish development of dissemination of PSI but does not directly link the developments to the PSI-directive. Instead, dissemination of PSI is seen as a national effort to increase transparency within the Spanish government. Janssen (2011) studied compliance of the PSI-directive implementation and the alignment with the guidelines and overarching ideas formed by the European Commission that led up to the PSI-directive. Janssen (2011) reviews the legacy of actions, ideas, and the foundation that formed the incentive for the PSI-directive from the first initial technical reports in the late 1990s to the national implementation of the directive. Janssen is targeting the legal compliance of the PSI-directive. She writes: The general principle in Article 3 shows the limited ambition of the PSI directive: The Member States have to ensure that, where the re-use of documents held by public sector bodies is allowed; these documents shall be re-usable for commercial or non-commercial purposes in accordance with the provisions of the directive. Hence, the directive does not impose any obligation on the Member States or the public bodies to allow re-use, but only to comply with the obligations of the directive if they choose to make their data available for re-use. However, the freedom of the Member States or public sector bodies to allow re-use is not unlimited. When they decide to make their documents available for re-use for one purpose outside of the public task, e.g. research, they have to allow any other type of re-use of these documents. It is no longer possible for public sector bodies to allow non-commercial re-use of their data, while prohibiting any commercial application" (Janssen and Kabel, 2005).
Janssen points out that the purpose of the PSI-directive was to commercialize PSI and that any additional considerations such as democratic participation, transparency, and accountability, are beneficial for society but not the goal for the PSI-directive.
In the Swedish case, the institutional resistance against the implementation of the PSI-directive has a variety of rationales and motives. Bureaucracies tend to defend status quo, and to add to the change aversion (Samuelson and Zeckhauser 1988). There are several Swedish structural obstacles that can, if utilized against implementation, be significant hurdles for legislators to override the administration's unwillingness to implement.
The use of the word handling for a general piece of information in the translation to Swedish of the directive 2003/98/EC enabled the administration to deny massive public release of public sector information.

Methodology
The methodology used in this study is document review. The claims are based by comparing wetted sources and inferences of sources, previous research, and documents at hand.

A Single Word Makes a Difference
A reoccurring Swedish issue is the conceptual and terminological blending of Offentlighetsprincipen, the Swedish established right to read a physical copy of any unclassified official public print or document once processed in the administration, and PSI. The Swedish Tryckfrihetsförordningen, the ordinance of freedom for the press, is derived from Offentlighetsprincipen. The FOIA (Freedom of Information Act) extends the rights from the Offentlighetsprincipen, which translated to English would be "the Principle of Open Government".
The core principle of the Offentlighetsprincipen was established in 1766 and it has been revised and updated several times since then. It was not freedom of information per se but instead freedom to read what the King already had authorized to be published and the right to republish documents already cleared by the monarch to be published.
For a citizen to see what was written by the King, and his administration, the document had to be a drawn up document that is finalized. In Swedish, this is referred to as "upprättad handling". In the 2003/98/EC directive the English word "document" is translated to "handling", which is defined directly in Swedish administrative law and was established in the courts already in the 1800s as a drawn up, finalized administrative paper document. A "handling" is the government agency, the department, or the county's official stand and can be challenged by the citizen's in court. The "handling" has its inherent legal standing.
By using the word "handling" in the Swedish translation of 2003/98/EC, the national administration refuted the idea it could be data, data bases, digital information, GIS information, but instead had to be a paper copy of a drawn up, finalized official document. The usage of "handling" with no further explanation, which leaves the definition to rely of court precedence, triggered a path dependent behaviour, which is a major reason why the implementation of 2003/98/EC was delayed a decade.
Offentlighetsprincipen is well-established in Swedish administrative law with, by international comparison, large precedence in court cases. By being unable to separate Offentlighetsprincipen, Tryckfrihetsförordningen and the 2003/98/EC PSI-directive within Swedish administrative agencies, it created opportunities to block, delay, or create rules that work against the purpose of the 2003/98/EC PSI-directive.
The fact that public documents have been publically available since 1766 has created a legacy of the pricing, access, and timeliness of access to documents that is counter-productive to PSIdirective and is described in the literature (Woods, 2001). Woods states: Many of the member states government agencies at present charge for PSI and there is a widely held view that rigorous freedom of information legislation weakens the grip of government on the information. However, Sweden has had a freedom of information policy in place since the mid-1700s and the high rates of charging for PSI by its government agencies refute this view. Significantly, of all the countries in the survey, it makes the highest investment in the production of public sector information, and perhaps as a result it also has the highest value of information, of between 0.5% and 2.8% of its GNP.
Woods states that the assumption that well-established public access to documents would guarantee dissemination and access of PSI at an acceptable cost is false. Wood (2001) uses Sweden as the example. He clearly states that the legal legacy would be an obstacle. The Swedish public discourse was the opposite. The Swedish government, and public discourse, firmly officially believed that they were spearheading the dissemination of public information based on the traditional right to read public paper documents. This would have been true, if the directive 2003/98/EC was limited to drawn up, finalized official documents on paper.

