6
Economic Analysis
6.1 INTRODUCTION
This chapter summarizes what economic theory has to say about how the terms on which government obtains data from, or provides data to, users—including through licensing or purchase—affect the well-being of society. The chapter is divided into five sections: first it describes the factors that affect whether data are likely to be produced and distributed efficiently; second, it emphasizes that licensing is just one of several options for government data procurement, and explains why some options are better suited to particular agency missions than others; third, it examines the advantages and disadvantages of licensing as a tool for achieving economic efficiency; fourth, it describes the economic considerations when agencies decide to negotiate licenses; and fifth, it discusses how agency decisions to charge a user fee may promote or hinder economic efficiency.
6.2 ACHIEVING THE GOAL OF ECONOMIC EFFICIENCY
The first step in any economic analysis is to specify goals: What is society trying to accomplish? Like most commentators, we identify economic efficiency as an important goal. In everyday life, economic efficiency has many different meanings. To an economist, however,
economic efficiency requires that there remain no unexploited opportunities to make someone better off without making someone else worse off.1 In the context of the provision of information,2 this subsumes two criteria. First, efficiency in production requires that society create information if and only if its cost is less than its combined value to all users. Second, efficiency in distribution requires that information be available to all users who value it at or above the marginal cost of distribution.3 For reasons discussed below, it is rarely possible to completely achieve either, let alone both, of these objectives. Nonetheless, society has developed a number of institutions that move it closer to their achievement. Where benefits are not quantifiable, one may have to be satisfied with achieving the agency’s goal (quantitative or not) at minimum cost.
Unlike most commodities, information goods such as geographic data have the property that once generated they can be used by one user without reducing the amount available for use by others. As previously noted, efficiency in production requires that society produce information if and only if the information’s combined value to all users exceeds the cost of production. For example, if user A values a particular piece of information at $5 and user B values the same information at $3, it is economically efficient to produce the information at any cost less than $8. For many types of information, the value to one user is independent of the number of other users. In these cases, the information’s value to society is simply the sum of its value to each individual user.4
Once information has been produced, economic efficiency requires that it be made available to any user who values it at or above the cost of dissemination. This condition can be met as long as the price charged to any individual is no greater than the amount that individual would be willing to pay for the information but no less than the marginal cost of distribution. Where the price exceeds the cost of distribution, however, the number of consumers who acquire and use the information is likely to be inefficiently reduced. When this occurs, the result is a deadweight loss because some members of society lose benefits without there being any gains to others.
It is also important to recognize that users of geographic data include not only final consumers but also other producers of information products or services, so that efficient distribution is likely to contribute to efficient production.5 In particular, the ability of other vendors, academic researchers, and government agencies to use geographic data in their own research and development activities may substantially enhance their incentives to invest in the creation of valuable new geographic products and services. Unless the conditions for efficient distribution are met, therefore, not only will some consumers fail to obtain information that they value at more than its incremental cost but the costs for some producers also will be increased. Because the production of information goods tends to be cumulative, with one creator building on the efforts of others, inefficient distribution may therefore raise the cost of subsequent research and development.6 Conversely, later innovators have lower costs when they can build freely on the work of their predecessors.7 Fostering the creation and development of geographic data and services requires striking the right balance between proprietary rights and free or open access to information.
5 |
See examples of the impacts of contrasting dissemination regimes on production described in Chapter 4, Section 4.3. |
6 |
The costs to later developers can be reduced either by limiting the rights of earlier ones or by lowering the costs of transactions between them. In Chapter 9, we discuss mechanisms that might lower transaction costs. |
7 |
The impact of intellectual property protection on the costs of innovation is discussed, for example, by W. M. Landes and R. A. Posner, 1989, An economic analysis of copyright law, Journal of Legal Studies 18:325–363; and S. Scotchmer, 1991, Standing on the shoulders of giants: Cumulative research and the patent law, Journal of Economic Perspectives 5:29–41. For a different view, see E. W. Kitch, 1977, The nature and function of the patent system, Journal of Law and Economics 20:265–290. |
Even if it were possible to achieve either efficiency in production or efficiency in distribution—no small task—achieving both simultaneously is even more difficult.
