The economics of business-to-government data sharing

Martens, Bertin; Duch Brown, Nestor Working Paper The economics of business-to-government data sharing JRC Digital Economy Working Paper, No. 2020-04 Provided in Cooperation with: Joint Research Centre (JRC), European Commission Suggested Citation: Martens, Bertin; Duch Brown, Nestor (2020) : The economics of business-to- government data sharing, JRC Digital Economy Working Paper, No. 2020-04, European Commission, Joint Research Centre (JRC), Seville This Version is available at: https://hdl.handle.net/10419/227699 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ JRC Digital Economy Working Paper 2020-04 The economics of Business-to- Government data sharing Martens, Bertin Duch-Brown, Néstor 2020 This publication is a Technical report by the Joint Research Centre (JRC), the European Commission’s science and knowledge service. It aims to provide evidence-based scientific support to the European policymaking process. The scientific output expressed does not imply a policy position of the European Commission. Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use that might be made of this publication. For information on the methodology and quality underlying the data used in this publication for which the source is neither Eurostat nor other Commission services, users should contact the referenced source. The designations employed and the presentation of material on the maps do not imply the expression of any opinion whatsoever on the part of the European Union concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Contact information Name: Néstor Duch-Brown Address: Inca Garcilaso 3, 41092 Seville (Spain) Email: Nestor.DUCH-BRO[email protected]pa.eu Tel.: (+34) 954.488.216 EU Science Hub https://ec.europa.eu/jrc JRC119947 Seville: European Commission, 2020 © European Union, 2020 The reuse policy of the European Commission is implemented by the Commission Decision 2011/833/EU of 12 December 2011 on the reuse of Commission documents (OJ L 330, 14.12.2011, p. 39). Except otherwise noted, the reuse of this document is authorised under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence (https://creativecommons.org/licenses/by/4.0/). This means that reuse is allowed provided appropriate credit is given and any changes are indicated. For any use or reproduction of photos or other material that is not owned by the EU, permission must be sought directly from the copyright holders. All content © European Union, 2020 How to cite this report: Martens, B., Duch-Brown, N., The economics of Business-to-Government data sharing, European Commission, Seville, 2020, JRC119947. i Contents Foreword ..................................................................................................................................................................................................................................................................... 3 Acknowledgements .......................................................................................................................................................................................................................................... 4 Executive summary .......................................................................................................................................................................................................................................... 5 1 Introduction..................................................................................................................................................................................................................................................... 7 2 The potential socio-economic welfare gains from B2G data sharing .................................................................................................... 8 2.1 The potential welfare gains from sharing data ............................................................................................................................................. 8 2.2 The economic impact of data sharing ..................................................................................................................................................................... 9 2.3 There is an emerging market for B2G data sharing operations ................................................................................................. 10 3 Barriers to B2G data sharing and how to overcome them ............................................................................................................................ 12 3.1 Monopolistic data suppliers and data pricing ............................................................................................................................................... 12 3.1.1 Pricing in monopolistic markets ............................................................................................................................................................... 12 3.1.2 Intervention in market pricing .................................................................................................................................................................... 14 3.2 Ex-ante transaction costs and ex-post risks .................................................................................................................................................. 16 3.2.1 Trusted technical intermediaries ............................................................................................................................................................. 16 3.2.2 Quality: management of the transformation of the transferred data ............................................................... 17 3.3 Externalities and incentive misalignment ......................................................................................................................................................... 18 3.3.1 Negative externalities and risks ............................................................................................................................................................... 19 3.3.2 Positive externalities and economic incentive misalignment ...................................................................................... 19 4 How to overcome these market failures and boost welfare gains?...................................................................................................... 21 References ............................................................................................................................................................................................................................................................. 22 Glossary of economic terms ................................................................................................................................................................................................................. 23 List of figures ..................................................................................................................................................................................................................................................... 26 Annexes .................................................................................................................................................................................................................................................................... 27 Annex 1. Monopolistic pricing and price discrimination ..................................................................................................................................... 27 Annex 2. Demand and data pricing regimes ................................................................................................................................................................ 