Archived

- Table of Contents - Previous - Next -

Annex 1
Risks and Uncertainties in Fiscal Projections

Introduction

In June 2005, the Minister of Finance tabled in the House of Commons an independent study by economist Dr. Tim O'Neill of the Government of Canada's approach to economic and fiscal forecasting. The report was commissioned to ensure that the Government continues to use the most up-to-date forecasting methods and to benchmark Canadian practices against the best in the world. One of the key recommendations of the report called for improving the transparency of the federal government's fiscal-forecasting procedures and information, in part by examining the key risks and uncertainties associated with fiscal projections.

This annex provides an overview of the sources of uncertainty in, and therefore the risks to, the fiscal projections. While Chapter 4 discusses specific risks to the current forecast, this annex explores the general implications of risks and uncertainties for fiscal planning.

Fiscal projections are inherently uncertain due to:

• Uncertainty associated with the economic forecasts that underlie fiscal projections.
• Volatility in the relationship between fiscal variables and the underlying activity to which they relate.
• Typically long lags with which final economic and fiscal information becomes available.

Ensuring that the annual fiscal targets of balanced budgets or better are met requires careful analysis of the risks and uncertainties in the fiscal projections.

The risks and uncertainties involved in meeting stated fiscal targets underscore the need for a prudent approach to fiscal planning. This includes the annual Contingency Reserve and an extra amount for economic prudence.

The $3-billion Contingency Reserve is small in relation to the $400 billion in combined revenues and expenses

• The budgetary balance is the difference between two very large numbers (revenues and expenses). Therefore small errors in forecasting either or both can lead to seemingly large forecast errors of the budgetary balance.
• The sum of revenues and expenses is about $400 billion and growing. A 1-per-cent underestimate of revenues and a 1-per-cent overestimate of expenses translates into an unanticipated $4-billion shortfall-greater than the Contingency Reserve.
• Every budget since 1994 has included a Contingency Reserve generally set at $3 billion. Although the nominal value of the Contingency Reserve has remained unchanged, its value as a proportion of gross domestic product (GDP) has fallen by about one-half from 0.4 per cent of GDP in 1994 to 0.2 per cent of GDP in 2004.
• Further, the revenue base that is subject to forecast error includes not only federal revenues but also provincial revenues and Canada Pension Plan premiums administered by the Canada Revenue Agency. These revenues total close to $70 billion. Since remittances initially received from the Canada Revenue Agency do not identify these streams separately, they must be estimated on a monthly basis. This adds to the complexity of the fiscal projection.

Fiscal projections are inherently uncertain

• There are three key sources of uncertainty associated with the fiscal projections: the economic forecast, the translation of the economic forecast into the fiscal projections, and information lags and data revisions.
• Economic forecasts contain an unavoidable level of uncertainty. Since the economic forecast underlies the fiscal projections, the uncertainty associated with the economic forecast carries over to the fiscal projections.
• Translating the economic forecast into fiscal projections is associated with uncertainties of its own. Although most fiscal variables are related to an underlying economic activity, the relationship between the two is sometimes volatile and hard to predict, given the reaction of taxpayer behaviour and cyclical developments. Furthermore, some fiscal variables are not related to economic factors, such as expenses related to the Government's legal liabilities.
• Finally, the typically long lags with which final economic and fiscal information becomes available can complicate the fiscal projections.

Economic forecasting is subject to error

• Most components of budgetary revenues and some components of expenses (i.e. employment insurance benefits and public debt charges) are influenced by the economic variables to which they relate.
• The gross domestic product of the country provides a broad measure of the Government's tax base and represents an important input for fiscal projections. Forecasts of GDP and other economic variables are subject to error. During periods of accelerating economic growth GDP tends to be underestimated, while during periods of decelerating or declining growth, GDP tends to be overestimated.
• Since 1994 the economic forecasts used for budget planning have been based on private sector forecasts. Forecasts are requested for calendar years starting with the budget year and up to six years thereafter.^[1] Over the last 11 years, the error between forecasts provided for budgets and currently published figures ranged from an overestimate of 2.6 percentage points in 2001 to an underestimate of 4.7 percentage points in 1999. The mean error was a 1.0-percentage-point underestimate.

