Pricing Government Information

Assuming that the good should be produced at all, marginal cost pricing insures that there will be an efficient level of output, based upon values expressed by a willingness to pay.[2] By pricing the good at the cost of producing the last unit, the rule insures that output will expand until there isn't anyone who values the good enough to justify an additional unit of production.

In the context of the dissemination of information, the term marginal cost pricing is commonly used to describe a policy of charging users only for the costs of disseminating information, without charging for the costs of collecting, editing, and organizing the information, since these costs are fixed, and do not vary with the number of data users.

In practice, marginal cost pricing is not entirely straightforward. One approach is to only include costs associated with the dissemination of the information. Another approach is to only charge the marginal cost of the dissemination of information. If an online system is used, the marginal cost of dissemination is likely to be quite low, approaching zero. Because of the large economies of scale in the dissemination of information, marginal costs are thought to be far below average costs of dissemination. Thus, marginal cost pricing, if followed strictly, requires significant public funding of dissemination systems.

Economies of Scale

The average unit costs of producing information goods and services often decline with output. There are often high fixed costs associated with the collection and organization of databases, and dissemination technologies also exhibit strong economies of scale. Consider the following example. Three years ago several agencies indicated that the costs of editing the data for a CD-ROM was about $10,000, plus an additional $1,500 to print a "master" disk. Additional copies of the CD-ROM could then be pressed for less than $2. Assuming these stylized facts were true, Table I illustrates the changes in average costs of production at different levels of output:

              output         average cost            average cost       marginal cost
                             (w/editing)                             (w/o editing)     
     


                     1         $11,502                     $1,502         $2

                     2           5,752                        752          2

                   100             117                         17          2

                  1000              13.5                        3.5        2

Scale economies in online systems are also extremely important. Costs for services such as the Department of Commerce EBB, the National Library of Medicine MEDLARS system, NTIS FEDWORLD, the House of Representatives LEGIS, and the new GPO Access system are not very sensitive to usage. Some of these systems can double their capacity for usage for less than 2 percent of the total operating cost.

Incremental Cost

Incremental cost is a concept similar to marginal cost, however it does not focus on the cost of the last unit of production, but rather the cost of the production which exceeds a certain level. Incremental cost is a term that is also used in the study of multiproduct production processes.

Prices based upon the incremental cost of dissemination do not require additional public funding of dissemination systems. In thinking about incremental cost pricing, however, it is often important to be clear about what is assumed about the nature of the system, before the calculations are made regarding the costs of adding public access. For example, if one divides dissemination costs into two categories, P for the public and G for the government, and assumes that the government needs would be met with or without public access, the incremental costs of providing public access are as follows:

                                  ICp = C(P,G) - C(G)

       where:         C(P,G) is the cost of providing dissemination for both
                             the public P and the government G, and

                      C(G) is the cost of providing a dissemination service
                             that is only used by the government. 

One could also create a variety of product groups. For example, suppose that L represented the output that was used for the federal Depository Library Program. If one took dissemination of information to G and L as given, the incremental cost of public access through a sales program S would be as follows:

                             ICs = C(S,G,L) - C(G,L)


       where:         C(S,G,L) is the cost of providing dissemination for
                             the public sales program S, the government
                             G, and the Depository Library
                             Program L, and 


                      C(G,L) is the cost of providing dissemination services
                              only to the government and the Depository Library Program. 

Because of the huge economies of scale in dissemination of information it is often critical to determine how incremental costs are calculated. Since the first customer is by far the most expensive one to serve given the high fixed costs of most dissemination systems, the costs borne by the public will be much less if one takes the government's use of the data as a given.

Incremental cost is a very useful concept in looking at markets for government information, because it allows one to think about who bears different portions of costs of dissemination systems, and it also provides tools for thinking about multiple products and cross-subsidies.

In the examples above incremental cost was considered in the context of several different users of the same product. But we could easily consider a situation where many products are offered, and one defined the incremental cost of a single product. For example, many online systems provide access to several different databases. Examples of this include the National Library of Medicine MEDLARS system, NTIS's FEDWORLD , the Department of Commerce's Economic Bulletin Board (EBB) or the new GPO Access program. The incremental cost of access to a particular database would be the cost of providing the entire online system, less the cost of providing access to everything but that database.

Stand Alone Cost

Stand alone cost is the converse of incremental cost. Quite simply, the stand alone cost of producing a product or output is the cost of providing that product or output by itself, without the benefits of other production or products.

Economies of Scope in production

While many people are familiar with the concept of scale economies, there is less appreciation for the important related concept of scope economies. Scope economies (or diseconomies) are concepts which are used to study multiproduct production processes. The issue is the degree to which different products benefit from shared inputs. Suppose that a firm produces two products, A and B. If the joint cost of producing A and B is less than the stand alone cost of producing both A and B separately, then there exist economies of scope in production.

The issue of scope economies first was used in early debates over the AT&T monopoly, when economists representing that firm argued that significant economies of scope justified the vertical integration of AT&T. Today the regional bell operating companies (BOCs) still assert that greater vertical integration will allow consumers to benefit from economies of scope. While there has been considerable skepticism about the existence of scope economies in the telecommunications field, there can be little doubt but that economies of scope are important, as evidenced by the fact that very few firms are build around a single product. Economies of scope can involve economies in management, finance, marketing and R&D as well as production processes.

Economies of scope are clearly important in information dissemination. Many of the fixed costs of online systems, such as staffing and software development, are relatively insensitive to the number of databases that are offered. For example, the Department of Commerce EBB uses 1.5 staff to provide an online service that is used by more than 15 different federal agencies, and the government could undoubtedly double the number of databases and files on the EBB without any increases in staff or hardware.

For commercial firms, there are also very important scope economies in marketing, which is extremely important, given the fact that marketing is typically considered the largest expense (by far) for commercial online systems.

Economies of Scope in consumption

One often observes a wide range of prices for online access to the same database. For example, the Department of Education ERIC database is available for $36 per hour in connect charges plus $.29 per record displayed from Dialog, and $20 per hour and $.05 per record from OCLC. But these differences are relatively minor compared to some disparities. The Bureau of the Census CENDATA database is available at $60 per hour in connect charges plus $.75 per record from Dialog, while the same database from Compuserve is priced at a flat connect rate of $4.80 per hour.

Table II

Selected Private Online Prices for ERIC and CENDATA
  

                                           hourly             record
                                     connect charge           display

Department of Education/ERIC


              Dialog                            $36                 $.29
              OCLC                               20                  .05

Bureau of the Census/CENDATA


              Dialog                             60                  .75
              Compuserve                          4.80              0

These pricing differences persist despite the fact that there are only small differences between the quality (the power and ease of use) of the front end user interfaces. There are several reasons that one firm can charge much more than another for the same information, but the most important ones concern the burdens incurred by a person using any information service, on such items as finding a particular database, learning a user interface and establishing lines of credit. That is, there are economies of scope in the consumption of online information services. It is much easier, and therefore "cheaper" in terms of time, training and bureaucratic hassle, to get several items from the same source, rather than from several sources. Full text search is also more efficient when the searching engine can scan the widest number of relevant databases.

