Ronald C. McMahon, "Cost Recovery and Statistics Canada"
Government Information in Canada/Information gouvernementale au Canada, Volume 2, number/numéro 4 (spring/printemps 1996)

Cost Recovery and Statistics Canada 1

Ronald C. McMahon, Saskatchewan Bureau of Statistics 2

In the mid 1980s, Statistics Canada adopted a policy of cost recovery, on grounds that those who directly benefit from government goods and services should pay for them. However, cost recovery involves charging for data that may be considered a public good. Moreover, revenues generated through cost recovery appear small relative to direct, taxpayer supported funding. It is recommended that Statistics Canada abandon cost recovery and adopt a system of charges based on the costs of moving data from its finished state to the user.

Vers le milieu des années 1980, Statistique Canada a adopté une politique de recouvrement des coûts, s'appuyant sur le fait que ceux qui prennent directement avantage des biens et des services du gouvernement devraient payer pour ceux-ci. Le recouvrement des coûts implique cependant des charges pour des données pouvant être considérées un bien public. De plus, les revenus générés par le recouvrement des coûts semblent minimes relativement à la capitalisation directe prise en charge par les contribuables. Il est recommandé que Statistique Canada abandonne le recouvrement des coûts et adopte un système d'exigence de prix selon les coûts de déplacement de données, de leur état fini, à l'utilisateur.


In the mid 1980's Statistics Canada along with other government departments began to implement a policy of cost recovery . This policy was later formalized in a Federal Treasury Board document which outlined both the policy and the procedures for its implementation. 3 The policy directs suppliers (i.e. departments) to "...identify the full costs of providing goods, services, property, and limited rights and privileges to external users; identify the market value of property; determine appropriate user charges; and impose them in accordance with the principles of [the] policy." 4

The rationale for implementing this policy included making government more business like and shifting more of the financial burden from the taxpayer to those who benefit most directly. In addition, it was noted that the policy would increase "...the responsiveness of supply to users' willingness to pay..." and it would therefore encourage "...efficiency in service delivery." However, "...the government [was] committed to an open process and [was to be] sensitive to the particular circumstances of individual sectors." 5

The policy was to apply to the items mentioned above but was not to apply to cash transfer payments or to charges between federal government departments. Any other exemptions were to be approved by Treasury Board. Charges were to be "...based on market value or a reasonable approximation of it..." subject to a number of exceptions. 6

Exceptions which allow for less than full cost recovery exist in order to maintain consumption and so that any affected party would not have their financial position impacted in a serious manner. Furthermore, less than full cost recovery was justified where demand was insufficient to cover the full cost or where users provided input or support at no cost. However, "...charges [were] not [to] be levied where the cost of the cheapest method of implementing and administering them would be excessive in relation to the revenue collected..." 7

Treasury Board viewed user charges as a good management practice especially since the increased revenue from these charges was, in general, to be used to reduce the appropriation required to finance the supply of those goods and services. The reduced programme need for tax dollars would aid deficit reduction which was part of the reason for the cost recovery policy.

To encourage the application of user charges certain incentives were put in place. Managers were allowed to reallocate person-year savings, revenue could be used to improve service while improved services facilitated an increase in user charges. Finally, new activity could be undertaken providing the corresponding revenue financed all or part of the new activity. 8

Public and Private Goods

The Government of Canada recognized that the services it provided generated broad public benefits (often public goods), however, some services benefited specific groups (private goods) and therefore financing these activities through user fees was deemed to be more equitable. The method of financing government output was to be based not only on equity but was to be efficient in allocating government resources and easy and cheap to administer. 9

According to economic literature 10 public goods display the following characteristics:

  • they are nonexcludable in that a person cannot be stopped from using the good even if they did not pay for it (free-rider problem);
  • the good is nonrival because the consumption by one person does not affect the supply available to others; and
  • generally the private market economy will not provide them because they cannot be produced and sold at a profit.

The marginal cost of providing the good to an additional consumer is effectively zero. Further, it is not necessary to exclude consumers or ration the good to conserve it. Because of these characteristics it may be costly or impossible to exclude individuals from consuming or using existing public goods.