Forecasts of interest rates are more accurate

• Public debt charges are strongly influenced by changes in short-term interest rates since about one-third of outstanding government debt needs to be renewed within the year.
• Forecasts of interest rates have generally been more accurate than nominal GDP forecasts.^[2]

Accurate forecasts are important since fiscal projections are sensitive to changes in economic assumptions

• Fiscal projections are sensitive to changes in economic variables, particularly to changes in real GDP growth, inflation and interest rates.
• Although the relationship between changes in economic variables and fiscal outcomes is volatile and hard to predict, rules of thumb can be used to get some sense of the sensitivities associated with economic shocks. These generalized rules of thumb assume that any increase in economic activity is proportional across the income components of GDP.
• The sensitivities shown below differ from those presented in Budget 2005 due to changes in the base levels of the various GDP components, changes in the levels of the respective revenues and expenses themselves, and changes in the assumptions regarding employment insurance (EI) premium rates. Rather than being fixed at a given level, EI premium rates are assumed to adjust such that EI revenues exactly offset program expenses; this assumption is consistent with the new EI rate-setting procedure introduced for 2006.
• In the following pages the fiscal impact of changes in three key economic variables are provided:

Estimated Impact of a One-Year, 1-Per-Cent Increase in Real GDP Growth on Federal Revenues, Expenses and Budgetary Balance

	Year 1	Year 2	Year 5
	(billions of dollars)
Federal revenues
Tax revenues
Personal income tax revenues	1.0	1.3	1.6
Corporate income tax revenues	0.3	0.3	0.3
Goods and services tax revenues	0.4	0.4	0.5
Other tax revenues	0.2	0.2	0.2
Total tax revenues	1.9	2.2	2.6
Employment insurance premium revenues	0.3	-0.7	-0.8
Other revenues	0.0	0.0	0.0
Total budgetary revenues	2.2	1.6	1.8
Federal expenses
Major transfers to persons
Elderly benefits	0.0	0.0	0.0
Employment insurance benefits	-0.6	-0.6	-0.7
Total	-0.6	-0.6	-0.7
Other program expenses	0.1	0.2	0.3
Public debt charges	-0.1	-0.2	-0.4
Total expenses	-0.6	-0.6	-0.8
Budgetary balance	2.7	2.2	2.6
Note: Totals may not add due to rounding.

• The increase in real GDP raises the budgetary balance by $2.7 billion in the first year, $2.2 billion in the second year and $2.6 billion in the fifth year.
• Tax revenues from all sources increase. Personal income tax revenues increase as employment and wages and salaries rise. Furthermore, due to the progressivity of the tax system, as individuals earn higher real incomes and move into higher tax brackets, they pay proportionally more of their income in taxes. Corporate income tax revenues rise as output and profits increase. However, the corporate income tax revenue increase is less than proportional as part of the increase in profits is offset through higher utilization of loss pools. Goods and services tax (GST) revenues increase as a result of higher consumer spending associated with the rise in employment and personal income.
• Since EI premium rates for a given year are set on projections carried out in October of the previous year, EI premium revenues are assumed to increase in the first year of the shock (due to higher employment) but decline thereafter, reflecting the adjustment to the break-even rate (changes in EI premium revenues do not exactly equal changes in EI benefits because the decline in administrative costs is not included).
• Expenses decline, mainly reflecting lower EI benefits (due to a decrease in the level of unemployment) and lower public debt charges (reflecting a lower stock of debt).

Estimated Impact of a One-Year, 1-Per-Cent Increase in GDP Inflation on Federal Revenues, Expenses and Budgetary Balance

	Year 1	Year 2	Year 5
	(billions of dollars)
Federal revenues
Tax revenues
Personal income tax revenues	1.3	1.3	1.4
Corporate income tax revenues	0.3	0.3	0.3
Goods and services tax revenues	0.4	0.4	0.5
Other tax revenues	0.2	0.2	0.2
Total tax revenues	2.2	2.2	2.4
Employment insurance premium revenues	0.4	0.1	0.1
Other revenues	0.1	0.1	0.1
Total budgetary revenues	2.7	2.3	2.6
Federal expenses
Major transfers to persons
Elderly benefits	0.3	0.3	0.4
Employment insurance benefits	0.0	0.1	0.1
Total	0.3	0.4	0.5
Other program expenses	0.4	0.5	0.8
Public debt charges	0.0	-0.1	-0.3
Total expenses	0.7	0.8	1.0
Budgetary balance	2.0	1.5	1.6
Note: Totals may not add due to rounding.