The ability to provide access to a broad scope of databases has historically been a significant drawing card for a commercial online system. The difficulty in finding data, learning new user interfaces, and establishing credit, has been a significant barrier for users who might otherwise "shop around" for data bargains. The fact that consumers prefer services with larger product lines makes it difficult for new entrants in database markets, and increases the market power of the few companies which dominate the online vendor market. Data Resources, Inc. (DRI) has the largest product line of economic statistics, and it is by far the most expensive. Westlaw and LEXIS are the only two firms that offer a complete array of legal databases, and they are very expensive. Dialog is the most complete vendor for bibliographic information, and it is considerably more expensive than several of its smaller rivals.

The fact that the services with the largest product lines are the most expensive is somewhat of a paradox, given the existence of economies of scope for the vendors in the provisions of the service. The current market structure is a problem. The barriers for new entry have led to high vendor prices and have discouraged innovation in new types of products.

Economies of Scope and the Internet

The rise of the Internet as a source of online information is providing new opportunities to change radically the way that citizens use databases. There are three major obstacles for consumers in obtaining online access to information. First, it is necessary to locate the information. Second, one has to learn how to use the online service. Finally, if there is a fee, it is necessary to establish a line of credit to use the service. The Internet has made the first and second items much easier, and there is considerable work underway to solve the third problem.

First, the Internet makes it much easier for users to find where information is located. Not only does the Internet feature a number of readily available indexing and locator services, such as Archie, Veronica, and the World Wide Web, but it also makes it much easier for consumers to communicate with each other and share information about sources of information. This in turn dramatically changes the concept and economies of marketing, since consumers have a way of finding vendors, and consumers who are price sensitive can and do share information, including information on where government information may be available for free.

Second, the Internet has a number of sophisticated tools, such as gopher, WAIS, mosaic and lynx, which provide standardized front ends for searching data. These software products are powerful, easy to use, and widely available. The federal government has funded the development of a number of these software products, which are typically available for free or a nominal price from developers. As a result of the widespread deployment of standardized user interfaces, it is much easier for data users to search and retrieve data from different "publishers" of data. This dramatically lowers the entry barriers for new firms, and increases the competition among vendors who provide access to the same database.

On the third issue, establishing a line of credit, the federal government is funding a number of research projects aimed at making it possible to conduct financial transactions over the Internet.[3] The goal of these projects is to create a form of digital cash, somewhat akin to a fare card for a subway system, which will be used to pay for transactions in real time. Persons working on this problem are interested in developing methods of payment that make it possible to "pay" as little as $.20 for a transaction, making it feasible and possibly profitable to market low cost services with broad public appeal.

The incumbent database vendors such as DRI, LEXIS, WESTLAW or Dialog are alarmed by many of the developments on the Internet, which they see as a threat to their markets.

There is an addition issue which deserves some consideration. Many of the indexing and searching services on the Internet exist in part because it is possible to search various decentralized Internet resources without paying for telecommunications charges. Under the High Performance Computing Act of 1991 NSF was instructed to fund research aimed at developing methods of charging for network usage. Over the past several years a number of firms and researchers have been investigating ways that users can be charged for their use of Internet services. Any movement toward usage based pricing on the Internet may lead to profound changes in the current infrastructure of Internet discussion groups and indexing and searching services.

Subsidy Free Prices

The private sector and the government both engage in widespread price discrimination. WESTLAW, LEXIS, Dialog, Compustast and other vendors provide deep discounts to education institutions. The federal government charges commercial users higher fees under FOIA than it charges non-commercial users, and the GPO provides 1,400 federal Depository Libraries free access to its publications. It is sometimes asserted that price discrimination in general and non-profit discounts in particular result in one group subsidizing another. In fact, the issue of price subsidies is more complex. In the economics literature subsidy free prices are defined as a set of prices which require each group of users to pay no more than the stand alone price of the service, and no less than the incremental cost of service.

To illustrate this concept, suppose that there are two groups of customers, commercial users and non-commercial users. Suppose further than the technology for disseminating the information consists of fixed costs of $5,000, plus variable costs of $500 for each group served. The stand alone cost of serving either group is $5,500, the cost of serving both groups is $6,000, and the incremental cost for either group is $500.

If each group was charged an equal amount for access, the charge would be $3,000. Suppose, however, that the non-commercial users are only willing to pay $2,000 for access, while the commercial users are willing to pay much more. If each group pays an equal share, the non-commercial users would not participate, and the cost to the commercial users would be the stand alone cost of $5,500.

Suppose instead that the non-commercial users were allowed access for $1,500, somewhat less than their maximum willingness to pay. The total cost of serving both groups would be $6,000, and the cost for the commercial users would be $4,500, or $1,000 less than the case where costs are shared equally.

Indeed, any price for the non-commercial users that is greater than their incremental cost of $500 or less than their maximum willingness to pay of $2,000 will benefit the commercial users more than a system of sharing costs equally. On the other hand, if the non- commercial users pay less than $500, the commercial users would end up paying more than their stand alone cost of $5,500, and it would be accurate at that point to say that the commercial users were subsidizing the non-commercial users.

The existence of a subsidy is determined by the net contribution toward joint costs, rather than the differential in prices. As long as a user pays more than the incremental cost, a net contribution is being made toward to the joint costs, which will benefit the other users (as long as prices are limited to costs).

Peak Load, Congestion, and Time of Day Pricing

Online information systems not only benefit from large economies of scale, but the costs of the system depend upon how the systems manage peak and off peak use. A system that supports 50 simultaneous users could double the number of simultaneous users that it supports, or it could also provide incentives for users to shift access to off peak hours. For example, the Department of Commerce EBB discourages peak load use by charging higher fees in the mornings, and less for evening and weekend use ( $9 per hour for weekday use from 8 am to noon, and only $3 per hour weekdays after 6 pm and all day on weekends). A number of commercial vendors, such as Dialog and LEXIS, also limit certain educational discounts to hours which do not conflict with peak usage of the system.

There two important reasons for charging different prices for peak and off peak usage. As noted, by using prices to shift users to off peak hours, it is possible to lower the cost of providing access in peak periods. Differential pricing by time of day also allows a vendor or government agency to employ a type of price discrimination that does not rely upon intrusive tests of user means or goals. For example, the EBB found that business users, who have the highest willingness to pay, were particularly interested in accessing the EBB in the mornings, while many non-commercial users were happy to access the service at lower rates in off peak hours. The peak load pricing policy at the EBB gives the commercial users better access (less congestion) during the hours which are important to them, and it provides a much lower cost alternative for those who can wait. Since both groups are paying more than their incremental cost (most of the EBB costs are fixed), and less than the stand alone cost of the service (the EBB limits fees to costs), the prices are subsidy free. Moreover, if a substantial number of off peak users would have abandoned the service under a single price rule, the differential pricing may actually lower prices for both groups, since both peak and off peak users make net contributions to the joint costs.