27 ii 3 Foreword As part of a JRC-CNECT research programme on digital platforms and data, CNECT/G1 requested the JRC to contribute an economic input to the on-going reflections in the B2G data sharing Expert Group 1 . The Expert Group was set up to (a) identify good practices, (b) assess the legal, economic and technical obstacles preventing B2G data sharing, (c) advise on actions to promote B2G data sharing for public interest purposes and (d) give recommendations to the Commission on how to further develop its B2G data sharing policy. The subject of B2G data sharing was first discussed in the Commission Communication (2018) "Towards a common European data space". The Staff Working Document that accompanies the Communication sets out the basic principles of B2G data sharing and discusses some legal and technical conditions that may facilitate these operations. Starting from these principles, the B2G Expert Group discussed further refinements and details and explored how they could be cast into a more operational governance framework for B2G data sharing operations. The current report complements the debates in the Expert Group with an economic perspective on several of these principles. 1 https://ec.europa.eu/digital-single-market/en/news/meetings-expert-group-business-government-data- sharing 4 Acknowledgements The authors would like to thank the comments and suggestions to earlier versions of this document provided by Alberto Gago-Fernandez, Johan Bodenkamp, Julie Sors, Maria Rosaria Coduti and the rest of colleagues from DG CNECT G1. Similarly, this report has greatly benefited from inputs provided by the Business-to- Government Data Sharing High Level Expert Group, and in particular from Alberto Alemanno. In addition, Michael Lutz (JRC B6) also provided important suggestions for improving the readability of the final version. Finally, Fiammetta Rossetti provided excellent research assistance. Remaining errors and omissions are the sole responsibility of the authors. Authors Bertin Martens Néstor Duch-Brown 5 Executive summary Data and information are fundamental pieces for effective evidence-based policy making and provision of public services. In recent years, some private firms have been collecting large amounts of data, which, were they available to governments, could greatly improve their capacity to take better policy decisions and to increase social welfare. Business-to-Government (B2G) data sharing can result in substantial benefits for society. It can save costs to governments by allowing them to benefit from the use of data collected by businesses without having to collect the same data again. Moreover, it can support the production of new and innovative outputs based on the shared data by different users. Finally, the data available to government may give only an incomplete or even biased picture, while aggregating complementary datasets shared by different parties (including businesses) may result in improved policies with strong social welfare benefits. The examples assembled by the High Level Expert Group on B2G data sharing show that most of the current B2G data transactions remain one-off experimental pilot projects that do not seem to be sustainable over time. Overall, the volume of B2G operations still seems to be relatively small and clearly sub-optimal from a social welfare perspective. The market does not seem to scale compared to the economic potential for welfare gains in society. There are likely to be significant potential economic benefits from additional B2G data sharing operations. These could be enabled by measures that would seek to improve their governance conditions to contribute to increase the overall number of transactions. To design such measures, it is important to understand the nature of the current barriers for B2G data sharing operations. In this paper, we focus on the more important barriers from an economic perspective: (a) monopolistic data markets, (b) high transaction costs and perceived risks in data sharing and (c) a lack of incentives for private firms to contribute to the production of public benefits. The following reflections are mainly conceptual, since there is currently little quantitative empirical evidence on the different aspects of B2G transactions. Monopolistic data markets. Some firms -like big tech companies for instancemay be in a privileged position as the exclusive providers of the type of data that a public body seeks to access. This position enables the firms to charge a high price for the data beyond a reasonable rate of return on costs. While a monopolistic market is still a functioning market, the resulting price may lead to some governments not being able or willing to purchase the data and therefore may cause social welfare losses. Nonetheless, monopolistic pricing may still be justified from an innovation perspective: it strengthens incentives to invest in more and better data collection systems and thereby increases the supply of data in the long run. In some cases, the data seller may be in a position to pricediscriminate between commercial buyers and a public body, charging a lower price to the latter since the data would not be used for commercial purposes. High transaction costs and perceived risks. An important barrier for data sharing comes from the exante costs related to finding a suitable data sharing partner, negotiating a contractual arrangement, re-formatting and cleaning the data, among others. Potentially interested public bodies may not be aware of available datasets or may not be in a position to handle them or understand their advantages and disadvantages. There may also be ex-post risks related to uncertainties in the quality and/or usefulness of the data, the technical implementation of the data sharing deal, ensuring compliance with the agreed conditions, the risk of data leaks to unauthorized third-parties and exposure of personal and confidential data. Lack of incentives. Firms may be reluctant to share data with governments because it might have a negative impact on them. This could be due to suspicions that the data delivered might be used to implement market regulations and to enforce competition rules that could negatively affect firms’ profits. Moreover, if firms share data with government under preferential conditions, they may have difficulties justifying the foregone profit to shareholders, since the benefits generated by better policies or public services fuelled by the private data will occur to society as a whole and are often difficult to express in monetary terms. Finally, firms might be afraid of entering into a competitive disadvantage if they provide data to public bodies – perhaps under preferential conditions – and their competitors do not. Several mechanisms could be designed to solve the barriers that may be holding back B2G data sharing initiatives. One would be to provide stronger incentives for the data supplier firm to engage in this type of transactions. These incentives can be direct, i.e., monetary, or indirect, i.e., reputational (e.g. as part of corporate social responsibility programmes). Another way would be to ascertain the data transfer by making the transaction mandatory, with a fair cost compensation. An intermediate way would be based on solutions that seek to facilitate voluntary B2G operations without mandating them, for example by reducing the 6 transaction costs and perceived risks for the provider data supplier, e.g. by setting up trusted data intermediary platforms, or appropriate contractual provisions. A possible EU governance framework for B2G data sharing operations could cover these options. 