Apart from special surveys and possibly non-standard tabulations from its data bases, Statistics Canada's cost recovery activities involve copyright and charging for data that may be considered a public good. These and similar products and services provided by the federal government are informative by nature and they would probably not be provided by the private sector. For the most part these goods and services may be classified as public goods and as such should probably not be subjected to the user pay philosophy. It is this class of goods and services which is the subject of the remaining sections of this paper.

It should be noted that not all goods and services sold by government are of this type; for example ferry tolls and fees for camping spots in a national park. There are also intellectual property and related products subject to copyright that would be considered private goods. Published novels, music CDs and Mickey Mouse wearing an RCMP uniform would probably fit into this category. Here, those holding the copyright for the Mickey Mouse logo and the RCMP uniform would in all likelihood charge licensing fees for the use of these symbols.

Crown Copyright

Federal government cost recovery policy flows from the fact that the Canadian federal government has copyright in the works it produces including intellectual property. 11 The right is perpetual and is rooted in seventeenth century Britain at the time of the restoration of Charles II well before England's first copyright law was passed in 1710. 12

In the United States the federal government does not have copyright to the works it produces. However, U.S. state and municipal governments can claim copyright as can their Canadian counterparts at the provincial and municipal government level. In Canada both the federal and provincial governments are not bound by the copyright act and are immune from suits for infringement; this is not so with their U.S. counterparts. 13

Copyright is a territorial right and even though the U.S. federal government is unable to copyright material it produces in the U.S. it can be granted copyright outside the U.S. where the laws in that jurisdiction allows that protection -- which is the case in Canada. This could pose a potential problem for Canadian users of the Internet who access U.S. federal government databases located in the U.S. Is the copyright law in Canada being circumvented by moving these data from the U.S. to Canada? Will these actions eventually be covered by NAFTA or the World Trade Organization agreement of 1994 on the basis that differential treatment is not justifiable? 14

In Canada the federal government has produced a document concerning the use of the Internet by government organizations. It notes that Crown copyrighted material may be used by anyone without securing prior permission providing the material is not being made available for resale and providing that the originating agency has included a notice to this effect in their hard-copy publication. (In the past Statistics Canada required users of its data to acknowledge the source of any information reproduced or quoted from their public documents. 15) In addition, all agencies must respect any agreements the originating agency may have with the private or public sector concerning the use of software and information. 16

Statistics Canada and Cost Recovery

Statistics Canada, being an entity of the federal government has the right under law to copyright intellectual property along with the works it produces. In addition, it may recover all or a portion of the costs it incurs in producing output in accordance with Treasury Board directives.

However, it should be noted that in producing statistical output the costs are sunk and raising prices will not call forth additional product unless the production of copyright material becomes a business activity. When this happens in a for-profit atmosphere there is a serious question as to whether the government ought to be in this type of business.

Because the statistical output of Statistics Canada may be deemed a public good, and since there is high social value in the free flow of information any attempt to maximize economic rent through user fees may not maximize social or aggregate economic welfare. Therefore, for allocative efficiency the output should be financed by government taxes. 17

Statistics Canada practices cost recovery for the following purposes: 18

  • to determine the real demand for goods since free goods have no value and do not indicate real demand ;
  • to have those who benefit directly pay for a good rather than the taxpayer;
  • to ensure that resources are spent on statistical programmes as opposed to spending it on customized outputs;
  • to recover out-of-pocket costs for serving information requests;
  • to cover the costs of conducting special surveys; and
  • to comply with Government cost recovery policy.

The bulk of Statistics Canada's revenue comes from their parliamentary appropriation for Census and non-Census purposes. These appropriations vary from year to year depending on the Census cycle. However, the appropriation for the Census is much larger in the year that the Census is conducted. Based on information received from Statistics Canada, the non-Census appropriation reached a peak of $267 million in the 1992-93 fiscal year and has since declined in the face of fiscal restraint imposed by the federal government. Revenues generated from cost recovery have remained fairly stable since the beginning of the 1990's and have not replaced the funds lost through budget cuts (see Tables 1 to 4). 19

Statistics Canada generates approximately $13 million from the sale of both Census and non-Census products and services. The sale of Census material accounts for less than one percent of total revenue while the total sales of all products and services adds less than five percent to revenue from all sources. The revenue generated by Statistics Canada from the sale of products includes revenue paid by another government department to enable Statistics Canada to place statistical material in Canada's designated depository libraries. It also includes revenue raised from specialty products such as international trade data on CD-ROMs. These revenues are not net of marketing costs.