Price Discrimination and Metering

Most commercial data vendors engage in extensive price discrimination. There are many reasons why price discrimination exists. A firm with market power is better off if it can charge different users different prices, based upon their "willingness to pay." Consider a market with two users, one which is willing to pay $1 for data and another that is willing to pay $3 for the data. If the price is set at $1, two units are sold for a total sales of $2. If the price is set at $3 one unit is sold for total sales of $3. But if the vendor charges the first user $1 and the second user $3, it will have total sales of $4. So long as the incremental cost of supplying either user is less than $1, the vendor will maximize profits using price discrimination.

The term "reservation price" is the term used to describe the highest price that a consumer is willing to pay. In most cases firms do not have perfect information about each consumer's reservation price, although partial information may allow the firm to place consumers into different groups according to perceived differences in the willingness to pay.

The ability of a firm to conduct price discrimination depends on a number of technological, legal, logistical and practical considerations. If individuals have low willingness to pay and businesses have high willingness to pay, a firm may try to charge a different price for each group, but this is not always possible. For example, it would be difficult to charge individuals $10 for a book, and charge a firm $20 for the same product, if the firm could buy the book from the same retail outlet. Online services, however, offer a number of opportunities to discriminate among users, typically using features of the service or the billing mechanism to distinguish between types of users. For example, Dialog offers an off-peak service that is marketed to individuals, at significant discounts over the normal commercial rates. But since this service is only available after normal working hours, and billed to an individual Visa or Mastercard, the service is less likely to be used by corporate libraries and other large users of Dialog's more expensive day time services.

Online services provide opportunities to monitor and meter the use of services. Vendors may uses metering to charge intensive users more than occasional users. This represents an important change from a book or a diskette, where the vendors typically cannot monitor the number of readers or intensity of use. This change in the way that information is sold has created new ethical dilemmas for libraries, since each user of a metered online service imposes new costs. Libraries, which are constrained by funding, are faced with difficult choices, such as rationing access to information, or charging fees to patrons for searches.

The metering systems used by online vendors can be simple, like a hourly rate for using the service, or complex, such as the algorithms used by the National Library of Medicine, LEXIS, WESTLAW or Data Resources, Inc. (DRI). The sophisticated metering systems not only allow the vendor to measure the intensity of use, as measured by time or bytes of data, but also the specific information that has been retreived.

For example, to obtain an SEC filing from LEXIS, using the standard contract, one must pay a flat monthly fee of $125, a $39 per hour connect charge, a fee of $15 to $50 per document, depending upon the type of filing, plus 1.8 cents a line to download to view the data. LEXIS charges less or more for other libraries on LEXIS, depending upon its perception of willingness to pay. DRI requires a steep minimum payment and connect charges, plus fees for individual data points (individual numbers), as well as other charges. The DRI prices for data points vary greatly from number to number. For example, national GNP figures are much cheaper than the MSA level employment numbers, because DRI assumes that there are fewer sources and thus less competition for the regional data than for the national GNP figures.

Monopsony Power, Government Procurement

Monopsony power is the term used to describe the market power exercised by a large buyer or a group of buyers acting together.[4] Large firms or organizations representing groups of users often bargain for discounts and lower rates.

One reason that large buyers obtain discounts is that there may be economies of scale in providing the service. For example, marketing costs, which are generally considered the most significant factor in the overall cost structure for database vendors, may be lower for large buyers. Large purchase orders may also allow the vendor to benefit from greater economies of scale in the production of the product or service.

Larger buyers also exercise greater bargaining power, particularly when buyers can play vendors off against each other. The bargaining power of the large buyer is particularly important in the market for database services, since prices are typically far above marginal costs. Large buyers can also make more credible threats to develop their own products or services in-house, since the "stand alone" cost of the service for large buyers is much lower, on a per-unit basis, than is the case for an individual.

In an extreme case, where there is only one buyer, the monopsonist can choose to procure the service from one or more vendors, using competitive bids or negotiated contracts.

The largest consumer of federal government information is the federal government. The federal government, through its procurement, could exercise significant monopsony power, but for the most part this hasn't occurred. Under the Library of Congress FEDLINK program, for example, the federal government negotiated a flat hourly rate of $125 per hour for LEXIS and WESTLAW searches, which is about one half the cost to the public under the standard commercial contracts. But the federal government could have done much better if it had used the procurement process to make the market for computer assisted legal research (CALR ) more competitive.

The federal government could, for example, insist that winning vendors use non- proprietary citation systems for federal caselaw, eliminating an important barrier to competition which exists due to West Publishing's claim of a copyright on the page numbers of published federal caselaw. The federal government could also require, in the procurement, that the winning bidder provide the government with a free and clear title of the database of caselaw, so that in subsequent procurement, new firms would find it easier to enter the CALR market. These approaches have been recommended by Tax Analysts and other smaller firms. Some experts predict that better procurement policies would lower the hourly cost of online legal research to about $30 per hour, or about one quarter the FEDLINK cost. Moreover, the public, which is now paying about $240 per hour for CALR, would also benefit from the greater competition.

The federal government could also uses its substantial power in procurement to improve the level of competition in many other database markets. For example, the federal government produces and uses literally hundreds of databases of research abstracts. These include well known products such as Medline, AGRICOLA and the NTIS Bibliographic database, as well as many others that are less known and have smaller numbers of users.

In some cases the government produces its own CD-ROM products, and in other cases it buys products from private vendors. The private vendor costs are often very high. Silver Platter, a commercial vendor, charges $825 per year, plus a one time charge of $950 for backfiles, for a single workstation subscription to the National Agricultural Library's (NAL's) AGRICOLA database on CD-ROM. The major customers for this product are Department of Agriculture employees and publicly funded universities.[5]

The cost of mastering a CD-ROM is about $1,500, which is less than the cost of two Sliver Platter subscriptions. Since the government buys a large number of subscriptions, the government's "stand alone" costs of production are far less than what it pays Sliver Platter. It would be difficult to imagine a private firm creating a database, turning the data over to commercial vendors, and then buying the data back from the vendor at retail or close to retail prices, but that is what the federal government often does, at great cost to the taxpayers.

The federal government could take a number of steps to make procurement more competitive. It would be very helpful if the government would provide greater standardization of the records formats in which federal research abstracts are stored and disseminated. The government should also develop (in-house or through contractors) public domain software for CD-ROM and online searching that will work across all federal research abstract databases, so that each agency doesn't confront the same problem from scratch. This would make it far cheaper for the government or private vendors to create new CD-ROM or online products, lowering both the government's costs, and the costs of market entry, thereby enhancing actual and potential market competition. If the software is placed in the public domain, it will also be available for databases of privately copyrighted materials, making that market more competitive as well.

The idea of using the government to develop non-proprietary record formats and public domain software is not an untested proposition, as evidenced by the huge success of public domain software searching mechanisms deployed on the Internet, including Mosaic, which is an extremely powerful and sophisticated multimedia software program for online publishing. The federal government and data users would benefit richly from a Mosaic type software program designed for searching research abstracts or numeric data accessible on CD- ROMs or the Internet.