13 business, if not the firm will run a loss 14 . This implies that some social welfare losses are inevitable. This is illustrated in annex 1. A B2G example of data quality discrimination concerns a public transport company that may buy a cheaper more coarse-grained mobility dataset from a telecom operator, rather than a more expensive fine-grained dataset, in order to reduce the cost of data procurement. This may result in a less optimal transport schedule for consumers. The mobility data provider may be willing to provide more fine-grained data at a higher price, thereby extracting more welfare surplus from the transport company and its consumers. Price differentiation is often linked to quality differences. In some cases, the data seller may be in a position to price-discriminate between commercial buyers and a public body, charging a lower price to the latter. An example of this type of B2G price discrimination comes from the JRC's own experience with the procurement of a unique dataset on online digital transactions from a private market analysis firm. While these data have a high commercial value for firms operating online stores, the supplier agreed to sell it at a much lower price to the JRC because it would not be used for commercial purposes. This led to a price negotiation between the two parties, still above marginal cost however. There are also examples from our own experience where the JRC was unable to procure data at a reasonable price and had to cancel the transaction. In some instances, even if the data seller accepted a reduction in price, the price proposed was still above the available budget. In some others, after some negotiation, data holders decided not to sell the data perhaps due to the fear of reputational losses. Pricing hinges on the assumption that the demand curve reflects true demand for the data (see annex 2 for a detailed graphical exposition). However data are often an intermediate input into a service produced by a downstream party. In this case, the willingness to pay for the data will depend on the derived demand in the downstream application. For example, mobile phone data may be an input by a public transport company to improve its bus or train schedules to better accommodate consumer demand at different places and times in a city. In that case increased demand for transport services may reflect the true welfare value of the intermediate data delivery. The cost of the data input can be recuperated via ticket prices. If the data input is competitively procured, then the transport company pays a market price to the data provider. If there is only a single provider, the transport company may have to pay a monopolistic price. This gives the upstream data provider an opportunity to extract a large share of the consumer value produced by the downstream transport service provider. In short, we cannot examine the upstream data market and downstream services market separately; we have to look at them as an integrated entity. If there is competition among different data providers, B2G data pricing is likely to be close to marginal costs plus some mark-up or return on investment, depending on the extent of competition in the market. For example, a public body can tender for mobility data from several mobile phone operators in a country. If this were part of the firms’ main line of business, fixed costs may interfere in this process and exert upward pressure on prices because in that case a firm also wants to amortize its fixed costs. When additional data processing is done for a specific user, that user may be charged the full marginal cost but also the additional data processing cost, unless competition is strong and puts downward pressure on mark-up pricing. It is not clear to what extent monopolistic data pricing actually occurs in B2G sharing transactions since there is a lack of empirical evidence on this area and there will be high variation across sectors. One can think of examples of monopolistic data supplies in B2G data sharing situations. For example, a city authority may request access to private room reservation data from an online booking platform that operates in the city, for reasons related to tourist taxes, safety inspections and the implementation of other city council regulations with regard to private short-term rentals. Every booking platform has exclusive access to its own room reservation data and is therefore in a de facto monopoly position. If sharing the data with the city authority would be mandatory under the city regulation, that would reduce the price to zero for the public body 15 , 14 This can also be the case in markets where there are a relatively large number of firms but selling differentiated products, i.e., products that are not identical for different consumers. This feature gives firms some market power because they will be the sole producers of their variety, while the large number of competitors would suggest a market close to perfect competition. This market structure is referred to as monopolistic competition, and is characterized by the fact that firms price according to their average costs, but they retain some market power since the prices will be above marginal costs. 15 DG COMP has a mandate under Art 101 TFEU to request data from firms, not only for specific competition cases but also for wider sector inquiries that are not necessarily case-related. It used this mandate in 2017 to collect firm data in the framework of an e-commerce sector inquiry. Attempts by DG GROW to obtain a similar mandate to access firm-level data in the context of the Single Market (the SMIT proposal) failed because there were insufficient legal grounds for such a mandate. 14 although the data supplier would still bear the costs of data collection, processing and transmission. If it is not mandatory, the city authority may negotiate a price with the private provider by seeking a lower price than the monopoly one 16 . 3.1.2 Intervention in market pricing The Commission SWD (2018) already suggested that pricing of B2G data transfers could be linked to the public interest purpose pursued and the social need it aims to fulfil. In other words, the compensation paid to the private data supplier should vary in function of the expected social welfare benefits. Firms supplying data to a more important, urgent and valuable public purpose would receive a lower price or compensation for their data because it would be used to tackle a highly relevant public interest matter. The B2G Expert Group has further developed this reasoning and suggested a range of public interest data sharing projects where free data, marginal cost pricing or marginal cost plus a return-on-investment mark-up would apply as pricing rules in all B2G data transactions, irrespective of the data market structure 17 . Payment of normal market prices would still be allowed in a certain number of circumstances. The economic logic of marginal cost pricing for data was first developed in the PSI Directive as a means to facilitate G2B data sharing and wider use of government data. The underlying assumption was that (high) prices could reduce demand and the number of G2B data sharing operations. Users with a low willingness and/or ability to pay for the data would not be able to access them. Lowering the price paid for data could thus increase the number of operations. Pollock (2008, 2010) investigates the economic impact of moving from average to marginal cost pricing rules in G2B data sharing (PSI) for some specific cases of government agencies. Pollock (2010) then extrapolates the micro-findings to the macro-economic level by means of a multiplier effect. These studies do indeed show that marginal cost pricing lowers the social welfare losses compared to monopolistic pricing (See Figure A1a). These PSI studies do not consider the supply side impact of lower prices on the incentive to invest in data collection because the data were already collected by the government for other purposes, irrespective of any alternative uses. The incentive effect could therefore be ignored. A similar reasoning can be applied to data collected by firms for reasons related to their main line of business and without regard to alternative uses. The cost of collecting, storing and processing the data is already covered by the main activity of the firm. From this perspective, sharing the data with others at a marginal or zero cost would not affect the profitability of the main business of the firm. Data holders may initially not be using their data for other