Permitting producers to charge high prices increases their incentives to produce information. At the same time, high prices exclude some would-be consumers from the market.8 Thus, the benefits of encouraging additional innovative activity generally must be balanced against the social costs of discouraging efficiency in distribution. Although high prices may be necessary to encourage the production of information, high prices create inefficiencies by excluding users who are willing to pay at least the cost of distribution but not the prices being charged by producers.
6.2.1 Other Sources of Inefficiencies in Information Markets
The deadweight loss that results when the price of information exceeds the cost of distribution is not the only type of inefficiency that can occur in a market for information. This section describes six additional inefficiencies that also can contribute to deadweight loss.
First, information is an “experience good.” This means that a potential buyer may be unwilling to pay for a particular piece of information before inspecting it because the buyer cannot place a value on it. On the other hand, the buyer also may be unwilling to pay afterward if inspection reveals all of the needed information.9 This problem can be overcome in cases where a small subset of data is enough to demonstrate quality. Such is typically the case in the geographic data market. Furthermore, users who repeatedly sample without buying eventually will be denied the opportunity to sample, and suppliers who attempt to charge high prices for information of limited value eventually will find that their customers go elsewhere. However, sampling and the use of reputation
may be unable to eliminate completely the inefficiency that arises from this source.
Second, there may be strong incentives for users to underreport the value of information in order to obtain lower prices from the producer. If a large proportion of users succeed in this strategy, however, the information may not be created at all or the amount may be inefficiently small. In practice, high-volume users often accept the fact that, because small-volume users will “free ride,” they must pay a disproportionate share of the cost of producing information. As long as the contributions of these users are sufficient to cover these costs, the information will be produced, although perhaps less will be produced than if other users had also paid to do so.
Third, when different users place different values on the information, vendors may have to charge different prices to each user, that is, practice price discrimination. This explains why some vendors have attempted to ascertain how individual users will employ their data.10 In the case described earlier, the only single price at which both A and B will purchase the information is $3. However, if the cost of producing the information exceeds $6 and price discrimination is impossible, no private entity would supply it despite the fact that it would be efficient to do so as long as the cost of production is less than $8. Even though users may try to disguise the true value of the information to them, licensors sometimes can use objective criteria to identify consumers willing to pay higher prices.11 Nonetheless, effecting price discrimination is generally quite difficult and, even where it is accomplished, it is unlikely to satisfy fully the conditions for efficient production and dissemination of information.
Fourth, it may be hard to exclude nonpaying users, especially when paying customers choose to share information. That is, it may be impossible for producers to prevent free riding once their information goods have been disseminated to a large number of users. Vendors can respond by (1) suing those who engage in or facilitate such behavior, (2) using technical protections such as providing software in copy-protected form or requiring that playback devices be designed to limit copying, (3)
10 |
For example, see footnote 44 of Chapter 3. |
11 |
For example, academic journal publishers typically impose higher subscription rates on libraries than on individuals. Similarly, music-performing license organizations impose higher fees for performing rights on establishments such as nightclubs and bars with large revenues than on the same types of establishments where revenues are smaller. Finally, book publishers charge higher prices for hardbacks than for paperbacks, presumably because those who wish to obtain a book as soon as it is published are willing to pay higher prices. |
obtaining indirect compensation by imposing levies on recording media or devices,12 (4) pricing products at low rates that discourage unauthorized copying, or (5) imposing high fees that capture some of the value that otherwise would be lost to sharing.13 Nonetheless, some free riding is likely to occur, with the result that efficiency in the production of information may not be fully achieved.