Flat Rate Pricing

Every subscriber to an online information service has their own "demand curve," or schedule of willingness to pay for online time or searching. If the pricing contract sets a charge for minutes of use, searches, bytes or whatever, the level of usage will generally be lower than if the user paid a fee which did not depend upon the intensity of use. Users will respond to metered usage by economizing on their searches. If the usage based fees exceed the marginal costs of usage (which are typically zero or trivial for online systems), the result will be socially inefficient, since the users value the service greater than the marginal cost of providing access.

Flat rate contracts allow subscribers to pay a lump sum for unlimited searching privileges.[6] Many institutions prefer flat rate pricing contracts because they offer predictability in budgets, and reduce internal incentives to charge back users, which can be a difficult and undesired task.

If priced correctly, flat rate contracts are more socially efficient than metered alternatives, since usage isn't constrained by fees which exceed marginal costs. However, the efficiency benefits of flat rate contracts depend upon the level of the flat rate. If only flat rate contracts are available, and the flat rate is too high, some data users will be priced out of the market altogether and consume nothing, a result which is generally socially inefficient, and may lead to significant losses in the potential customer base.

The Department of Commerce's EBB provides users with the option to choose between contracts which are based upon flat rates or hourly charges. Intensive users are generally better off under the flat rate contracts, and end up paying lower rates, on an hourly basis, than the persons who buy the service under through the hourly contracts.

Economic efficiency and Ramsey Pricing

Ramsey pricing is a term named after Frank Ramsey, the mathematician and economic theorist.[7] Economist have long been interested in optimal rules for economic efficiency in taxation and in the pricing of public utilities. The problem which is addressed is the following:

in a world with different products and different consumers, what are the most efficient set of prices that will allow society to finance fixed costs (of production) which are not covered by marginal cost pricing?

This is a particularly relevant issue for online information system, since marginal cost pricing will virtually never cover total production costs (which are mostly fixed).

Ramsey did not consider the wide range of pricing strategies that we observe in online markets. Rather, he focused on the issue of how "unit" prices for different goods (or different consumers) should exceed marginal costs, in order to finance a given fixed cost. The solution to the Ramsey pricing problem (in the simplest case where cross-elasticities of demand are zero), are a set of prices where the mark-up over marginal cost is highest when demand is inelastic, and lowest when demand is elastic. This is also referred to as "inverse elasticity" pricing. One consequence of Ramsey pricing is that consumers with the highest willingness- to-pay contribute the most toward the fixed costs.

The proof that this is an efficient result requires sophisticated mathematics, but the intuition is simple. It is efficient to produce goods or services when consumers value the good or service more than its marginal cost. Prices above marginal costs are socially inefficient, but the efficiency costs are much less when the level of consumption in not sensitive to changes in prices. In a practical sense, the consumer who responds to prices the least is asked to pay the most.

There are aspects of Ramsey pricing which deserve further explanation. Ramsey and other economists who advocate this rule define efficiency in such a way that transfers of wealth between different consumers or between consumers and producers are irrelevant, since the objective is to maximizes society's net benefits, without regard to who wins and loses in a particular pricing regime. This is not, therefore, a good rule to follow blindly if one cares about income distribution, or fairness or equity of access to information in general. Indeed, it is often pointed out that under Ramsey pricing, the government should place extremely high taxes on insulin and other medicines, since demand for therapies which are essential for one's health are highly inelastic. Of course, few practical persons believe that high taxes on medicines would be a just or wise public policy, no matter how "efficient" the outcome using neoclassical cost benefit criteria.

Ramsey pricing does agree, however, with public policies that would charge business users higher fees for access to government databases than non-commercial users, since the business users would, presumably, have a less elastic demands.

GOVERNMENT PRICING RULES

There are a plethora of rules and policies for the pricing of federal government information. It is helpful to begin with the rules which were long used to price information in paper formats, the Title 44 printing statutes and the Freedom of Information Act (FOIA). Then we examine the authorizing statue for the National Technical Information Service (NTIS) and a number of special agency rules and provisions.

The Government Printing Office

The traditional rules for pricing federal information are found in Title 44, the section of the U.S. Code which provides the rules for agency printing and the Government Printing Office (GPO). Under this system, the GPO prints documents on behalf of federal agencies, selects some publications for distribution to the federal Depository Library Program, and also selects documents for a sales program which serves the general public. In 1993, Congress amended Title 44 when it passed the GPO Enhanced Electronic Access Act (GPO Access), which requires GPO to provide online access to some federal publications,[8] and modified the GPO pricing rules as they apply to online services.

The Depository Library Program

The 1,400 member Depository Library Program (DLP) receives a large number of GPO information products and services for free. While DLP members have historically received delivery of information "products" for free, it appears as though DLP members may be responsible for their own telecommunications charges for GPO's online services. The cost of the DLP is paid for through public appropriations (the fy 1993 federal budget for the DLP was $25.1 million). The DLP, which was created at the beginning of the 19th century, is designed to provide a decentralized system of universal access to federal government information.[9] Most of the members of the DLP are non-federal institutions, that enter into contracts which require the member library to provide free public access to its federal government document collections, in return for the library's participation in the free distribution of the information.

While all federal agencies are required by law to provide the DLP with copies of their publications, there is a substantial "fugitive document" problem, as many agencies ignore the legal requirements for DLP distribution. This has been a particularly important problem for electronic products and services. The 1993 revision of OMB's Circular A-130 has compounded the fugitive document problem by telling agencies that while they are encouraged to provide electronic products and services to the DLP, OMB considers this an optional, rather than a mandatory, requirement.

The GPO Sales Program

Information products which are included in the GPO sales program (not including online services), are priced at 150 percent of the so called "rider rate," of the publication. That is, the agency pays for the initial press run, and GPO "rides" the order for the DLP and the sales program. There has been some controversy in recent years over the method used to calculate the 150 percent of cost (cost plus 50 percent). GPO now claims that it receives a 50 percent mark-up on all costs associated with the printing, handling and distribution of the publications.[10] The GPO "rider" rate is conceptually the same as the incremental cost of public dissemination, with the government's output taken as a given.[11]

GPO Access Sales Program

Under the GPO Enhanced Electronic Access Act (GPO Access), GPO can charge no more than the "incremental cost of dissemination," for its online sales program. One important issue that is not resolved is whether GPO will adopt a pricing strategy that will make it easier for individuals and researchers with episodic research interests to obtain access to databases which are sold for fees.[12] There was broad initial interest in having GPO provide one- stop-shopping for federal online information, in much the way that GPO offers "one-stop- shopping" for information products printed on paper, with users able to pay for a variety of databases through a single invoice and a single line of credit.

The GPO Access program officially begins operation on June 8, 1994. The initial product line is the online version of the Congressional Record, the Federal Register, a Federal Locator, an electronic storage facility, enrolled bills, and the GPO Federal Bulletin Board.