The Financial Upside for the Administration
Swedish government agencies are funded via the national budget, but to avoid agencies and department returning small incomes to the treasury it has been the rule that when an agency or department generates any side-income this money can be kept by the government entity and utilized (SFS 1992:191). The rule originated in the 1970s when citizens started to ask for photocopies of public documents, providing a limited income for the agencies that charged 2 SEK (0.25 EUR) a photo copy. The lawmakers of the 1970s thought it was irrational to send back these incomes derived from SEK a photocopy to the treasury, so the agencies were allowed to keep any revenue that was raised from selling information or copies thereof leading to the Avgiftsförordningen. (SFS 1992:191). The Avgiftsförordningen (translated to English "the law of dues and tariffs"), allows agencies to actively sell their information and raise money, which can be used by the agency at their will within the law, and it will not affect their budget.
Thus, if an agency denies a private entity reuse, it can raise money and commercialize the PSI themselves. This was an administrative opportunity protected in the 2010 legislation.
Seen from this perspective, the Avgiftsförordningen law, which was designed to allow agencies to keep a few thousand Euro of photocopying fees, would allow the agencies to sell data for millions of Euro (see Lakomaa 2013 p 343). Several agencies started in the early 2000s to refine, process, and retail data themselves with the support of the Avgiftsförordningen. Lantmäteriverket, the Swedish mapping, cadastral and land registration authority, SMHI, the meteorological agency, and Bolagsverket, the national office of incorporation, adopted this idea and commercialized PSI data themselves. If the government themselves commercialize the data and become retailers of commercialized data it derails the intent with the 2003/98/EC directive, because the intent with the directive is to initiate a private enterprise utilization of the re-used data. A government commercialization would also limit the actual reuse, as the reuse would be within the boundaries of the government agency's field. As an example, an entrepreneur could design a dating application or direct marketing tools from cadastral land data, which the land survey agency will not do. The commercialization through the government agencies themselves retailing processed data thus has a limited economic potential.
The Swedish national statistics board SCB (2015) supplies raw data, but imposes charges for the computation of non-standard datasets. This is a form of commercialization as the data is held in the governmental sphere and any non-standardized analysis would be a matter of commercial activity from the agency.
The Swedish discourse within government agencies has been mainly focused on the opportunity to raise additional revenue and not the opportunity to create public good through increased utilization of data. Due to the path dependent relation to the document as a "handling" there has been, until the legislative session leading to the new legislation of 2015, limited discourse beyond the concept of a drawn up final official paper document. The state-run Swedish Institute for Economic Growth (ITPS) produced a report "PSI-direktivet -politik och potential" (ITPS, 2008) that avoids the larger context and instead only focuses on the commercialization of geo-spatial data in a limited market space. In the report "Fritt fram att avtala om offentlig information?", The Swedish Agency for Public Management, Statskontoret, describes processes for dissemination of PSI and what contractual concerns there might be. However, it does not separate Swedish Offentlighetsprincipen from the European PSI-directive leading to a distorted picture of the legal foundation (Statskontoret, 2005). This steering document issued by the central Swedish government is misleading and lacks understanding of the European objectives. Statskontoret (2005) considers that Sweden, by the older legislation, already meets the requirements from the PSI-directive, which is clearly inaccurate and evidenced by the European Commission's non-acceptance of the 2010 legislation and forcing compliance leading to the 2015 legislation.
The older Swedish legislation outlines the right to read legislation that requires the citizen to know that the resource exists, where it is stored and in which format it is stored, before it can be accessed. Additionally it requires an outlay to request the information, upon which the agency can deliver the information in a non-machine readable format, as long as the document is finalized and reaches the threshold to be considered a drawn up official document "handling". This extensive freedom for the agency to set its own rules and keep the proceeds can be used to sell digital services by the agencies themselves. The Swedish incorporation registry, Bolagsverket, acts as an independent retailer of its own information database of company ownership, the composition of the corporate boards, and any corporate filings. The Bolagsverket sells the information retail and provides a database for other information merchants, which is limited compared to the database the administration themselves market at retail pricing. There is no access to the complete database as raw data, and, if requested as freedom-of-information, then the Bolagsverket can deliver the database as several tons of printouts at a price of 2 SEK. The option to refuse machine-readable format and the legal right to invoice 2 SEK a printed pages ensures that Bolagsverket can block any competition that would undermine the commercialization business model of the agency.