Fifth, a special difficulty may be created when users require access to information that is owned by different entities in order to have anything of value, a difficulty variously known as the “complements” or “anticommons” problem. In these circumstances, each individual owner may attempt to capture a disproportionate share of the combined value of the information. However, if all owners attempt to do this, the transaction may never occur, or occur at an inefficiently high price, so that no value is created, or the value that is created is inefficiently low.14 Recognizing this problem, individual producers may attempt to prevent or discourage such behavior by internalizing the distribution of profits, perhaps by merging, or by forming joint ventures or consortia. Alternatively, as in the case of patents, producers may cross-license or form patent pools in which each producer agrees to make licenses to its patents available at little or no cost in return for an agreement on the part of other producers to do the same. Nonetheless, these institutions may deal only imperfectly with the complements problem.
Finally, transactions may not occur because the task of negotiating and administering contracts entails high transactions costs. Transactions involving information goods may be especially complicated (and expensive) compared to conventional sales in which the vendor transfers complete title at the time of payment.15 Strategies for reducing transacttion costs include (1) widespread adoption of standardized contracts and other business practices, (2) blanket licenses that give the licensor access to all of the works in the owner’s collection in return for a fixed fee,16
and (3) creating centralized marketplaces or other institutions where buyers and sellers can easily find each other. Finally, producers might accept, or even promote, limitations on their own intellectual property protection if, in return, that also will increase their access to the information of others.
6.3 THE CHOICES: GOVERNMENT MISSIONS AND PROCUREMENT OPTIONS
Licensing is one of several tools available to agencies when they procure geographic data,17 and consideration of all options is important in any economic analysis. The suite of workable options in each case is determined by what the agency is trying to achieve, and will be influenced by whether the agency’s mission requires broad redistribution, limited redistribution, or only internal use. In practice, most of the potential policy benefits from licensing tend to involve the last two categories. Here, government agencies often can obtain useful (though limited) rights at prices that are generally below the costs of outright purchase or in-house production.18 However, the savings may be small if there are relatively few limits on the rights that are transferred. As a result, licensing is unlikely to be chosen when an agency’s mission involves acquiring data that it wishes to disseminate to the public as a whole without restriction.
6.4 WHEN IS LICENSING ECONOMICALLY EFFICIENT?
In this section we weigh the advantages and disadvantages to a government agency of obtaining data from the private sector through
|
include the American Society of Composers, Authors, and Publishers and Broadcast Music, Inc. (performance rights for music) and the Copyright Clearance Center (reproduction rights). For a (somewhat dated) description of these organizations, see S. M. Besen and S. N. Kirby, 1989, Compensating Creators of Intellectual Property: Collectives That Collect, R-3751-MF, Santa Monica, The Rand Corp. See also J. H. Reichman and P. Samuelson, 1997, Intellectual property rights in data? Vanderbilt Law Review 50:51. |
17 |
See Chapter 4. |
18 |
Recall that we have defined purchase to mean that there are no limits on the rights of the acquirer to use or transfer data. |
licensing, as opposed to purchasing the data from the private sector or producing the data in-house, using economic efficiency as our criterion.
6.4.1 Economic Advantages of Licensing
Licensing data from the private sector is likely to be economically efficient when government does not have enough information to decide what data society needs. In this case, entrepreneurs may have better or different information about which investments make sense. Private markets are a powerful mechanism for eliciting and organizing information held by large numbers of scattered actors, in which case government may be just one of a large number of customers for the data that are produced.
When private firms have superior information about potential demand for data, it is likely to be economically desirable to decentralize the investment decision to dozens and perhaps hundreds of private-sector entrepreneurs.19 Moreover, large specialized private-sector data producers may benefit from economies of scale and scope that might not be available to individual government agencies for which data acquisition are a small part of their missions. Markets also guard against the possibility that government-funded programs will take on a life of their own; forcing actors to invest their own money eliminates the danger of program managers, employees, or outside contractors acquiring a personal stake in continuing unproductive programs. This problem is likely to be reduced, however, when society’s need for a particular dataset enjoys widespread consensus (e.g., data for basic navigation or emergency response), in which case outright purchase or in-house production may be preferred.
The foregoing analysis is quite general, and there may be additional economic efficiencies from acquiring data with licensing restrictions in specific circumstances. We provide four examples.