The GPO Federal Bulletin Board has been running for some time. The service consists of more than 5,000 files from more than eighteen government agencies, including the executive branch, Congress and the federal Courts. Users are currently charged by the size of the file which is downloaded, with no connect charge. The data charges are as follows:

Table III

GPO Federal Bulletin Board Prices

                   file size (kb)                        fee

                             < 50                      $2

                     50 > , < 100                       3

                    100 > , < 150                       4

                    150 > , < 200                       5

                    200 > , < 250                       6

                    250 > , < 300                       7

                    300 > , < 400                       8



                   $1 per additional 100 kb ($10 per megabyte) thereafter

GPO's prices for the Federal Bulletin Board (FBB) are based upon the uncompressed size of the files, even though many are stored in compressed formats. The FBB prices are high relative to the prices charged by the Department of Commerce EBB, but less than the prices charged by the best known commercial data vendors. Last year GPO was asked by the Taxpayer Assets Project to provide a system with lower prices for off-peak usage, but declined to do so.

On June 8, 1994 GPO announced its prices for the online versions of the Federal Register and the Congressional Record. Both services are provided over the Internet using a WAIS server, and are only sold through a subscription.[13] The cost of the subscription for a single work station is $30 for a single month, $200 for six months, or $375 for one year. GPO increases the prices for "subscriptions" for networks. The schedule of prices is as follows:

  Number of Work Stations      Annual Subscription                      Average
                                                                        Cost*

       1                                   $375                          $375
       2 to 10                             $750                          $75
       11 to 50                            $1,875                        $38
       51 to 100                           $2,815                        $28
       101 to 150                          $3,755                        $25

       151 to 500                          $3,755 plus 
                                           $188 per 10 work stations     $21

       501 to 1000                         $10,335 per year plus
                                           $94 per 10 work stations      $15

       1001 or more                        inquire

*Average cost calculations are based upon the high end of the range of allowable workstations.

GPO's pricing for the Federal Register and the Congressional Record are controversial. While the annual subscription cost for unlimited searching is less than the cost of using existing commercial services (far less in the case of the Congressional Record), assuming one searches the database extensively, it is still far too high for private citizens or most small businesses and non-profit organizations.

The pricing schedule for the Register and Record are interesting for a number of reasons. First, GPO offers only the flat rate contracts, rather than a choice of flat rate or hourly rates. As a result, the prices are cheap for intensive users (a little more than $1 per day, for services that cost more than $100 per hour on Dialog), but extremely expensive for occasional users. Under this pricing regime GPO will only sell the subscriptions to intensive users, and virtually no individuals will have access to either product, except through institutions which are subscribers.

Secondly, there are huge differences between the costs of adding users in a large organization, and the prices for an individual subscription. The average cost either product for a 10 user contract is just $75, or one fifth the cost to an individual. The largest user can add users for only $9.40 per year for unlimited use, which is 1/40th the price faced by individuals.

There are also huge "border" problems in the GPO pricing schedule. The marginal cost of going from 1 to 2 work stations is $375. From 2 to 10 workstations, the cost is zero. From 10 to 11 stations, the marginal cost is $1,125, while from 11 to 50 the cost is zero again.

The flat rate pricing model for unlimited searching is an increasingly popular way to market online services to libraries and other large users,[14]but at the prices announced by GPO it is an entirely inappropriate way to price the service to individuals and episodic researchers.

GPO's failure to provide a method of pricing that allows episodic users to access the database represents an alarming misunderstanding of the purposes for which the GPO Access program was enacted. Congress clearly intended the GPO Access program to enhance public access to important collections of public documents which are vital to a democracy.[15] The Federal Register and the Congressional Record were singled out precisely because they were considered documents that are fundamental to participation in national debates over public policy issues. GPO personnel deliberately designed a pricing system which will only serve a handful of large organizations, raising questions about their understanding of the agency's mission.

According to GPO personnel, the business plan for the first year of the GPO Access program calls for the sale of less than 2,000 subscriptions. At the minimum price of $375 for a single work station contract, each 1,000 subscriptions generates $375,000 in revenue. Since the GPO Access operating budget is under $1 million, the entire program could be funded through less than 1,500 subscriptions to each product. To put this into perspective, there are more than 100 thousand libraries, 15 thousand corporations regulated by the SEC, 19 thousand EPA employees, 2 million members of the armed forces and 276 million people in the United States, so a program that is designed to make these two critical documents available to a few thousand organizations is only scratching the surface of the potential readership.

On July 5, 1994 the Taxpayer Assets Project, OMB Watch, the Congressional Accountability Project and People for the American Way met with the Public Printer to discuss these concerns, and ask for a number of remedies, including such items as contracts for searching by the hour, lower priced or free subscriptions for k-12 schools and smaller public libraries, and discounts or free use in off-peak hours. Michael DiMario, the GPO Public Printer, promised to consider the suggestions.

Free After Six

Among the suggestions for changes in the GPO pricing scheme for the Register and the Record is a proposal that GPO provide free access to its entire produce line from 6pm to 8am on weekdays and all day on weekends. This proposal is referred to as "free after six." Proponents of the "free after six" proposal say that businesses and non-profit organizations that are buying subscriptions so they can search the database during normal business hours (8am to 6pm EST) will continue to do so, even if the service is free in off- peak hours.

The supporters of the "free after six" proposal also anticipate that free access in off peak hours would help to introduce new users to the system, some of whom will become familiar with and dependent upon the service, and lobby their employers to purchase subscriptions for day time access. A number of private sector vendors now encourage free use of systems as marketing mechanisms. LEXIS and WESTLAW, for example, are often free or discounted for law students, who then come to rely upon the services (get hooked), and become paying customers through employers upon graduation from law school. America Online, Prodigy, Delphi, many internet providers, premium cable channels, newspapers, newsletters, magazines and many other commercial information services also offer free trial use in order to gain paying subscribers.

The costs for GPO in providing the free after six service are trivial, since most of the costs of the program are fixed, and GPO is already providing 24 hour 7 day a week access to the service for the existing subscribers.

If GPO adopts the "free after six" proposal, it will have found a clever and important way to accommodate both the practical requirements that the agency seek user fees to fund the program's operation, and the public interest in broadening access to government information for individual citizens, most of whom will be deterred by any pricing system. The use of the time of day is a non-intrusive method for price discrimination that achieves an important public purpose without reliance upon means tests or agency judgements about the benefits of the search to the public interest. Since the benefits of "free after six" would be available to everyone, it would satisfy most criteria for fairness. This mechanism is only available for online information systems, since dissemination of information in paper, microfiche, diskette, or CD-ROMs are not generally metered according to the time of usage.

GPO's Start-up Problem

GPO has struggled with its pricing policies for online services. When the Federal Bulletin Board was first established, access was limited to federal depository libraries. When the system was opened to the public, the agency tried to adopt prices to fund the cost of the program. But since the initial customer base was quite small, the relatively high user fees did not generate much income. In contrast, the Department of Commerce EBB was first created as a free program and did not introduce user fees until it had attracted a large user base, allowing it to set fees which were as low as $3 per hour via telephone or $6 per hour using the Internet.[16]

When the GPO Access legislation was passed, Congress did not provide any appropriations for start-up, requiring GPO to fund the program from savings elsewhere, or through user fees. Congress is currently considering a request for $1.5 million for additional start-up funds for next year's budget. The problem for GPO concerns the difficulties of funding the start-up of the program and developing economies of scale and scope.