purposes than those needed to sustain their main business, but they may become increasingly aware of the potential value of selling their data for other businesses. They can start developing new data-related revenue streams outside their traditional business model. For instance with the help of APIs they can launch new data-driven services in the hope that these will scale and amortize the fixed set-up costs in order to produce new revenue streams. The banking sector is a good example. Spanish bank BBVA started an API market as an intermediary platform that enables new B2B commercial data transactions related to banking operations and customers, using standardised interfaces. Selling data can therefore become a significant source of new revenues for data-intensive firms. Most of these new data market opportunities are situated in the B2B realm, though some may also involve governments and public bodies as clients. In order to scale up the volume of B2G transactions, the Expert Group has considered the possibility of intervening in the pricing conditions for B2G transactions, i.e. a marginal cost pricing rule, possibly with a return-on-investment top-up margin. The underlying reasoning would be that lower prices would increase the purchasing power of governments to buy more data and thus expand the use of private data in the public interest. The firm would in any case continue to collect, store and use the data to pursue with its normal main business and could also start selling the data in B2B settings as 16 Another example from the JRC’s own experience: The JRC bought very detailed data on online and offline sales of electronics products in several European countries from a marketing firm for the purpose of informing an EU policy decision. The firm was in a monopoly position since the data were quite unique and not available from other firms. The JRC therefore had to negotiate a reasonable price with the monopolistic supplier and, despite the efforts, still paid a considerable amount of money. The data provider sold the data to other firms interested in understanding their niche markets. Hence, we can safely assume that the price paid was close to monopolistic levels. 17 The concept of "marginal cost" is a theoretical economic construction that is often not so easy to calculate in practice. The additional costs generated by preparing a dataset for a new purpose may not be easy to distinguish from other applications of the dataset. 15 additional, new business activities. Private firms could sell this data at competitive market prices to other firms and they would in this new market activity not be deprived of additional data revenue. Another consideration concerns competition in public procurement markets. Many government activities require the public procurement of goods, services and labour. This is usually done on market terms through tender and procurement procedures, including for data procurement. For example, some European Commission services, and particularly the JRC, buy access to private data sources for their recurrent activities through competitive tender procedures 18 . Governments may divert from these procurement procedures when there is no competitive market to supply the goods or services and they may replace it with alternative mechanisms such as framework contracts, supply arrangements, joint ventures or strategic alliances, among others. The structure of the market is thus an important issue in the choice of the appropriate mechanism, which in turn can have an impact on the final prices to be paid 19 . There is an implicit assumption that competitive procurement is the most economically efficient method, but this is not necessarily always the case. Indeed, there are many situations where the conditions for competition are not suitable or projects are complex or urgent to use public procurement. For example, if there is no active market on the type of data that governments seek to reuse, public procurement is a difficult endeavor. Companies usually tend to give a very high estimation of how much their data is worth. However, the actual value of data is not based on the data itself but on the outcomes of the potential combinations of that data with other data sources for a specific purpose. Having this in mind, the usefulness of data for the intended public interest purpose cannot be guaranteed ex-ante, which makes public procurement procedures delicate as well. If that data is finally not useful, there is a clear societal loss. Furthermore, without knowing that usefulness, governments find it difficult to set a fair maximum price in the call for tenders. Critically, many of the public interest purposes that may be attained through B2G collaborations require access to many datasets from many different companies. This is either due to their inherent complex purpose or to the methodological need to ensure representativeness of the results. For example, if a government agency would like to understand the increased mortality of bees, officials and/or researchers would need data from various sources (e.g. data from the beekeepers, Common Agricultural Policy registrars, pesticides sellers, land owners, agricultural data, weather data, etc.). Multiple parallel procurement procedures cannot possibly ensure that all these data are accessed on time, not to mention the high cost of each individual procured dataset that in itself is not useful unless combined with the rest of the datasets. If one or several of those potential sources would reject access to its data, there is a direct impact on the real value of all procured datasets, with important societal loss. Finally, with procurement of data, it is never clear whether a significant number of well qualified, interested contractors may participate in the tender or that the tenders will be of a sufficient size in order to justify the high transaction costs implied by procurement procedures. For these reasons, for many public interest purposes, procurement of data does not provide welfare gains for society and alternative methods should be considered. Data are not legally protected by IPR rights, unless they would fall under the EU Database Directive that recognizes a copyright on original creatively designed databases or a sui generis right on other types of databases, subject to strict conditions 20 . However, apart from the legal argument, the economic argument of the incentive effect of protecting databases is well recognised. Indeed, this was the main motivation for setting up the EU Database Directive. The incentive argument is probably stronger for firms that sell data as their main business compared to firms who sell data as a by-product of their main activity. Indeed, the CJEU 18 EU Directive on Public Procurement, EU 2014/24. It is not clear to what extent the provisions of the EU Directive on public procurement apply to data procurement. The Directive underlines that there is considerable legal uncertainty as to how far contracts concluded between entities in the public sector should be covered by public procurement rules. The Directive lists many exceptions to the rules, including for "activities of general economic interest" (Whereas 20) and for "electronic communications" (Art 8). EU commitments under the WTO Government Procurement Agreement should also be taken into account. See https://www.wto.org/english/tratop_e/gproc_e/gp_gpa_e.htm 19 The European Commission's own procurement rules for example explicitly require a detailed market description and justification of the monopolistic position of the supplier before a direct agreement can be negotiated with a supplier, by-passing the standard procurement procedures. 20 Governments have the possibility to request a court to grant a compulsory license for the patented product at non-monopolistic preferential prices and in the public interest. This may happen for medicines in developing countries and in medical emergencies. Normally, government health services or governmentsupported medical insurance schemes pay the monopolistic price and may subsidize low-income consumers to access the medicines. 