Laments are sometimes heard about the quantity of tax research in Canada being so much lower than in the United States. However, many types of tax research cannot be done without access to tax data, and Canadian governments have been reluctant to share their data with tax researchers and the general public. This can be expected: governments naturally seek to protect themselves from the criticisms that could be the outcome of such research. Also, it is difficult to provide data to researchers while limiting the access of political parties, interest groups and provincial governments. Recent actions by different arms of the federal government suggest the data-access problem is getting worse with respect to aggregate statistics but better with respect to “microdata—”data which is from the tax returns of individual taxpayers but which is completely anonymous (stripped of names, social insurance numbers and addresses).
Tax researchers in Canada have long been envious of their counterparts in the United States, who have the benefit of the public-use tax microdata files produced by the Statistics of Income Division of the Internal Revenue Service. Because these files contain tax-return data on individual taxpayers—including some files following particular taxpayers over time—it is possible to do all kinds of statistical studies estimating the elasticities (responsiveness) of investment, charitable donations, tax-assisted saving for retirement, taxable income, and so on to changes in the tax system.
The only Canadian public-use tax microdata file— Statistics Canada’s Social Policy Simulation Database and Model (SPSD/M)—is designed solely for simulating tax and social policy changes. It is excellent for that purpose, as it contains unemployment and asset-debt-expenditure information not found on any other Canadian tax database. However, the SPSD/M cannot be used for the behavioural studies that have made the US Statistics of Income data so valuable because, to avoid invading taxpayers’ privacy, each record on the file is a composite of the data from several individuals. For example, the moving expense figure may come from one individual, while the child-care expense figure may come from another person with a similar income and age. Behavioural responses can still be examined using elasticities imported from elsewhere, but to actually measure these elasticities requires data on actual individuals.
The privacy issue is a serious one in light of the “Big Brother” criticisms of the proposed Longitudinal Labour Force File of Human Resources Development Canada and the current concerns about the release of 1908 and 1911 census data. Also, although public-use microdata files do not include individuals’ names, there remains a concern that someone who has access to another private database on individuals, such as drivers’ licence data or an employer’s payroll records, could merge the two files through common data elements and identify particular individuals. The US Statistics of Income Division tries to prevent this from happening by blurring and masking the data prior to public release, but the effectiveness of these procedures is uncertain. Therefore, the SPSD/M is the best Canadian tax microdata likely to be made publicly available.
Fortunately, a counterpart to the Statistics of Income data does exist in Canada—the microdata file on individual taxpayers produced by the Canada Customs and Revenue Agency (CCRA) for the use of the Department of Finance and Statistics Canada. Recently, researchers have had some success in obtaining access to this data in ways that resolve privacy concerns—either directly through on-site visits by the researchers or indirectly through statistical work executed by employees of these departments as time and departmental work priorities permit. For the Department of Finance, this is strictly on a caseby-case basis with no written guidelines, and the factors that are likely to be considered include the Department’s own interest in the issue and the potential for the findings to cause political embarrassment to the minister. For instance, projects involving RRSPs and the shifting of corporate taxable income among provinces have been approved.
Rather than approaching the Department of Finance, researchers should consider going though Statistics Canada because of its commitment to public data access and its lack of concern about potential political controversy. Also, Statistics Canada is considering establishing regional data centres so researchers do not have to travel to Ottawa, although whether this would extend to the highly sensitive tax data remains to be seen. Access is through the Small Area and Administrative Data Division. Michael Veall of McMaster University has published two papers using this source, one of which was written with Mary-Anne Sillamaa.
Other sources of tax microdata also exist. The provinces generally have both personal and corporate income tax data, and Quebec has allowed access to its provincial personal income tax data for a study on the alternative minimum tax. CCRA itself also seems willing to entertain proposals for access to both its personal and corporate data. Statistics Canada also has corporate income tax data, which it uses to produce an annual tax statistics publication.
Access to microdata through these channels would seem to be consistent with the spirit of the federal Access to Information Act and similar provincial statutes. However, such access cannot actually be enforced because the Act is usually interpreted to mean only that analysis that has already been done be made public, and not that new analysis be done.
In the absence of full public access to microdata, tax researchers rely on aggregate data. Beginning in 1979, and now annually, the Department of Finance publishes a tax expenditure account giving the total revenue cost of selected features of the Canadian tax system such as RRSP deductions and the charitable credit. While this publication is certainly valuable in understanding the importance of different aspects of the tax system, it tells us nothing about how the use of these features is distributed by income level, province, age and gender. For that information, and for further information by major source of income, gender and locality, it is necessary to go to the other source of aggregate tax data, the annual volume of statistics on the personal income tax system published by the CCRA.
First published in 1946, this volume is known to generations of tax policy analysts as the “Green Book,” although its official title has varied from “Taxation Statistics” to the current less informative title of “Income Statistics.” Its readership includes the news media, Members of Parliament, officials of federal, provincial and municipal government departments, academic researchers, and anyone else interested in the workings of the personal income tax system.