There are several important chicken and egg problems. For GPO to lower the unit costs, it must benefit from scale and scope economies. To attract users, it is helpful to have a large product line, and low unit prices. For GPO to build a product line, it must work with federal agencies, whose participation in the GPO Access program is voluntary. Agencies, on the other hand, are more likely to cooperate with the GPO Access program if it has a large user base. All of these problems suggest that GPO needs funding for its start-up, allowing it to initially price below cost (which is allowed under the statue) and build a larger user base, before the program is expected to generate significant funding from user fees. As noted above, GPO should avoid pricing policies which require the purchase of large blocks of searching time, or it will discourage access by persons who only use databases occasionally, and decrease the GPO Access user base.

Reinventing GPO's customer relations

GPO has repeatedly been asked to engage the public in debates over the product line and prices for GPO Access services, but the only group which it has made a serious effort to listen to are government documents librarians, who generally have a more limited view of the pricing issues than do the general public.[17] Under title 44, federal Depository Libraries have online access to the GPO Access program for free, and the pricing structure for the Federal Register and the Congressional Record generally meets the needs of non-depository libraries which are frequent users of these documents. GPO has rejected requests that it set up an Internet discussion list to allow the general public to discuss the GPO Access product line and pricing issues. If GPO had done so it would undoubtedly have decided to adopt peak load discounts for the GPO FBB, and it would have likely provided a better method of providing episodic users with access to products such as the Federal Register and the Congressional Record. Moreover, some of GPO's new products make little sense in terms of market demand. GPO has announced that it will provide the "enrolled bills" of the 103rd Congress for $60 per year. Not only is the $60 price prohibitive for ordinary citizens, who may only want access to a single bill in the course of a year, but GPO also picked the least important group of bills to publish. The "enrolled bills" have already passed Congress, and citizens can no longer have any opportunity to influence the legislation other than to ask the President to veto or sign the legislation. The public wants copies of all versions of the bills which are pending before Congress.[18]

GPO's problems are illustrative of a general failure of federal agencies to maintain contact with users to better appreciate the way data are used and the consequences of pricing policies. A model for engaging the public in regular constructive dialogues about agency product lines and prices is found in HR 629, the Improvement of Information Access Act.

The Freedom of Information Act

The federal Freedom of Information Act (FOIA), section 552 of Title 5 of the U.S. Code, spells out the general rights of the public to obtain agency documents, including documents which are not considered "publications," such as memos, notes or other documents which are not otherwise publicly available from the agency. Agencies must adopt schedules of fees applicable to the processing of requests under FOIA, as well as guidelines and procedures for waiving or reducing fees.[19]

Commercial Use under FOIA

Under FOIA, the fees can be no more than "reasonable standard charges for document search, duplication, and review, when records are requested for commercial use."[20]

Non-commercial use under FOIA

When a FOIA request is for non-commercial use, by "an educational or non- commercial scientific institution, whose purpose is scholarly or scientific research; or a representative of the news media," the fees "shall be limited to reasonable standard charges for document duplication."[21] If a non-commercial request is made by a requester who does not represent a scholarly educational, scientific or news media organization, the fees shall be limited to "standard charges for document search and duplication."[22]

Fee Waivers

Moreover, a request by anyone, which is not principally for commercial purposes, and which would result in a disclosure that is "in the public interest because it is likely to contribute significantly to public understanding of the operations or activities of the government," must receive a waiver or reduction of fees.[23]

The FOIA rules regarding the fees for access to information are designed to limit agency discretion regarding the pricing of information, and to promote broad public access to materials which are important for democratic debate and scholarly or scientific research. There are a number of important limitations on the public's rights under FOIA. The most important limitation with regard to pricing concerns the dichotomy between published and unpublished information. Some courts have found that agencies do not have to use FOIA pricing rules when the information is available from the government as a publication for sale.[24]

NTIS

The National Technical Information Service (NTIS) has broad responsibilities for the collection and dissemination of federal information. As a result of efforts by the Reagan Administration to "sell" NTIS to the private sector, Congress enacted "compromise" legislation which requires much of the agency's activities to be funded through user fees. NTIS must use fees from the dissemination of government information to fund a wide range of activities:

Operating costs for the National Technical Information Service associated with the acquisition, processing, storage with bibliographic control, and archiving of information and documents shall be recovered primarily through the collection of fees.25]

The NTIS budget for fy 1993 was $46.9 million, including $39.1 million in user fees. NTIS has adopted very aggressive pricing policies. Many NTIS activities do not generate sufficient income to cover costs, so the agency seeks to make up the difference by charging very high prices on certain information products. While the agency as a whole breaks even, individual information products are often priced far above costs. For example, a subscription to the Environmental Protection Agency's Enforcement Document Retrieval System (EDRS) Master file is $1,440 for four quarters, or $360 for a single file. The EPA Facility Index System (FINDS) file is priced at $2,360. Health related files are also very high priced. For example, the Food and Drug Administration's Vaccine Adverse Event Reporting (VAERS) database is distributed on two floppy disks, and costs $1,080.

There seems to be little relationship between the cost of disseminating the data and prices that are charged by NTIS. Not all NTIS prices are high. In general, however, NTIS prices are considerably higher than are GPO's prices, even for the same items. Table V is a comparison of prices for 19 databases which are sold by both GPO and NTIS. Of the 19, which were selected on a non-discriminatory basis from GPO and NTIS catalogs, the NTIS prices are much higher in every case. The closest prices, in percentage terms, is for a database that is priced at $34 by GPO and $90 by NTIS.

     DATABASE                  			NTISPRICE	GPOPRICE
Annual Energy Review 1991     			$90		$17
Commercial Building Energy Consumption 		$140		$20
Survey, 1989
EPA Toxic Chemical Release Inventory 		$1,800		$174
1989
EPA Toxic Chemical Release Inventory by 	$55		$15
State
Field Size Distribution for U.S. Oil and 	$55		$15
Gas Provinces, 1989
Historical Monthly Energy Review Data File	$140		$22
International Coal Statistics Database		$90		$17
Monthly Electric Utility Sales, EIA #826	$90		$15
Monthly Energy Review, Current Month		$55		$15
Monthly Power Plant Report, EIA #759  		$195		$15
Natural Gas Annual, 1988, Volume II  		$90		$17
Natural Gas Annual, 1989, Volume I  *		$90		$34
Nonresidential Buildings Energy Consumption	$140		$20
Performance Profiles of Major Energy Producers 	$90		$17
Residential Energy Consumption Survey 1990,	$90		$15
Consumption and Expenditure Tables
Residential Transportation Energy Consumption	$55		$15
Survey, 1988
State Energy Data Systems, Per Census Region	$50		$15
State Energy Price and Expenditure Data System,	$50		$15
Per Census Region
U.S. Crude Oil, Natural Gas, and Natural Gas	$55		$15
Liquid Reserves

* This document is sold in two parts by the Government Printing Office. The total of the two GPO documents is $34.