In addition, Statistics Canada conducts special surveys under contract to other government organizations and currently generates more than $25 million in revenue from this activity. According to Statistics Canada the revenue raised from this source covers all identifiable costs including overhead. 21

While these prices may not necessarily appear to be excessive, a comparison of the cost for an equal amount of Canadian and United States data puts the pricing question in perspective. According to Statistics Canada 450 megabytes of geographic information on a CD-ROM can be purchased from the U.S. Bureau of the Census for $7,500. Statistics Canada would charge $127,000 for the same amount of Canadian data. 22 It also should be noted that Statistics Canada depends on cities and provinces to update its street network files. These files are used to conduct the Census in Canada. Perhaps Statistics Canada should provide the provinces with standard Census products as a quid pro quo for this support.

As a further example of the pricing question a U.S. firm purchased the complete pre-processed Census data from the U.S. Census Bureau for the regular price of $8,100. These data were originally stored on 82 CD-ROMS. The company reduced this to five CDs by processing the data into an SPSS format which it sells for $1,000. 23 In addition, all of the U.S. Census product is available on the Internet free of charge at

The Canada Census population totals or the totals for any population attribute are derived by first calculating the totals for individual geographic areas within a province or territory. The cost of preparing these data was originally financed by a special parliamentary appropriation for the Census. In other words it is paid for by the Canadian taxpayer. Cross classified population aggregates at the provincial/territorial level, such as age by sex, are generally made available by Statistics Canada free of charge. However, if you require the same data for a sub-provincial area you will be faced with cost recovery. This has restricted the use of these data and has generated very little revenue. In fact, over the life cycle of the Census (a five year period) Statistics Canada has raised less than one percent of its total revenue from the sale of Census products (see Table 2).

Statistics Canada has further restricted the usefulness of its data by introducing licensing agreements which have effectively limited the distribution by third parties of both its electronic and hardcopy products. 24

However, Statistics Canada has entered into a number of innovative agreements within its cost recovery framework in an attempt to make its products less costly to some users. For example, universities are able to access CANSIM (Statistics Canada's statistical data base) for $995 per year with access being restricted to university account holders. High school students are also being exposed to Statistics Canada's data via E-STAT which is a CD-ROM containing a number of data series. The cost for this product is probably far above the actual transfer costs, i.e. the cost of placing the data on the CD-ROM. Finally, out-of-date publications are being sold to students at a reduced price. 25 Unfortunately, this approach has given privileged low cost access to some groups in society while excluding others.


Statistics Canada's cost recovery initiatives have not gone unnoticed. Professor Foot of the University of Toronto feels that the implementation of cost recovery at Statistics Canada has hindered the free flow of information. This in turn, he states, will lead to less informed users. 26 Peter Calamai, a member of the Open Government Campaign which is asking for better access to government information, recently proposed that governments should provide free on-line electronic access to data collected at public expense. He also decried what he considered to be predatory pricing policies and the use of exclusive licensing arrangements for information gathered with money provided by the Canadian taxpayer. 27 Canadians also provide data to Statistics Canada without charge and pricing policies which are viewed as unnecessary or excessive may eventually erode public support for the Census and other household surveys.

The December 1994 issue of the University of Kansas Science Bulletin notes that the exorbitant cost of Canadian data in comparison to U.S. data has resulted in Canadian researchers making policy recommendations based on U.S. data. In addition, the article observes that Canadian students are becoming more familiar with U.S. data because Canadian data are too costly to access. 28

Statistics Canada has been told directly that its pricing policies discourage the use of its data. In a review of the Industry Division, by a private consultant, members of a focus group stated that it was cheaper to get information outside of Statistics Canada. They felt that they were paying for economic analysis as part of the data they were trying to access. 29

The practise of cost recovery also becomes very troubling in the area of analysis. In recent years Statistics Canada has become more involved in analysing, interpreting and describing the public policy aspects of the data it collects rather than simply commenting on the data itself. 30 Although shelf-tables may be made available at no cost to the user these tables will more than likely be designed to provide the foundation for Statistics Canada's interpretation of the data in question. Any outside researcher wishing to challenge the findings may be faced with large costs as they attempt to retrieve data based on a structure incompatible with shelf-tables .