16 has made a difference between these situations in its case law on the sui generis right of the EU Database Directive, and has given a narrow interpretation to this right 21 . Therefore, the 2018 Commission evaluation on the EU Database Directive has concluded that the sui generis right does not apply to most situations in the data economy 22 : databases which are the by-products of the main activities of an economic undertaking (‘spin-off’ databases) are in principle not protected by the sui generis right, as they would not fulfil the ‘substantial investment’ threshold of article 7 23 . It is sometimes claimed that the public interest of government activities is a justification for preferential pricing 24 . In fact, all government activities are meant to produce goods and services that are in the public interest and welfare enhancing for society; if not there is no reason for government to undertake that activity. The Expert Group discussed a range of interpretations of "public interest", from humanitarian emergencies and climate change to more practical applications such as statistics and ordinary government policy decisions. The public interest can be defined in economic terms as any activity that increases the welfare of society as a whole or the welfare of a disadvantaged group in society in order to address equity concerns. There are many economics handbooks that propose methods to assess and quantify the public interest (for example Auerbach and Feldstein, 1985; Diewert, 1983; World Bank, 1998). Public bodies, including the Commission, do impact assessment in order to assess the welfare impact of proposed policies. 3.2 Ex-ante transaction costs and ex-post risks An important barrier for data sharing comes from the ex-ante costs related to finding a suitable data sharing partner, negotiating a contractual arrangement, re-formatting and cleaning the data, etc. Potentially interested public bodies may not be aware of available datasets or may not be in a position to handle them or understand their advantages and disadvantages. Given that data is an experience good, there may also be expost risks related to uncertainties in the quality and/or usefulness of the data, the implementation of the data sharing deal, ensuring compliance with the agreed conditions and the risk of data leaks to unauthorized thirdparties and exposure of personal and confidential data. There are at least two ways to deal with these exante transaction costs and ex-post risks 25 : (a) using trusted technical intermediary platforms to manage the data transfer operation and (b) upstream management of the "quality" of the data that are transferred to a downstream user. 3.2.1 Trusted technical intermediaries Technological and institutional arrangements, and regulations, can reduce ex-ante transaction costs and expost risks and thereby enable more voluntary transactions to take place, without necessarily making them mandatory (IDC, 2015, p 27). The presence of an intermediary that reduces ex-ante transaction costs and expost risks, for example by means of standardized contractual provisions, may improve the enabling environment for data sharing. Verhulst, Young and Srinivasan (2015) discuss several means to reduce transaction costs such as APIs with standardised datasets and access protocols, data cooperatives and data 21 Judgment of the Court (Grand Chamber) of 9 November 2004. Fixtures Marketing Ltd v Organismos prognostikon agonon podosfairou AE (OPAP). Reference for a preliminary ruling: Monomeles Protodikeio Athinon - Greece. Directive 96/9/EC - Legal protection of databases - Definition of database - Scope of the sui generis right - Football fixture lists - Betting. Case C-444/02. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62002CJ0444. 22 Commission Staff Working Document, Evaluation of Directive 96/9/EC on the legal protection of databases, 2018/04/25, SWD(2018) 146 final. 23 Regarding the sui generis right, the investment must be qualitatively or quantitatively substantial. The CJEU has ruled that the expression «investment in the obtaining of the contents» must be understood to refer to «the resources used to seek out existing independent materials and collect them in the database», and that «it does not cover the resources used for the creation of materials which make up the contents of a database» (CJEC, Grand Chamber, 2004/11/09, The British Horseracing Board Ltd and Others v William Hill Organization Ltd, C-203/02 ; Fixtures Marketing Ltd v Oy Veikkaus Ab, C-46/02). 24 Public interest should not be confused with public goods and services. For example transport, education and health services are produced by a mix of private and public providers in most countries. They are not public goods because they are rival and excludable and can thus be monetized. This allows private service producers to step into the market and compete with government services. 25 “Ex-ante" refers to "before the B2G data transaction is concluded; "Ex-post" means after the contract has been signed, during the implementation phase. 17 pools that facilitate exchanges, etc. For example the creation of "data stewards" 26 on the business and government side of the market could increase the volume of voluntary market-based B2G data sharing operations. Data cooperatives and pools, such as SODA ("Scalable Oblivious Data Analytics") and OPAL ("Open Algorithms") are trusted intermediary institutions that may enable a wider range of data suppliers and users to set up a common arrangement to share data, and to share the costs and benefits of this sharing arrangement. The OPAL initiative "sends the algorithms to the data, not the other way around" to avoid misuse and exposure of personal data. SODA seeks to establish a trusted intermediary that can carry out data analytics in an environment that requires a combination of data protection and multi-party data sharing and avoids unauthorised access to commercially sensitive data. The Frauenhofer Institute's Industrial Data Space 27 offers similar trusted intermediary facilities for multiple parties with secure data processing and strictly regulated access and billing. Reducing the lack of data-analysis skills in government may also reduce transaction costs but this does not matter so much when the data are delivered indirectly as a service rather than a direct raw data delivery. Third-party technical platforms as trusted intermediaries may facilitate data transfers, pooling, processing and management. As such they reduce transaction costs and risks for the parties involved in the B2G data transfer. However, technical platforms are not a substitute for reaching an agreement on access and use conditions between the parties (who has access to which data under which conditions) and the allocation of costs and benefits between the parties (who pays for what and who can claim which revenue or compensation). Such an agreement requires negotiations between the parties involved. Data suppliers to the pool or platform will have to find satisfactory arrangements that deal with uncertainties and potential costs associated with these uncertainties. In general, initiatives that reduce transaction costs and risks for data providers can help to promote voluntary B2G transactions. Sometimes a platform operator may impose a standard agreement to which all parties have to adhere. There are examples however where such standard conditions for B2G data sharing arrangements failed because of insufficient attention to ex-ante transaction costs and ex-post risks. One example is the Hitachi – Copenhagen City Data Exchange. Following a public tender, the city of Copenhagen provided seed financing to Hitachi to build a data sharing platform for many types of data between many providers and users. Hitachi became the platform operator and broker between data suppliers and users. It invested in reducing transaction costs for users by searching for data, formatting and preparing them for posting on the platform, ensuring access, protecting privacy and security, and bringing the data together in a single platform that made it easier for potential users to find what they were looking for. However, the platform could not set the price and access conditions to the data. The City imposed as entry requirement that any user would get access on equal conditions. That prevented suppliers and users from negotiating specific access conditions and prices for each user. Suppliers had no leverage over who could access the data. Consequently, ex-post risks and opportunity costs were not manageable in the platform. This made data providers reluctant to make their data available. The incentives for data sharing were too weak compared to the perceived risks. As a result the platform did not scale up (IDC, 2015, p15-19). 