First, acquiring data through outright purchase or in-house production in order to avoid restrictions on use may not be efficient if the number of potential users is small. In such cases, users who are most directly affected are likely to press an agency to be able to distribute the data freely even when the benefits they receive are smaller than the costs that the agency incurs in being able to do so. An important corollary is that, from a purely logical standpoint, agencies should fund data acquisition at the smallest
level of government that embraces all potential beneficiaries,20 although it may be difficult to apply this principle when, as is often the case, the potential beneficiaries cannot be identified in advance.
Second, agencies may find it efficient to accept licensing restrictions that allow them to “piggyback” data collection or distribution activities onto commercial programs even where the result is that the agencies do not acquire the precise data that they need. One example of the synergies is the NEXTMap Britain project, which supports its mapping activities in Britain with revenues from private and public sources.21 Another example is the National Aeronautics and Space Administration’s partnership with Orbital Image Corp. on the Sea-viewing Wide Field-of-view Sensor satellite data.22 License agreements are a natural way to implement such government–industry collaborations and can result in substantial cost reductions, albeit at the cost of some compromises in the data that are available to an agency.
Third, there may be insufficient funds in an agency’s budget to support every worthwhile acquisition. In this case, the agency may decide that acquiring data through a license that provides very limited redistribution rights—despite its drawbacks—may be appropriate.
Fourth, although in theory every agency has an incentive to coordinate purchases with others in order to lower its costs, it may be prohibitively expensive to achieve the necessary coordination. Even where coordination is feasible, some agencies may be tempted to ignore potential savings because of institutional rivalries. In either case, the government may end up acquiring redundant or unnecessarily expensive data and some data may not be acquired at all. There may be cases where carefully crafted licenses, in which each of a number of agencies acquires limited rights in data, can overcome these difficulties.
6.4.2 Economic Disadvantages of Licensing
The principal economic disadvantage of obtaining data through licensing, rather than through outright purchase, is that the licensing restricttions limit the ability of an agency to distribute the data it has acquired in an economically efficient manner. Whereas efficiency in distribution
20 |
At the limit, local governments would encourage very small groups to organize on their own. |
21 |
That is, an insurance company (Norwich Union), and the U.K. Environment Agency (testimony of Michael Bullock, Intermap). |
22 |
See Chapter 4, Section 4.2.1. |
requires that no user who is willing to pay at least the incremental cost of distribution is excluded, that may not be possible when there are limits on the ability of any agency to redistribute data that it has acquired from the private sector. For this reason, licensing almost always sacrifices distributional efficiency by leading to overpricing (and undersupplying) geographic data. The amount of actual exclusion depends on individual circumstances. This effect is particularly important in cases where demand is price sensitive, that is, where many users are likely to drop out of the market each time prices move upward. Except for the rare case when every user is identical, or perfect price discrimination is possible, “deadweight loss” is unavoidable.
Government can almost always avoid the deadweight loss associated with exclusionary pricing and inefficiencies in distribution of information by using methods to procure data that permit unrestricted redistribution and providing these data to any user willing to pay the marginal cost of reproduction. Avoiding inefficiencies in production is likely to be more difficult. There is a broad consensus that governments should play a large role in the provision of geographic information, although the line between private and government provision remains contentious. The main challenge is to decide whether the advantages of licensing (efficiency in production) outweigh its drawbacks (efficiency in distribution). In the words of Thomas Jefferson, agencies should be careful to “draw…a line between the things which are worth to the public the embarrassment of an exclusive patent, and those which are not.”23
6.5 NEGOTIATING THE LICENSE STRUCTURE: ECONOMIC CONSIDERATIONS
If an agency decides to acquire data through licensing, it must still negotiate terms. In this section, we note some economic considerations for the negotiations. The challenge facing agencies is to maximize the difference between benefits and costs. The nature of benefits will depend on the agency’s goals, but would include the ability to use the data to fulfill the agency’s mission at the lowest possible cost. The costs include royalties paid, the deadweight loss resulting from restrictions on government’s ability to use and redistribute the data, and contract administration expense.