• An increase in nominal GDP resulting solely from a rise in prices would raise the budgetary balance by $2.0 billion in the first year, $1.5 billion in the second and $1.6 billion in the fifth.
• Higher prices result in higher nominal income and, as a result, personal income tax, corporate income tax and GST revenues all increase, reflecting gains in the underlying nominal tax bases. Compared to the impacts of the real GDP shock, the effects on personal income tax revenues are more pronounced in the initial year due to limited pass-through of higher inflation (tax brackets are indexed to the percentage change in the Consumer Price Index for the 12-month period ending September 30 of the previous year). For the remaining years, the impacts on personal income tax revenues are higher in the real GDP shock, reflecting higher real income and the progressivity of the tax system. For the other revenue streams the effects are similar.
• EI premium revenues increase in response to higher earnings in the first year but dissipate thereafter as premium rates are assumed to adjust to offset the impact of higher earnings. However, unlike the real GDP shock, EI benefits do not decline since unemployment is unaffected by the rise in prices.
• Partly offsetting higher revenues are increases in the cost of statutory programs that are indexed to inflation, such as elderly benefit payments, and increases in federal wage and non-wage expenses, which are assumed to increase in line with prices. Public debt charges fall due to the lower stock of debt.

Estimated Impact of a Sustained 100-Basis-Point Increase in All Interest Rates on Federal Revenues, Expenses and Budgetary Balance

	Year 1	Year 2	Year 5
	(billions of dollars)
Federal revenues	0.4	0.5	0.8
Federal expenses	1.3	2.0	2.7
Budgetary balance	-1.0	-1.5	-1.9
Note: Totals may not add due to rounding.

• An increase in interest rates would lower the budgetary balance by $1.0 billion in the first year, $1.5 billion in the second and $1.9 billion in the fifth.
• The deterioration stems entirely from increased expenses associated with public debt charges. The loss on debt charges rises through time as longer-term debt matures and is refinanced at higher rates.
• Moderating the overall impact is a rise in revenues associated with the increase in the rate of return on the Government's interest-bearing assets, which are recorded as part of non-tax revenues.

Translating the economic forecast into a fiscal projection is an uncertain process

• Even if the GDP forecasts were perfectly accurate, translating the economic forecast into a fiscal projection is not straightforward, as illustrated by the variation in the ratio of federal revenues to nominal GDP. This ratio can be thought of as the effective tax rate for the economy as a whole.
• There is a cyclical element to the revenue ratio. It tends to decline during downturns and to increase during recoveries, reflecting the progressive nature of the tax system and the cyclical nature of corporate profits and capital gains. It is also affected by the impact of tax policy changes.
• Over the last 20 years, this ratio has gone from a high of 17.9 per cent in 1991-92 to a low of 15.3 per cent in 2003-04. For an economy that produces $1.3 trillion worth of output every year, each 0.1-percentage-point change in the ratio produces a $1.3-billion change in revenues. Over the past 20 years the ratio has changed annually by an average of 0.4 percentage points.

Growth in tax bases does not translate in a predictable way into tax revenues: corporate income taxes

• The relationship between broad revenue sources (namely, personal income taxes, sales taxes and corporate income taxes) and their respective bases is subject to large variations.
• Provisions in the Income Tax Act allow corporations to carry forward losses for up to 10 years, and current-year losses can be carried back up to three years in order to offset taxes already paid in previous years.
• Such provisions contribute to the weak link between corporate income tax revenues and corporate profits.

Growth in tax bases does not translate in a predictable way into tax revenues: personal income taxes

• While over long periods of time there is a fairly stable relationship between personal income and personal income tax revenues, in any one year this relationship may not hold. The reasons for this include:
• The response of individuals to tax changes is uncertain.
• The take-up of tax-deferred income plans, such as registered retirement savings plans, can vary.
• The National Accounts measure of personal income does not include capital gains and withdrawals from tax-assisted retirement savings in registered retirement savings plans (RRSPs) and registered pension plans (RPPs).
• For example, there were several years during the late 1990s when personal income taxes grew faster than personal income, in part due to a strong stock market performance and the associated capital gains (but also due to the then only partial indexation of the tax system to inflation). In 2001-02, personal income rose almost 5 per cent while personal income tax revenues declined more than 7 per cent, reflecting the tax reduction measures under the Five-Year Tax Reduction Plan, as well as capital losses associated with a decline in the stock market.