Administrative Resistance through Confusion
In the 2010 legislation to implement the 2003/98/EC directive, there was a confusion that linked offentlighetsprincipen and the PSI-directive, that lead to the reliance on established, but irrelevant, legal precedence, which in turn prevented the implementation of the PSI directive. Offentlighetsprincipen has been legally codified with extensive published legal research (Lundell and Strömberg, 2009;Bohlin, 2010). This confusion creates a belief that there is a legal standing to claim that the requirements of the PSI-directive are met. As stated, the core problem is that the Swedish freedom-of-information legislation and PSI-directive are not interchangeable and different from a legal perspective.
Offentlighetsprincipen and supporting of auxiliary Swedish freedom of the press legislation, is a legal tradition since the mid-1700s that aims to allow the public to share a public decision made by the royal authorities, once determined and published by the authorities themselves, and this right is extended over time to all public documents that are not classified. The citizens' right is to read the public document. The citizen will be provided with an opportunity to read the public documents at any Swedish government agency. If the citizen wants a copy to read at home, it becomes a financial transaction between the citizen and the agency.
Pricing is based on the interest to see one sheet of paper, or a few, and time is not of the immediate essence. The precedence is that a request should be processed within 4 to 6 weeks. The agency can freely choose the method of disseminating the information. A request to obtain a copy of the computerized data set of court dockets could at the court's discretion be printed out on thousands of papers and delivered 4 to 6 weeks after the request at a price of 2 SEK a page. The method of delivery may reduce the functional usability of the data, rendering it useless for re-use purposes. The PSI-directive was initiated to imitate the U.S.'s success in creation of information services based on re-used public sector information. The EU and European Commission identified that Europe was far behind the U.S. in capitalizing on its public sector information assets to drive the evolution of the information society. In the U.S., these data sets are either free or made available at low cost.

The Swedish Definition of Document
In the original English version of the PSI-directive, the word "document" is used (European Commission, 2003a), which in the United Kingdom has a wider interpretation than the Swedish "handling" that is used in the translation to Swedish in the official European documentation (European Commission, 2003b). This apparent translation error triggers a chain of confusion. The focus in the Swedish translation becomes "handling" which in the legal precedence in Swedish is a sheet of paper that contains written information. The misconception proliferates into the Swedish implementation by using the freedom of press definition of "handling" which dates all the way back to the mid-1700s as an expedited final written act of the government. The word "handling" is well defined in the national Swedish legal literature (Lundell and Strömberg, 2009;Bohlin, 2010). The main construct in the legal framework of the Swedish freedom of press is defined as the right of any citizen to ask and see the underlying documents of a decision, once a proceeding is finalized through the final decision of a public servant, provided these documents are not classified. This request has to be processed in a "reasonable" time. By using the obsolete and misleading interpretation, raw data could be seen as being outside of the scope of the PSI-directive. The absence of Swedish PSI-directive supporting literature and the amassed volume of precedence and literature defining offentlighetsprincipen create an obstacle for releasing raw PSI-data to commercialization. The key term is the Swedish "upprättad" which can be translated to the English "prepared" or "expedited". In the Swedish offentlighetsprincipen, an expedited and prepared ("upprättad") document cannot be in legal doctrine an unprocessed data point because the process itself generates the preparedness that constitutes the threshold for being considered "expedited and prepared".