Cost Recovery at Statistics Canada -- Does it Work?

The government implemented cost recovery to achieve the following purposes:

  • fairness;
  • managing government resources; and
  • deficit reduction.

One of the stated purposes of cost recovery was to achieve fairness so that costs would be shifted from the taxpayer to those who most directly benefit. This action was also backed by the Chief Statistician. 31 On the surface this appears to be a reasonable goal. Unfortunately, the amount of revenue generated by Statistics Canada through cost recovery (less than five percent of its total revenues) does not support this ideal. If the revenue received from other government departments for the sale of products and services (see Table 1) is factored into the government source of revenue for Statistics Canada then it is probable that the taxpayer funds more than 97% of their budget. For example, the funding for the placement of Statistics Canada's publications in depository libraries comes from another government department yet is viewed as revenue from cost recovery. This policy fails the fairness test.

Has the government improved the management of resources? Cost recovery was supposed to introduce a market type discipline on the demand for and supply of goods and services provided by the government. Since in economic terms Statistics Canada's outputs are public goods, the type of discipline envisioned by this policy is impossible to attain. Instead we have users who complain, refuse to pay and generally attempt to find alternative sources for their information needs. This policy fails the improved management of resources test.

Finally, has cost recovery contributed to deficit reduction? In the case of Statistics Canada this is a very difficult question to answer. If we were able to identify non-government cost recovery revenue and Statistics Canada's expenditures for enhanced marketing services this question could be answered. Given the additional effort involved in policing and formalizing licensing agreements and in actively pursuing the sales of products and services, the revenues generated through cost recovery less associated expenditures probably net to zero. It may give the illusion that additional revenues are being made available for expenditures elsewhere in the organization, however, in all likelihood this policy fails the deficit reduction test.


It is recommended that Statistics Canada discontinue the practise of cost recovery for goods and services that have been generated in the pursuit of its taxpayer funded activities and move to a system of charges based on transfer costs . Here transfer costs are defined as those costs incurred in moving data from its finished state to a medium as required by the data user. In addition, the use of covenants such as licensing agreements which restrict the use or distribution of goods and services provided by Statistics Canada should be discontinued.

Furthermore, it is recommended that Statistics Canada provide access to its major data bases such as CANSIM via the Internet or other access modes without charge. This approach will facilitate the democratization of data and should lead to a more informed and enlightened public most of whom provide data free of charge to Statistics Canada many times during their lives.


Statistics Canada Revenue From All Sources
By Selected Fiscal Years
Table 1Revenue Flowing to Statistics Canada Including the Census
(Millions of Dollars)
ParliamentaryParliamentarySales ofTotalTotal
FiscalAppropriationAppropriationSpecialProducts &ExcludingIncluding
YearFor CensusNon-CensusSurveysServicesCensusCensus
($ Millions)
1990-9154.0227.619.30.211.711.9258.6 312.8
1992-9329.8267.423.01.711.513.2301.9 333.4
1993-9431.9239.526.91.611.112.7277.5 311.0
1994-9535.0243.627.00.512.312.8282.9 318.4
Source: Statistics Canada, Communications and Operations Field
Table 2Percent of Revenue, by Source, Including the Census
ParliamentaryParliamentarySales ofTotalTotal
FiscalAppropriationAppropriationSpecialProducts &ExcludingIncluding
YearFor CensusNon-CensusSurveysServicesCensusCensus
Source: Statistics Canada, Communications and Operations Field

Statistics Canada Revenue Sources Excluding Census Revenue
By Selected Fiscal Years
Table 3Revenue Flowing to Statistics Canada Excluding the Census
(Millions of Dollars)
ParliamentaryParliamentarySales ofTotalTotal
FiscalAppropriationAppropriationSpecialProducts &ExcludingIncluding
YearFor CensusNon-CensusSurveysServicesCensusCensus
($ Millions)
Source: Statistics Canada, Communications and Operations Field
Table 4Percent of Revenue, by Source, Excluding the Census
ParliamentaryParliamentarySales ofTotalTotal
FiscalAppropriationAppropriationSpecialProducts &ExcludingIncluding
YearFor CensusNon-CensusSurveysServicesCensusCensus
Source: Statistics Canada, Communications and Operations Field


[1] May be cited as/On peut citer comme suit:

Ronald C. McMahon, "Cost Recovery and Statistics Canada," Government Information in Canada/Information gouvernementale au Canada, Vol. 2, no. 4.3.