* This document is sold in two parts by the Government Printing Office. The total of the two GPO documents is $34.

NTIS often places restrictions on how information is redisseminated or used. For example, the Environmental Monitoring Methods Index (EMMI) is produced by a joint venture between NTIS and Viar & Company, in cooperation with EPA. The EMMI is a relational database system for personal computers that lists "2,600 EPA regulated chemical substances, methods for their analysis, and regulatory and office based lists on which they appear." The price of the system, which only requires 8 megabytes of disk space, is $385 for a single user, $995 for an office LAN that supports 2 to 5 users, and $1,995 for an office LAN with an unlimited number of users.

NTIS Leased Databases

A number of NTIS databases are not available for sale at all, but rather are "leased" to vendors who resell the data to the public. According to NTIS officials, the agency currently has 85 lease agreements with various vendors.[26] There is a wide variance in the terms of the lease agreements across different databases. Lease terms for the NTIS Bibliographic Database, the AGRICOLA database and the Registry of Toxic Effects of Chemical Substances (RTECS) database are given below:

              1.  Annual Fee for 1994 current files         $9,500

              2.  Backfile fees:1993                         8,500
                                1992                         7,500

                                1991                         6,500

                                1990                         5,000

                                1989                         3,500

                         1980 - 1988                         1,500

                         1964 - 1979                           750


  (customers who purchase entire backfile receive 20 percent backfile discount)


              3.  Connect Time                        $30 per hour

              4.  Online Record Display or Print


                         full record                          $.30
                      partial record                           .25
                      minimal record                           .03

              5.  Off-Line Print                              $.30

              6.  Search term              $.03 per term per usage
              (any word, word fragment, number of other item that represents the interests of
              the user, and which constitutes one of the basic semantic units of a search
              statement.)


              Plus additional fees for use of selective dissemination of information (SDI)
              profiles, off-line retrospective search and republication of records.

              1.  Annual Lease, 1994 data  $2,750

              2.  Backfile data, 1970 - 1993         $200 per year

                                (less 35% discount for all backfiles)

              3.  Connect time                         $6 per hour

              4.  Online Display or Print                               
                         full record                          $.10
                      partial record                           .05
                      minimal record                           .01

              5.  Off-line print                              $.10

              6.  Search Terms                                  $0

Plus additional fees for use of selective dissemination of information (SDI) profiles, off-line retrospective search and republication of records.

              1.  Annual Lease fee                          $2,500

              2.  Connect time                        $22 per hour

              3.  Online and Off-line Display or Print of Data

                         full record                          $.17
                      partial record                           .15
                      minimal record                           .03

              4.   Search Terms                               $.03

Plus additional fees for use of selective dissemination of information (SDI) profiles, off-line retrospective search and republication of records.

NTIS does not provide public online access to these databases except through private vendors. Table IX below gives examples of various vendor prices for the NTIS, AGRICOLA, and RTECS databases.

       NTIS Bibliographic Database


              DIALOG                   $90 per hour connect time plus $.75 per record
              DATASTAR                 $87 per hour connect time plus $.67 per record
              BRS                      $70 per hour connect time plus $.62 per record 

       

       AGRICOLA


              DIALOG                   $45 per hour connect time plus $.55 per record
              OCLC                     $20 per hour connect time plus $.75 per record 

       RETECS


              DIALOG                   $84 per hour connect time plus $.50 per record
              DATASTAR                 $56.40 per hour connect plus $.25 per record 

FEDWORLD

Until recently NTIS avoided the development of online information systems. In 1993, however, NTIS created FEDWORLD, an online information system that was started on a shoestring with a 486 computer. NTIS plans on vastly expanding the FEDWORLD program. When FEDWORLD was first put online, it was a simple bulletin board service with a clever way to connecting users to other agency computer systems, using standard modems. NTIS has recently increased the FEDWORLD staff from two to seven, and it plans to make a number of large databases available, including a growing collection of databases that will only be available for a fee.

At present there are three databases on FEDWORLD published by NTIS that require fees. These are the Davis-Bacon Wage Determination Database, which is priced at $500 per year for up to 3 hours of searching per day, the FAA Airworthiness Directives, which are priced at $145 per year, and the FDA Important Detention List, which is priced at $360 per year, and delivers data through electronic mail. Like the GPO Access online Federal Register and Congressional Record, these products are priced for intensive users, and the pricing scheme creates a barrier that will prevent most episodic users from searching the three NTIS online databases that require fees.

NTIS also "sells" disk space to agencies who want to use FEDWORLD to disseminate their data for free. At present NTIS charges $5,000 for 5 megabytes of disk space on FEDWORLD, plus an additional $250 per megabyte for space greater than 5 megabytes. Since disk storage costs less than $1 per megabyte, the $1,000 to $250 per megabyte charge for storage is a profitable service for NTIS. According to NTIS officials, the agencies compare their costs of maintaining their own bulletin boards, including the costs of responding to data users who encounter problems, and compare that to the cost of using NTIS's FEDWORLD. GPO officials say that they are planning a similar service, which they will price significantly below the NTIS FEDWORLD price.

NTIS and OMB Circular A-130

OMB Circular A-130 requires federal agencies to disseminate information at prices which do not exceed dissemination costs. The present funding of NTIS precludes the agency from following OMB A-130, however, at least with respect to individual databases. Since agencies can use NTIS to disseminate their data on an exclusive basis, and NTIS shares profits with agencies, the provisions of OMB Circular A-130 are largely meaningless in terms of any restraint on the pricing of government information.

Copyright of Government Information

The federal copyright law prohibits the copyright of United States Government Works. According to 17 USC Section 105, "Copyright protection under this title is not available for any work of the United States Government, but the United States Government is not precluded from receiving and holding copyrights transferred to it by assignment, bequest, or otherwise." Section 101 of the copyright act further defines the "work of the United States Government" to be a "work prepared by an officer or employee of the United States Government as part of that person's official duties." This provision does not apply to information which is prepared by a federal contractor, or to state or local government bodies.

Even though the federal government may not copyright most public documents, agencies such as NTIS have been able to exercise copyright type control over data using contracts, which specify how information can be disseminated. Persons who violate the provisions of the these contracts cannot be sued for copyright violations, but NTIS or other agencies may attempt to limit access to the data in the future, although there is controversy over the legality of such contracts.

Private vendors often claim copyright of databases of federal government information, through a variety of mechanisms, such as the copyright of a compilation or arrangement of data, copyright of grammatical corrections of written court decisions (West Publishing), copyright of page numbers of published court decisions (West Publishing) or other assertions of property rights. In a number of cases, commercial data vendors insert messages which assert copyright of data which is clearly not subject to copyright. In other cases, private vendors obtain copyrights from data that is funded through public appropriations, under the sections of the copyright law which allow private contractors to obtain copyrights. Some federal agencies in turn ask that contractors assign copyrights back to the government. Yet another copyright issue concerns the use of copyrighted software to control access to government data which is stored in proprietary formats.