Like any private-sector business, agencies should insist on paying the lowest possible price for data, and should be willing to haggle or invoke
competitive bidding to do so. Agencies should not attempt to ensure that private-sector firms receive sufficient royalties to “cover their investment” in data. Not only is such a policy unfair to taxpayers, but it sacrifices the supposed efficiency of using markets in the first place.
Agencies also must be willing to present creative solutions in which the “standard” package of license rights is adjusted to avoid paying for rights that the agency does not need. From a practical standpoint, agencies should seldom, if ever, pay more to license commercial data than to independently re-collect the same information.24 Although this principle might seem to invite wasteful duplication, the threat will almost never be carried out. Instead, rational vendors will adjust their prices to stay competitive.25
Agencies can sometimes obtain reduced license fees in return from providing nonmonetary incentives, such as access to one-of-a-kind resources, to private licensors. Examples include access to raw and processed data in government archives, coordination with existing agency research programs, access to skilled agency employees, access to agency facilities, and possible synergies between private and public research agendas. In the New Economy, online “content” and the “ability to attract eyeballs” have also become assets. Agencies with prominent Web sites may be able to extract price concessions in return for posting advertisements or links to private vendors.26
6.6 LICENSING GOVERNMENT DATA TO THE PRIVATE SECTOR
We have already explained why efficient distribution requires government to make data available to all users who are willing to pay at least the marginal cost of distribution.27 Nonetheless, there may be practical
24 |
Agencies occasionally may pay a premium when there is no time to collect an independent dataset. With appropriate planning, such cases should be few and far between. |
25 |
More specifically, a rational vendor understands that “sunk costs are sunk.” It is better to sell data at a loss than not to sell them at all. |
26 |
Advertising and links are a dominant source of revenue for private Web sites. In 2002, Yahoo earned roughly 63 percent of its first-quarter revenues ($192,700,000) from advertising (Subscriber Fees a Boost for Yahoo, San Francisco Chronicle, April 11, 2002, available at <http://sfgate.com/cgi-bin/article.cgi?f=/chronicle/a/2002/04/11/BU78117.DTL>. |
27 |
Agencies also should avoid restrictions on users’ ability to create extensions and improvements to their data. This is true even when—as is usually the case— |
obstacles to achieving this objective. First, transaction costs may make it impractical (i.e., uneconomical) to collect any fee, with the result that some users may obtain data that they value at less than the marginal cost of distribution.28 Second, government databases sometimes include data that have been licensed from the private sector and the restrictions in those licenses prevent pricing at the marginal cost of distribution. Third, legislation or budget constraints may require some agencies to charge user fees that attempt to cover all or a portion of their data acquisition costs. Subject to these constraints, agencies should attempt to achieve efficiency in distribution by avoiding excessive user fees and unnecessary use restrictions.29
6.7 SUMMARY
Society makes geographic data investment decisions through two very different institutions: governments and markets. Deciding which sector should acquire and distribute a particular product has profound implications for economic efficiency. This chapter has reviewed the strengths and weaknesses associated with each institution. In general, markets are a good solution when the initial decision to invest in a particular product is controversial or uncertain. Conversely, government procurement is most useful when uncertainty about whether to invest is small, so that distributional efficiency becomes the dominant concern. Beyond these generalizations, additional considerations may apply to particular cases.
Agencies affect the government/market balance each time they acquire or distribute data. The challenge is to make these choices consciously with an eye toward economic efficiency. License design can be an important tool for setting this balance. For example, suppose that an agency
|
intellectual property rights protect the extensions and improvements [see footnote 7 in this chapter for related references]. Consumers who choose to pay for “improved” public data are better off than they would be in a world where intellectual property rights did not exist. Additionally, when government makes data widely available, the data products of would-be competitors are more likely to be competitively priced. The threat of actual or potential competition is a powerful constraint on the price that consumers pay for commercial data and services. |
believes that certain geographic data products have uncertain value, so that investment decisions are best left to private markets, and that vendor pricing will not seriously limit society’s use of whatever data are produced. Under these circumstances, the agency usually will wish to acquire data from private producers through licenses that give the agency modest use and redistribution rights. In general, such licenses promote markets by allowing the original suppliers to pursue additional sales due to the diminished likelihood that data provided to government will become widely available to others through actions of government.