3.2.2 Quality: management of the transformation of the transferred data Data can be transferred in different varieties. The data can be transferred "raw", i.e. in the original format in which they were collected. They can be anonymized and/or aggregated to a more coarse-grained level from which the original raw data cannot be reconstructed. Data analytics can be applied to derive insights from the original data that can be sold as a service to the client. These variations in the data taxonomy can also be used to manage data governance for the purpose of reducing transaction costs and risk, or for reaching a negotiated agreement on data access and use conditions. Data are usually not consumer end products. We do not search online for a train schedule because we like looking at train schedules but because we want to purchase a transport service, i.e. a train journey. We therefore say that data are usually intermediate inputs into the production of a service or a service transaction. Data transactions create a link between an upstream data market and a downstream data-driven services market. This implies that there are two modalities for sharing or trading data: direct and indirect. With direct data trade the data holder delivers data as an input to a service producer. With indirect data trade 26 See the GovLab initiative on data stewards http://thegovlab.org/data-stewards-data-leadership-to- address-the-challenges-of-the-21st-century/ 27 See http://www.industrialdataspace.org 18 the data holder delivers a data-based service as an input for the downstream service provider. With direct trade, data are changing hands; with indirect trade there is no transfer or sharing of the original (raw) data. Many B2G data sharing examples circulated in the Expert Group are examples of indirect data trade. For example, using mobile phone data to help public transport companies improve their schedules or help tourism promotion services assess the presence of foreign tourists in a city. Only derived, aggregated and processed outputs are shared with a public body in these cases, not the original raw data. The debate in the B2G Expert Group also confirms that there are other dimensions too, apart from the direct-indirect access dichotomy. For example, data sharing can include site or remote access, sending an algorithm to the supplier to extract the required insights, sharing via a trusted third party, continuous direct access via APIs, among others. The distinction between direct and indirect data (service) trade is not always a crystal-clear black-and-white border line, as the train schedule example demonstrates. A search engine processes the underlying raw data of a train schedule to make them more readable and useful to the user who gets better insights into his travel options. Making them searchable by route, date and time and adding a ticket payment service to the search function creates an additional value-added service for consumers. The train schedule data transfer becomes a more indirect data service as more value-added features are added to the original raw data. In general, further processing of the data and adding other complementary services increases the value to the end user. The B2G Expert Group proposes a four level data taxonomy: raw or primary data, pre-processed (cleaned and structured data), processed (for a particular purpose) and data-driven insights (no data, only data services) 28 . Higher levels in the taxonomy may sometimes coincide with a higher degree of data aggregation and anonymization of personal data. Trusted intermediary data platforms can play an important role in reducing transaction costs and risks in a B2G data transfer by doing the processing in-house and transferring only insights or data-based services to the public body that is the end user. A good example of such a set-up is the Dalberg Group 29 , a firm that offers B2G data-based services and insights to public bodies in developing countries. These services are usually paid for by aid agencies. The processing-based approach to data taxonomy or data quality may give the impression that adding value to data is a cumulative linear process. In many cases however, value is added when datasets from various sources are aggregated and result in economies of scope (Rosen, 1983, Duch-Brown et al, 2017): the insights and value derived from analysing the combined set is greater than the value that could be gained from analysing each dataset separately. For example, the value of online advertising services can only be realized by combining consumer search data with data on the supply of ads offered by advertisers. Using each dataset separately would not permit setting up a targeted advertising service. Another example are urban transport and mobility platforms that require data aggregation across several transport service providers (trains, buses, metro, taxis, etc.) in order to manage peak transport demand in dense urban areas. If economies of scope are important, the choice between a direct or an indirect data service depends on which party is in a better position to do the aggregation, the original data holder or recipient. In a B2G setting, governments may want to combine acquired data with other datasets that they already hold. For example, they may want to combine mobility data with infrastructure, land ownership, income tax or education data. In that case a trusted intermediary data platform (OPAL, SOPA etc.) may be an appropriate solution to deal with the complexity of data confidentiality, strictly regulated access conditions, etc. (see below under transaction costs). 3.3 Externalities and incentive misalignment In this section we analyse the impacts of externalities and incentives on B2G data sharing initiatives. 28 The Expert Group’s data taxonomy specifies that data, information and insights are not synonyms and that the latter are generated by analysing information and drawing conclusions. "Insights" also consist of data points that are shared with a recipient. The value added by the service resides in the processing and analysis of the original primary data that has led to these insights. The original data can usually not be reverse engineered from the insights. From a mathematical perspective some data entropy has been lost in the processing. Extracting "insights" can therefore be seen as reducing the entropy of a dataset. See https://en.wikipedia.org/wiki/Kolmogorov_complexity 29 See https://www.dalberg.com/ 19 3.3.1 Negative externalities and risks Firms may be reluctant to share data with governments because it might have a negative impact on them. This could be due to perceived risks of market regulation and competition in their upstream or downstream markets. An example of perceived regulatory risks comes from our own experience at the JRC. When we sought to procure data from parcel delivery firms about cross-border deliveries of e-commerce packages in the EU the firms refused to provide the data because they feared they might be used by the Commission to impose new regulatory measures on the sector. Indeed the Commission was exploring these questions as part of its Digital Single Market policy package and