Concluding Comments

Government information is a property type different from other assets in terms of important economic, political and social characteristics. Unlike physical property, information can be shared without diminishment. Access to government information is essential for citizens who want to monitor and influence the government, and if access is too expensive it will deter low income persons or disinterested citizens from taking an active role in governance. Access to information is also often tied to the agency mission so closely that policies which create barriers to access to information undermine the usefulness of the agency itself.[27]

free or fee

There should be no presumption that information should be sold rather than given away. The government has always provided a number of mechanisms for the free distribution of information, running the gamut from Congressional franking privileges and White House press releases to the free distribution of bills and GAO reports (in paper formats), as well as countless other items. Other information, much of it not very different in type, is sold for a price. Often the decisions to sell or disseminate for free depend upon the whims of agency officials or lower level technicians, who may or may not have considered the broader consequences of their actions.

In determing whether or not to charge for information products and services, we recommend that agency officials consider a number of factors, including:

1-1. Does the dissemination of the information enhance the agency mission?

1-2. Is public access to the information important for public accountability and democratic participation?

1-3. To what degree is the information for the benefit of a specific group? What is the public interest in disseminating the information to this specific group?

1-4. What will it cost the government to disseminate the information for free?

1-5. What will it cost the government to create an accounting and billing mechanism so that the information can be sold?

1-6. What will be the impact of a system of user fees on the dissemination of the information and how will that affect the agency mission?

1-7. How will a system of user fees change the ways the public obtained access to agency information in the past?

limits on fees

If, after consideration of these factors, a decision is made to charge user fees for information products and services, agencies should be directed to observe certain constraints.

2-1. Prices should not exceed the incremental cost of providing public access. If the agency would have developed the information product or service for its own purposes, it should only require the public to pay for the incremental cost of obtaining access.

2-2. If public access to information is essential for agency accountability and democratic participation, provisions should be provided for waivers or reductions of fees for non-commercial or public interest uses of the information.

2-3. In setting fees, agencies should search for mechanisms, such as "free after six," which allow the agency to balance the goal of financing dissemination through user fees with the need to broaden public access to information resources.

2-4. Agencies should consider a wide range of pricing options to accommodate different types of users, and allow the most efficient use of the information services.

Finally, it is extremely important that agencies be required to provide mechanisms for dialogues with citizens over agency product lines, prices and other information management policies. Too often agencies fail to adopt to changes in technologies or are extremely insensitive to user concerns and needs. Provisions for periodic review of agency policies and performance (as envisioned in HR 629) provide a useful prod for accountability and dynamic change which is sorely needed today.

Endnotes:

[1] In the case of the exploration of oil leases, the state of Alaska created a mechanism that allowed firms to join exploration "units", which pooled exploration efforts. Failure to join an exploration unit resulted in more rigorous diligence requirements. Thus, the landowner (the state of Alaska) was able to avoid many free rider problems associated with exploration information externalities.

[2] Marginal cost pricing does not ensure that the total willingness to pay will exceed the total cost of production. This will depend upon the fixed costs associated with production.

[3] Such as the U.S. government support for CommerceNet.

[4] A monopsony exists when there is only one buyer for a product or service.

[5] The NAL receives limited discounts for copies of the CD-ROM used in house, but not for other Department of Agriculture users.

[6] Mixed systems of fixed payments and usage based fees are also widely used. DRI, for example, offers contracts for some libraries that combine large fixed payments with deep discounts on data point charges.

[7] Frank P. Ramsey, "A Contribution to the Theory of Taxation," Economic Journal 37 (March 1927):47-61

[8] GPO is required specifically to provide online access to the Congressional Record , the Federal Register and an electronic directory of federal government information. GPO is also instructed to provide online access to appropriate publications distributed by the Superintendent of Documents, and information provided to GPO for online dissemination from other federal agencies.

[9] "Depository Library Program" Title 44 United States Code (USC) , Chapter 19, 1988 ed.

[10] In 1993, GPO reported sales revenue of $84 million. Expenses were $ 23 million for the cost of the goods, $11 million for postage and $5.3 million for surplus publications, for a total of $39.6 million. This leaves a gross profit of $44.5 million, for a gross margin of 53 percent, or a mark-up of more than 100 percent over cost. However, GPO assigns another $40 million in general and administrative expenses to the sales program, leaving net earnings of only $3.6 million

[11] The government pays for the initial press run.

[12] This issue was raised in debates over the GPO ACCESS legislation. See James P. Love, "The Marketplace and Government Information," Government Publications Review 19 no. 4 (July/August 1992): 398-412; and James P. Love, "A Window on the Politics of the Government Printing Office Electronic Information Access Enhancement Act of 1993," Journal of Government Information 21, no. 1 (January/February 1994):3-13.

[13] GPO provides dial-in access for persons who do not have their own Internet access.

[14] There are more than 10,000 libraries in the United States, of which about 1,400 will receive free online access to the GPO ACCESS program through the federal Depository Library Program.

[15] See: Love, "A Window on the Politics of the Government Printing Office Electronic Information Access Enhancement Act of 1993," Journal of Government Information 21, no.1 (January/February 1994): 3-13.

[16] For usage on weekends or after 6 p.m. EST on weekdays.

[17] GPO has been monitoring and occasionally participating in discussions on GOVDOC-L, an Internet list made up principally of federal government documents librarians. While GOVDOC-L is a very good list to follow serious debates over federal information policy, it is not a general interest list, due to the heavy traffic of messages that are of very specialized and limited interest concerning document shipments.

[18] At one point GPO was considering publishing Postcript versions of the bills on the FBB. A very small percentage of data users have Postscript printers, so this choice of format would have severely curtained public access to the legislation

[19] "Public Information; Agency Rules, Opinions, Orders, Records, and Proceedings," Title 5 United States Code, Sec. 552 (4)(A)(i).

[20] Title 5 USC, Sec. 552(4)(A)(ii)(I).

[21] Title 5 USC, Sec. 552(4)(A)(ii)(II)

[22] Title 5 USC, Sec. 552(4)(A)(ii)(III)

[23] Title 5 USC, Sec. 552(4)(A)(iii)

[24] In SDC Development Corp v. Mathews, 542 F 2d 1116 (1976), Justice Anthony Kennedy, then a judge for the 9th Circuit Court, ruled that computer tapes created by the National Library of Medicine and available for sale to the public for $50,000 were not subject to the pricing rules set out under FOIA.

[25] "Clearinghouse for State and Local Initiatives on Productivity, Technology, and Innovation," Title 15 USC Sec. 3704b-2

[26] According to personal communication with Nancy Dehncke, Manager of the NTIS Announcement Products Division, on June 8, 1994.

[27] Consider, for example, the Bureau of the Census, the SEC Full Disclosure Program, MEDLINE, AGRICOLA, LANDSAT or the Federal Elections Commission.