Growth in tax bases does not translate in a predictable way into tax revenues: GST revenues

• While the GST tax base is relatively stable and generally grows in line with the economy, GST revenues can also be difficult to predict.
• GST revenues are the difference between two large numbers: gross receipts and refunds. The GST is a value-added tax, so businesses collect tax on their sales (gross GST receipts) and are refunded the tax paid on their input (GST refunds).
• Small changes in either or both of these can translate into a large change in the difference between the two. In some instances GST refunds may not be claimed in the same year as associated GST receipts, causing the growth in the tax base to differ from GST revenue growth. However, over a period of several years, the differences tend to lessen, and net GST receipts grow broadly in line with the underlying consumption tax base.

Accrual-based financial reporting meets the highest standard, but it increases forecast uncertainty

Liabilities and Assets Subject to Measurement Uncertainty

- Environmental liabilities - Legal liabilities - Employee benefits - Capital/amortization - Loans and loan guarantees

• In 2002-03, the Government changed its basis of accounting from modified accrual to full accrual. Full accrual accounting is internationally recognized as presenting a more complete picture of a government's financial position, improving information for decision making and accountability.
• Under full accrual accounting, expenses and corresponding liabilities are recorded in the periods to which they relate rather than when cash is disbursed. Changes in these liabilities are often difficult to predict since they generally do not bear any direct link to underlying economic activity. Adjustments to the Government's liabilities are audited by the Office of the Auditor General of Canada.
• In the case of environmental liabilities, the Government accounts for costs and liabilities related to the management and remediation of contaminated sites when the contamination occurs and estimates of the costs are available. For legal liabilities, the Government records an allowance for claims and litigation where it is likely, based on the legal assessment of the Department of Justice Canada, that there will be a future payment and a reasonable estimate of the loss can be made. In both cases, the estimated liabilities are subject to adjustment as circumstances change and new information becomes available.
• Since the Government recognizes costs of employee benefits in the year in which the liability is incurred (as opposed to on a pay-as-you-go basis, under which liabilities are recorded only when benefits are paid out), liabilities for current services are adjusted each year based on actuarial assessments.
• Capital expenditures made each year by various departments must be replaced by an estimate of the depreciation or amortization of the Government's overall capital stock.
• The solvency of outstanding loans and the possibility that loan guarantees may be called are also assessed on an ongoing basis with appropriate adjustments made to their value.

Lags in data availability add uncertainty to the fiscal projections

March Fiscal Monitor Compared to Final Outcomes

	1997-98	1998-99	1999-00	2000-01	2001-02	2002-03	2003-04	2004-05
	(billions of dollars)
March Fiscal Monitor: surplus	4.2	9.1	14.9	19.8	9.8	10.4	7.6	9.8
Changes to final outcome
Budgetary revenues
Personal income tax revenues	1.4	-1.0	1.8	1.0	2.0	2.1	1.8	1.1
Corporate income tax revenues	2.1	1.3	-0.1	0.6	-0.6	0.9	1.1	0.1
Goods and services tax revenues	1.3	-0.2	-0.2	0.2	-0.4	-0.1	0.0	-1.4
Other revenues	0.9	0.2	-0.2	-0.4	-1.8	1.1	1.5	1.7
Total	5.8	0.2	1.3	1.3	-0.8	4.1	4.3	1.6
Expenses
Policy initiatives	4.4	3.5	2.5	1.4	0.0	5.5	0.6	5.1
Atomic Energy of Canada   Limited environmental liabilities								2.3
Other	2.2	3.0	1.4	2.6	0.0	2.0	2.2	2.4
Total	6.5	6.5	3.9	4.0	0.0	7.5	2.8	9.8
Net adjustment	-0.7	-6.2	-2.6	-2.7	-0.8	-3.4	1.5	-8.2
Surplus	3.5	2.9	12.3	17.1	8.9	7.0	9.1	1.6
Note: The final outcome for 2002-03 was on a full accrual basis while The Fiscal Monitor was on a modified accrual basis. Totals may not add due to rounding.