Originally prepared for presentation at the Federal/Provincial Committee on Data Dissemination, December 5, 1995 and published at <>


Ronald C. McMahon
Saskatchewan Bureau of Statistics
5th Floor
2350 Albert Street
Regina, SK
S4P 4A6

[3] Decision of the Treasury Board Meeting of December 7,1989, document T.B. 812553.

[4] Ibid. p. 3.

[5] Ibid. pp. 2-3.

[6] Ibid. pp.5-7.

[7] Ibid. Appendix A pp. 1-2.

[8] Ibid. Guidelines pp. 14-21.

[9] Decision of the Treasury Board Meeting of December 7,1989, document T.B. 812553, Executive Summary, pp. 3-4.

[10] See: John F. Due and Ann F. Friedlander, Government Finance Economics of the Public Sector , Richard Irwin Inc., 5th ed., 1973, p. 3. J. C. Strick, Canadian Public Finance , Holt, Rinehart and Winston of Canada, 4th ed., 1992, p. 10. Robin W. Boadway and David E. Wildasin, Public Sector Economics, Little, Brown and Co., 2nd ed., 1984, pp. 58-60.

[11] Treasury Board document 812553, Appendix B, p. 4. [Intellectual property includes property such as Crown copyrighted material and patentable designs and processes, and all other interests related to the disposition of such property (e.g. licensing arrangements).]

[12] David Vaver, Copyright and the State in Canada and the United States, Osgoode Hall Law School, York University, Toronto pp. 1-2.

Available at:

[13] Ibid. pp. 1-2, 9.

[14] Ibid. p. 11.

[15] In the early 1980's Statistics Canada publications carried the following note: "Statistics Canada should be credited when reproducing or quoting any part of this document."

[16] Eleanor Zazulak, The Internet: A Guide to Internet Use in the Federal Government, Information Management Partnership Publishing, 1995, p. 16.

Available at:

[17] W.T. Stanbury, Aspects of Public Policy Regarding Crown Copyright in the Digital Age, Faculty of Commerce and Business Administration, University of British Columbia, pp. 5-6.

[18] Letter dated July 21, 1995 from Yvon Goulet, Assistant Chief Statistician, Communications and Operations, Statistics Canada, Ottawa to Ron McMahon, Director of the Saskatchewan Bureau of Statistics, Regina, Saskatchewan.

[19] Ibid.

[20] Ibid.

[21] Statistics Canada, 1991 Census Catalogue (catalogue 92-302E).

[22] Letter dated May 13, 1994 from Victor B. Glickman, Director Geography Division, Statistic Canada, Ottawa to Mr. David Mckie, Editor, Media Magazine, Ottawa in response to an article written by Peter Calamai for the April 1994 edition of Media Magazine.

[23] WESSEX Company is located near Chicago, Illinois and may be reached at 1-800-892-6906.

[24] Hardcopy publications contain a statement limiting the use of the data contained in the catalogue (see Statistics Canada Catalogue 13-001). Electronic products have a limited use licensing agreement at the front of the electronic file the contents of which you must agree to before you allowed to access the product.

[25] Statistics Canada, Perspectives on Labour and Income (catalogue 75-001E), Winter 1994, pp. 21-22.

[26] Ibid. p. 20.

[27] Media Magazine, Ottawa, April 1994, pp. 24-35.

[28] University of Kansas Science Bulletin, December 1994, Volume 6, No. 10.

[29] D.R. Hartley Consultants Ltd., Final Interpretive Report on the Industry Division, p. 35.

[30] For examples of the interpretive nature of recent Statistics Canada output see Violence Against Women Survey, November 1993.

[31] Statistics Canada, Perspectives on Labour and Income (catalogue 75-001E), Winter 1994, pp. 21.