Administrative Resistance through Delayed Action
In the information society, we are used to real-time access to data. The Swedish PSI-directive implementation of 2010 links the dissemination to the older established offentlighetsprincipen. The existence of confusion stems from older laws that have precedence for how much time is acceptable for the fulfilment of a public records request. The Swedish laws use the word "skyndsamt", for which the English translation would be "promptly", and is linked to the workload of the government agency. If the agency is busy, a "prompt" compliance with the request can take over a month to process due to the legal precedence being over a century old. Understandably, the pace in information processing was slower in the past as compared to today.
Government agencies can use this precedence to move their own retailing of processed PSI outside the scope of the PSI directive. If commercialized private entities want access to information, the information could be a month old; meanwhile the agency retails real-time information over the Internet at retail price. This does not comply with the PSI-directive, but, current Swedish law, in its 2010 version, allows the opportunity for agencies to attempt to maximize revenues. This feature in the implementation of the PSI-directive, combined with the Swedish administrative code Avgiftsförordningen, enables government agencies to use, without scrutiny, any revenue generated from the freedom of information.
The time of delivery has not been investigated in the peer-reviewed published literature, but is essential for the factual commercialization of PSI. Even if it will occur in later revisions of the PSIdirective, the implementation is still national.
When Neelie Kroes of the European Commission presents the future development of the PSIdirective in three major directions, the delivery time aspect is not mentioned (European Commission, 2011a): The Commission proposes to update the 2003 Directive on the re-use of public sector information by: Making it a general rule that all documents made accessible by public sector bodies can be reused for any purpose, commercial or non-commercial, unless protected by third party copyright; Establishing the principle that public bodies should not be allowed to charge more than costs triggered by the individual request for data (marginal costs); in practice this means most data will be offered for free or virtually for free, unless duly justified.
Making it compulsory to provide data in commonly-used, machine-readable formats, to ensure data can be effectively re-used.
Introducing regulatory oversight to enforce these principles; Massively expanding the reach of the Directive to include libraries, museums and archives for the first time; the existing 2003 rules will apply to data from such institutions.
The four obstacles identified are the price of the information, the need for machine-readable format, accessibility, and timing. The problem is that without set limits to restrict delays of dissemination of data in the Swedish setting, it would be up to each agency to determine what a timely dissemination could be. As the directives do not set any time limits for dissemination, the precedent from the field of FOI takes over. This means that the Swedish Department of Motor Vehicles may determine that providing 60-day old vehicle registration data is sufficient for compliance with the Swedish legislation or the National Weather Service (SMHI) may determine that the disclosure of the weather forecast for yesterday is sufficient for compliance, thereby undermining the directive. If the same agencies simultaneously act to sell processed, real-time data in retail, further damage is done to the spirit and wording of the directive.
Administrative resistance has also relied on privacy concerns based on EU protection of privacy (European Commission, 1995) to avoid non-government commercialization and usher in agency maximization through the government itself as the commercialization partner. The privacy issue is central as the Swedish government collects a massive amount of personal data on the citizens. Under the Swedish offentlighetsprincip such data is accessible through FOIA.
The Swedish FOIA handles the associated privacy risks by only releasing sensitive information on print paper. Traditionally this has prevented wide dissemination, re-use of sensitive data as cost and time issues have prevented persons and companies to acquire usable data sets, and prevented them from cross-referencing data.
The requirements of machine-readability in the PSI directive requires substantial changes in the Swedish privacy laws, notably in combination with personal data being purged from the Government databases in order to reduce privacy risks associated with personal data collected

Administrative Resistance Supported by Political Fear
International research has shown that resistance to transparency and to the release of large aggregations of bulk data increases if the political elite assumes there is a political risk involved (Coglianese, 2009). The release of data gives others an opportunity to see results that could have been ignored or missed by the agencies themselves . Politicians are more concerned about their legacy than they tend to express (Dobel, 2005;Ruscio, 2004). Political survival instinct therefore creates a political anxiety over the release of raw data in bulk. A British survey has shown that trust for government is lost initially when data and information are made freely available (Worthy, 2010).
Data from national government, counties, and municipalities that record program performances and policy implementation are likely to be thoroughly researched by interest groups, trade unions, legal activists, political opposition, and civil society. Those who set out on a journey to find failure are likely to find it according to Worthy (2010).
From this perspective, government authorities have strong incentives to control or delay the implementation of the PSI directive. It could also provide an explanation as to why Sweden, a nation that historically has been a quick implementer of directives, has not, twelve years later, fully implemented the 2003/98/EC directive (European Commission, 2003a;SFS 2010:566; The Government of Sweden 2015).
In comparison with the U.S. (Tolbert, Mossberger, and McNeal, 2008), not only is Sweden far behind in PSI dissemination, but also the rest of Europe. The U.S. and Australia (Creative Commons, 2015; The Government of Australia, 2015a; b) have taken steps to further increase accessibility to PSI data from federal, state and local government (U.S. Office of Management and Budget, 2004;2015c;U.S. Congress, 2003; State of Texas, 2015a; 2015b; Federal Funding Accountability and Transparency Act of 2006).

Conclusion
The Swedish delay in implementing the PSI-directive is probably multi-causal and might be explained by at least three factors: First, the translation of the English word "document" that was translated to "handling" triggering a path dependency of older freedom-of-information legislation that was not relevant. Second, administrative self-interest and agency maximization. Finally, weak political leadership combined with an inability to transpose visionary thinking to tangible results. Political visionary thinking and broad ideas can stumble -if the instructions to the administration are too open-ended or if the administration realize that there is no real political intent behind the different transparency postures presented by government political figureheads. The Swedish case illustrates the complexity of implementing European legislation pursuing information dissemination requiring the cooperation of an established path dependent administration. Obstacles could be found within the administration where removal of the ability to sell data would reduce autonomy and therefore require overcompensation. In addition, the implementation of the directive would also require extensive adaptations of other legislation, which would be costly, but also could be seen as an unwelcome divergence from established tradition.
An additional finding is that, if the administration is left with limited political supervision to tailor the implementation of the European public sector information directive, it is likely that only limited PSI will be released for commercialization. If the administration is able to commercialize data dissemination as a core activity of its own accord, then the administration will likely seek to prevent other actors, such as private enterprises, to enter the market and gain access to the data held by the administration. Due to factors of institutional inertia, early commercialization using the administration as the bridge to private commercialization could increase the resistance to enlarged dissemination.