eventually implemented new policy measures. Another example from the JRC experience concerns the procurement of data on app use from a private firm. The firm imposed strong purpose limitation conditions that would not have allowed us to use the data in a publicly available document, not even in an aggregated way. That stopped the transaction from taking place. It is often assumed that commercial risks only occur in B2B data sharing situations, not in B2G settings because public bodies are normally not in the business of competing with private firms. However, some public bodies may run commercial services that compete with private services, for example in public transport and utilities. Public bodies may also pool data from various private and public sources and share the insights with competing parties. Mobility-as-a-Service (MaaS) urban transport platforms (Carballa, 2018) are a good example of conflicting incentives for private firms to share data with public bodies. MaaS platforms combine information about, and access to, multiple transport service providers in a single interface. This requires private and public transport operators to share data with a publicly-operated platform. In doing so, these transport operators cooperate and compete at the same time: they compete over rides and market shares in transport while they collaborate to build a common data pool and service that can bring them all more rides. Carballa (2018) shows that, although every transport service provider has incentives to share data provided that a critical mass of competitors do likewise, the cooperative-competitive dynamics of data sharing can lead to situations where some transport operators will not have an incentive to share their data because they may lose market share and revenue. 3.3.2 Positive externalities and economic incentive misalignment While B2G data sharing may produce benefits for society at large or for particular target groups of the operation, the private firm that supplied the data may not experience any of these benefits directly. In other words, a B2G data transfer for a project that is in the public interest generates positive externalities that often cannot be directly monetized by the data supplier. If the data collection and transfer to a public sector body is the firm’s main business activity, there would be a misalignment of economic incentives. It would create a gap between private incentives to supply data to public bodies and the social benefits of the operation that do not accrue to the data supplier but are dissipated among the population at large. There is a long-standing economics research literature on the private provision of public goods 30 (see for example Bergstrom et al, 1983) that discusses in detail these incentive misalignments due to uncaptured externalities from public goods. It also explains that the private provider may have a personal preference for the provision of a public good and derive benefits from doing so. In that case it adds to his benefits and not only to his costs. In B2G data sharing operations this can be translated into Corporate Social Responsibility (CSR) and philanthropic motives for private firms to contribute data in the public interest. In that case firms may voluntarily choose to ignore the incentive gap between private costs and social benefits for CSR motives. Private contributions may be forthcoming but they are unlikely to reach the socially desirable volume of public goods production. Hence government intervention may be required to reach this socially desirable level of public good provision. The Com SWD (2018) and the Expert Group have invoked the concept of proportionality in an attempt to contain the gap between private costs and public benefits of B2G data sharing. Proportionality compares the costs for the data provider with the benefits for society. The SWD suggests that the ratio of costs to benefits should be kept within reasonable proportions. For example, the extent of fine-graining of the data should not 30 These are goods that are non-rival but also non-excludable. This means that the use by one individual does not reduce availability to others and that individuals cannot be excluded from use. See https://en.wikipedia.org/wiki/Public_good 20 exceed what is necessary for the purpose of the public project because that would make costs for the supplying firm disproportional to the benefits for society. In general, any activity in the public sector should meet the criterion that total benefit to society should be greater than the cost; if not then the activity is not social welfare enhancing for society. However, the point in B2G data transactions is not only to compare the total costs and benefits for society as a whole but also to examine who bears the costs and who enjoys the benefits. In the case of B2G data transfers, costs and benefits may be attributed to different parties: the private costs are born by the data provider while the social benefits are enjoyed by society as a whole or a selected target group. There are several mechanisms to compensate the costs incurred by private data providers that are ultimately a political decision. 21 4 How to overcome these market failures and boost welfare gains? There is a nascent market for B2G data transfers that contribute to the promise of socio-economic welfare gains from data sharing. However, most of these remain one-off experimental pilot projects that do not seem to be sustainable over time and the emerging market does not seem to scale. There are cases where the market fails to realise B2G data sharing transactions that are "in the public interest" i.e. that increase social welfare outcomes for society, often in cases where firms are not in the business of directly selling their data or making them available to third-parties. Provided the social welfare benefits of the operation exceed the private costs for the data provider (data production costs and possible adverse effects in the market), it would be in the interest of society to nevertheless go ahead with the data sharing transaction. That requires finding a means to overcome the obstacles to the transaction. If there is no operational market for the data yet, how can this missing market be created? In this note we discuss three types of obstacles: monopolistic data suppliers and monopolistic pricing, high transaction costs and risks for the data supplier, and incentive misalignment between the private provider and the public interest. We also suggested some potential solutions to these obstacles, including intermediary service providers that could reduce transaction costs and risks, and managing the taxonomy or quality of the data transfers to avoid transmission of personal or commercially sensitive data. One mechanism to solve some of the barriers that may be holding back B2G data sharing initiatives would be to provide stronger incentives for the data supplier firm to engage. These incentives can be direct (i.e., monetary) or indirect (i.e., reputational). Another way is to ascertain the data transfer by making the transaction mandatory, with cost compensation (e.g. marginal costs or marginal costs plus fair ROI). Mandatory data sharing and/or price regulations are an example of this second solution. An intermediate way would be solutions that seek to facilitate voluntary B2G operations without mandating them, for example by reducing the transaction costs and perceived risks for the provider data supplier with trusted data intermediary platforms, or appropriate contractual provisions, for instance. A possible EU governance framework for B2G data sharing operations could cover all these options. The main issue then is to set criteria for which transactions fall into which type of B2G governance box. For situations with monopolistic data supplies and high monopolistic pricing of the data, or refusal of access to the data, a preferential pricing rule and/or mandatory access to the data could be regulated, e.g. by introducing a specific and mandatory regulation