Suppose, on the contrary, that the agency thinks that certain geographic data products have proven their worth, but that high prices are preventing many people from using them. In this case, the agency may wish to make the data it acquires widely available by acquiring them through licenses that give the agency broad redistribution rights.30 Although such rights will limit any remaining private market for the product, that will be reflected in the price that the vendor demands, and the agency pays, for such a license. In return, efficiency in distribution is more likely to be achieved.
Finally, traditional licensing models are not the only—or, in some cases, the best—ways to promote economic efficiency. Other forms of private–public partnership are also possible. For example, some firms will not mount large data acquisition programs unless government agencies contribute resources. This can be done in a variety of ways, including cooperative research and development agreements, private–public partnerships, and licenses that obligate the agency to buy large volumes of data. From the agency perspective, such transactions may offer a good mix of efficiency in both production and distribution. Efficiency in production is achieved because the project must still realize significant commercial sales to be profitable. Efficiency in distribution occurs because the agency often has significant leverage to demand license terms that permit widespread dissemination on favorable terms, perhaps by requiring the vendor to donate its data to the public after a fixed period of years, or otherwise limiting the private partner’s ability to impose high prices. Alternatively, government might bear the entire cost of data production and acquire unlimited rights in order to promote efficiency in distribution.
VIGNETTE F.
A MAINSTREAM GEOGRAPHIC DATA MARKETPLACE DREAM
The geographic data marketplace, commonly referred to as Spatial Mart, has become a well-organized, convenient, and economical channel for reaching millions of customers. For many vendors, it is their primary distribution channel. Spatial Mart is a rich environment for commercial innovation and has gathered a critical mass of comprehensive content to meet broadening consumer demand. Innovators now focus on bringing new data products and services quickly and inexpensively to market rather than on maintaining complex selling and licensing arrangements.
All geographic data indexed by Spatial Mart adhere to licensing and metadata creation standards that ease automated search, retrieval, and financial functions. Additionally, Spatial Mart automatically tracks licensing terms in each transaction, and manages licensee payments and royalty distributions for any original and derivative works. Because of this, some commercial data suppliers now allow free downloads of substantial quantities of geographic data for product development purposes. These vendors’ primary source of revenue is from commercial redistribution of derivative products. Thus, hoping that profitable uses of their data will arise, they allow free experimentation with their data and its incorporation in other products and services licensed through Spatial Mart.
Bob Nathan, who runs a trucking company, is searching Spatial Mart for all available geographic data meeting his technical and geographic coverage needs. After selecting a product, he clicks on a combination of standard licensing paragraphs to define his preferred data usage rights. These choices address such matters as the ability to use, alter, and disseminate the data, or to further create, license, and sell derivative products. Commercial vendors have already established prices that depend on various combinations of standard provisions and the volume and geographic coverage requested (data with few or no use restrictions typically are priced higher than those carrying substantial restrictions). Consequently, within seconds, Mr. Nathan receives a price quote and proceeds to the online purchasing step where he can comparison-shop—much like in the online purchase of airline tickets—if he chooses. He is satisfied with the quote in comparison with the other options, and acquires the data.
Had Mr. Nathan concluded that no offerings satisfied his desired combination of price and licensing terms, he would have been asked to offer the price, terms, and technical specifications he would accept. Any vendor interested in supplying existing data meeting Mr. Nathan’s preferences could accept his offer—much like on priceline.com. Had Mr. Nathan needed custom data products or services rather than existing ones, Spatial
Mart’s electronic bidding system would have recorded his technical and licensing preferences, any additional purchaser/vendor contract preferences, and then invited qualified vendors to submit their bids by a specified date.
Bob’s dream is this: Can an operational infrastructure supporting efficient licensing and transaction interactions be developed that is open to all sellers and buyers of geographic data and services and will support an active and thriving marketplace?