Journalists are mostly interested in the Green Book for the data it provides on which localities in Canada have the highest and lowest annual incomes and the number of high-income individuals who did not pay federal income tax in a given year. Researchers have found other uses. For example, University of Western Ontario economist James Davies’ work on the changes over time in the average person’s marginal tax rate was based on this data. The present authors used this data to study how much parents reduce their tax payable by splitting property income with their children, for which the government’s solution was the “kiddie tax” introduced in the 1999 budget. Earlier this year, Robert Gagné, Jean-François Nadeau and François Vaillancourt used this data source to conclude that decreasing the marginal income tax rates of high-income taxpayers might raise revenue.
The CCRA’s recent treatment of the Green Book should give the tax policy community some concern. The first worrisome step was the discontinuance of the print version in June 2000. Henceforth, the data will be available only by direct request to the CCRA for copies of individual tables or online through the CCRA’s website (http://www.ccra-adrc.gc.ca/tax/individuals/stats/). Although a paper-based publication may seem like an unnecessary expense in an electronic age, the savings are likely only $100,000 per year and there are four disadvantages to web-only publication:
- Poor archiving. How will the data be accessed five years from now? Libraries archive printed documents but not web pages, and whether the web pages themselves will archive old data remains to be seen (although the CCRA is committed to doing so). The web is still in its infancy, and it may or may not prove to be an effective way to archive data for future use. Because of this uncertainty, very few academic journals are electronic-only; instead, electronic versions have been added without discontinuing the print versions.
- Readability. The data consists of more than 20 tables, each of which is accessed separately; new users would be better able to understand what data is being provided with the aid of a paper copy to browse through (or at least having the option of printing a single electronic file which would contain the whole publication).
- Accessibility. Statistics Canada reports that 11 million Canadians, especially older and low-income people, did not access the Internet in 2000. Although the vast majority of users of this publication probably do not come from this group, removing opportunities for access is undesirable.
- Lower public profile. Data added to a website attracts more public attention if it is accompanied by a paper-based publication.
Notably, neither Quebec nor the United States have cancelled the paper version of their personal income tax data.
Of course, none of this is intended to suggest that the CCRA stop adding the data to its website. Access through the web is rapid, free and more convenient than paper in terms of automatically loading data into spreadsheets for further manipulation and analysis. Moreover, provincial/territorial-level versions of the tables appear on the website, even though they were never in the print version. Through the web version, people around the world have become aware of Canadian tax data. Our concern is not with web publication but with web-only publication.
The low profile of the Green Book publication is also partly due to the lack of marketing effort by the CCRA. When a new edition becomes available, there is just an entry under “What’s New” on the CCRA website, and sometimes not even that. The CCRA should issue a press release, and not just one that states blandly that the data is now available but one that follows the Statistics Canada model of highlighting three or four newsworthy numbers. If the press is interested in the number of high-income Canadians who do not pay tax, then give them that number. Controversial issues should be dealt with forthrightly by providing full information along with any needed policy explanation, rather than by attempting to bury the entire publication.
Strangely, the move to electronic-only access has not produced more rapid publication. The Green Book has normally been issued in the fall of the year of its cover date; for example, the 1998 edition came out in early December of that year. However, the first electronic-only version—the 1999 edition—did not come out until June 2000, and the 2000 edition only recently appeared, in September 2001. This extra six to nine month delay in producing the last two editions is added to the normal and largely unavoidable delay of the data being for a year that is two years prior to the cover date; thus, it is the 1998 data that was released last month in the 2000 edition.
In 1998, the CCRA responded to criticism about delays by adding a second Green Book publication of “Interim Statistics” that was to be available near the middle of the year of the cover date. However, this publication has also proved to be tardy. The 1999 edition did not come out until November 1999, and the 2000 edition finally appeared, a year late, in July 2001.
The CCRA says that heavy staff turnover during the last two years has caused the delays. This is being corrected, and the CCRA promises that 2001 editions of the main and interim publications will be out by the end of this year. In the meantime, selected statistics are available from the CCRA by private request, and users can pay Statistics Canada’s Small Area and Administrative Data Division for more prompt data. Some RRSP and capital gains data that is to be available in the Green Book’s 2001 editions has been available from Statistics Canada since October 2000.
Tax statistics are vital to tax policy research. Producing a public-use microdata file for Canada similar to the one in the United States is a desirable long-term goal, but that idea seems more and more like a pipe dream. Still, it is encouraging that access to microdata is improving, since this is likely to increase the quantity and quality of Canadian tax research. Everywhere, the general attitude seems to be toward increasing data access, although cost recovery policies and delays in access may cause problems. On the other hand, current government actions concerning the Green Book are troubling. A publication that is subject to such delays and that has been given such a low profile does not have a promising future. As an important public document on tax policy like the tax expenditure account or the budget papers, the Green Book deserves better treatment. The CCRA needs to end the delays, reinstate the paper version as a supplement to web access, and make some attempt to have the media notice when the data becomes available.
We are happy to be able to conclude on a positive note: There is now some sign of action— as this paper goes to press, the CCRA has informed the authors that it is reinstating the press release and re-evaluating the decision to discontinue the paper version.
Portions of this paper appeared in the July 2001 issue of Canadian Tax Highlights. Reproduced by permission of the Canadian Tax Foundation.