for the purpose of B2G data transfers, including a (zero or marginal cost) pricing rule. There are already mandatory B2G data transfers in place. For example, governments make it mandatory for firms to hand over some of their accounting data for tax purposes, fill in questionnaires for statistical purposes and report on their compliance with environmental regulations, and banks have to provide detailed reports on their credit portfolios to financial supervisors. In the EU, DG COMP already has powers to force firms to hand over data in the context of specific competition cases but also for wider sector enquiries. This was used for example for collecting data in the 2017 E-commerce sector enquiry 31 . DG GROW recently tried to obtain similar powers in the context of the implementation of Single Market policies (the so-called SMIT proposal) but failed. A mandatory regulatory framework of B2G data transfers under specific conditions could thus be considered. This could also cover situations where a private data supplier faces strong negative incentives to participate in a B2G data transfer, despite the fact that the public interest or social welfare impact of an operation would considerably exceed the private costs of the supplier. If these private costs would not undermine the main business model of the supplier, a case could be made to move ahead with the operation and make it mandatory. Still, it would require some legal and possibly judiciary oversight on a case-by-case basis 32 . 31 See http://ec.europa.eu/competition/antitrust/sector_inquiry_final_report_en.pdf 32 In particular for its compatibility with rules on IP ownership and the protection of trade secrets. 22 References Alemanno, A. (2018), Data for Good: Unlocking privately-held data to the benefit of the many. European Journal of Risk Regulation 2. Auerbach, A. and M. Feldstein (1985), Handbook of public economics Bergstrom, T., L. Blume and H. Varian (1986), On the private provision of public goods, Journal of public economics. Carballa-Smichowski, B. (2018), Determinants of coopetition through data sharing in MaaS. Management and Data Science, vol. 2(3) 2018. Diewert, W. (1983), Cost-benefit analysis and project evaluation: A comparison of alternative approaches, Journal of public policy. Duch-Brown, N., B. Martens and F. Mueller-Langer (2017), Data ownership, access and trade, JRC Digital Economy Working Paper 2017-01. Deloitte (2011), Pricing Of Public Sector Information: Models of Supply and Charging for Public Sector Information, DG CNECT, October 2011. Drexl, J. (2016), Designing competitive markets for industrial data: Between propertisation and access, Max Planck Institute for Innovation and Competition Research Paper No. 16-13. Elkin-Koren, N. and M. Gal (2018), The Chilling effect of governance by data on innovation. University of Chicago Law Review 86. European Association of Remote Sensing Companies (EARSC) (2015) Copernicus Sentinels’ Products Economic Value: A Case Study of Winter Navigation in the Baltic, September 2015, available at http://earsc.org/news/between-24m-and-116m-per-annum-of-economic-value-generated-thanks-to-the-use- of-satellite-radar-images European Data Portal (2015), Creating value through open data. International Data Corporation and the Lisbon Council (2016), Opening Up Private Data for Public Interest, a European Data market study. Graef, I., S. Y. Wahyuningtyas and P. Valcke (2015), Assessing data access issues in online platforms, Telecommunications Policy vol 39, pp 375–387. Jones, C. I. and C. Tonetti (2018), Nonrivalry and the Economics of Data . Stanford Graduate School of Business Working Paper 3716. Newbury, D., L. Bently and R. Pollock (2008), Models of public sector information provision via trading funds Martens, B. and F. Muller-Langer (2018), Access to digital car data and competition in aftersales services, JRC Digital Economy working paper 2018-06. OECD (2015), Data-driven innovation: Big data for growth and well-being, OECD, Paris. http://dx.doi.org/10.1787/9789264229358.en Palfrey, J. and U. Gasser (2012), Interop: The Promise and Perils of Highly Interconnected Systems. Panzar, J. and R. D. Willig (1981), Economies of Scope, American Economic Review Vol. 71, No. 2, pp. 268-272 Pollock, R. (2007), Cumulative innovation, sampling and the hold-up problem. MPRA Paper 5022, University Library of Munich, Germany. Rochet, J. C. and J. Tirole (2006), Two-sided markets: a progress report, Rand Journal of Economics, 37(3), 645–667. Rosen, S. (1983), Specialization and human capital, Journal of Labor Economics, 1(1), 43–49. Verhulst, S., A. Young and P. Srinivasan (2015), An introduction to data collaboratives, New York GovLab. WISE DEMOS (2011), Big and open data in Europe: a growth engine or a missed opportunity? World Bank (1998), Handbook on the economic analysis of investment operations. 29 Annex 2. Demand and data pricing regimes How does pricing relate to production costs in a monopolistic market? This is explained in Figure A2. Typically for the digital economy, data producers face high fixed costs in setting up the data collection, storage and processing system. On the other hand, data systems often scale easily so that the marginal cost of processing an additional unit of data is very low or close to zero. It results in declining average costs. A profit-maximising firm with some degree of monopolistic market power will set prices at a profit-maximising level (Pmax where P-average cost is maximum) where it will sell a quantity Q1. In the presence of fixed costs and if a firm operates in a competitive market with little or no market power, it will price its products close to or equal to average cost (Pavg) and sell quantity Q2. In the absence of fixed costs, market pricing could go down to marginal costs (Pmc) and quantity Q3, or possibly Pmc plus a mark-up (also called a return-on-investment or ROI). The maximum quantity is reached when firms give their products away for free (Pzero). Total demand is satisfied only at Pzero, i.e. all potential users get access to the data at this point. In the presence of fixed costs, a preferential price is any price below Pmax (for a firm with market power) or below Pavg (for a firm without market power). Without fixed costs, any price below Pmc is a preferential price. The meaning of preferential pricing depends on the market structure and production costs. When the data market is monopolistic and the price is fixed at Pmax, consumer and producer welfare are defined by the blue and yellow areas under the demand curve. The difference between Q4 and Q1 represents the social deadweight loss, i.e. a number of potential data uses that are not addressed. In a competitive data market (either at Pavg or Pmc) a larger number of potential users are served (Q2, Q3 > Q1) and social welfare from the data expands, to the benefit of data users. Figure A2: Demand for data at different price regimes Source: Adapted from Newbury et al, 2008. GETTING IN TOUCH WITH THE EU In person All over the European Union there are hundreds of Europe Direct information centres. You can find the address of the centre nearest you at: https://europa.eu/european-union/contact_en On the phone or by email Europe Direct is a service that answers your questions about the European Union. You can contact this service: - by freephone: 00 800 6 7 8 9 10 11 (certain operators may charge for these calls), - at the following standard number: +32 22999696, or - by electronic mail via: https://europa.eu/european-union/contact_en FINDING INFORMATION ABOUT THE EU Online Information about the European Union in all the official languages of the EU is available on the Europa website at: https://europa.eu/european-union/index_en EU publications You can download or order free and priced EU publications from EU Bookshop at: https://publications.europa.eu/en/publications. Multiple copies of free publications may be obtained by contacting Europe Direct or your local information centre (see https://europa.eu/european-union/contact_en).