• The sometimes long lags with which information on fiscal developments becomes available add uncertainty to the fiscal projections.
• A significant amount of new information arises even after the end of the fiscal year. For example, in the case of personal income taxes, the published monthly data are based on cash receipts, but are adjusted to reflect estimated year-end receivables and payables. However, at year-end, these estimates are replaced with the final, full-accrual-based estimate of personal income tax receipts. This estimate is produced during the summer months and is based on personal income tax returns assessed as at May 31.
• Policy initiatives cannot be incorporated into the monthly results until they have received parliamentary approval. Also, specific liabilities cannot be included until the amounts can be quantified by departments. This adds uncertainty since approval is not guaranteed.
• The scope of the adjustments is illustrated by comparing the estimated surplus at the time of the March Fiscal Monitor, when fiscal data for all 12 months are available, to the final results published in the Annual Financial Report of the Government of Canada. Net adjustments range from a $1.5-billion upward adjustment to an $8.2-billion downward adjustment. In most years, the adjustment is negative.

As much as 56 per cent of annual corporate income tax revenues are received in the last four months of the fiscal year

• Corporations are required to remit monthly tax instalments based on either their actual liabilities in their previous taxation year, or an estimate of their tax liability for the current year. Usually, the instalment payments are "the lesser of." However, corporations are required to file their final tax payments after the end of their taxation year. To the extent that their profits exceed the estimate used to calculate their monthly remittances, corporations may be required to make large settlement payments when they file their final tax payments.
• Corporations with a taxation year ending October 31 (on average accounting for 9 per cent of federal corporate income tax revenues) and December 31 (on average accounting for 52 per cent of federal corporate income tax revenues) are required to file their final tax returns by the end of December and the end of February/early March, respectively. The end result of this process is that corporate income tax revenues tend to be back-loaded to the end of the fiscal year. Indeed, up to 56 per cent of total corporate income tax revenues can be received in the last four months (December to March) of the Government's fiscal year, and up to 48 per cent in the last three months (January to March).
• The amount received in the final months is quite volatile, making it hard to gauge the implications of the results from earlier months.

Data revisions can complicate fiscal projections

• Periodic revisions can complicate fiscal projections. The problem data revisions create is that forecasters, both economic and fiscal, may not have an accurate reading of recent economic activity at the time they make their forecasts.
• To incorporate the most recent information and methodologies, Statistics Canada must periodically revise its data. Each year, with the May release of the National Accounts, Statistics Canada makes revisions to data from the previous four years. In the case of methodological changes, revisions can go back even further.
• Once all the revisions have been incorporated, the final figure often deviates from the value reported in the initial release. For example, the current figure for nominal GDP in 1996 is almost 5 per cent higher than the figure originally reported in May 1997. Revisions to more recent years have been more modest, but the four-year revision period for them has not yet elapsed.
• Since GDP is being revised, so are its various components. Corporate profits, for example, are subject to particularly large revisions. This matters for in-year estimates because the underlying tax bases are part of the information used to make fiscal projections; using an inaccurate level for a particular tax base will lead to inaccurate projections.

Conclusion and Implications

• The Government remains committed to maintaining balanced budgets or better each and every year.
• The risks and uncertainties involved in making fiscal projections underscore the need for a prudent approach to fiscal planning. Although it is possible to provide an assessment of where the risks and uncertainties lie, it is impossible to predict ahead of time how large the risks really are and to surmise the extent of their consequences.
• Remaining out of deficit year after year requires careful analysis of the risks and uncertainties in the fiscal projections.
• In order to attain its objective of balanced budgets or better, the Government has committed to continue its practice of including a $3-billion Contingency Reserve and to set aside amounts for economic prudence, starting at $1 billion in 2005-06 and rising to $4 billion in 2009-10.

---

¹

In general, the forecasts for a given year were based on private sector surveys completed between December and early February for February budgets. The exceptions were the March 2004 budget, which was based on the March 2004 survey, and the December 2001 budget, which was based on the October 2001 survey and further consultations with forecasters in December 2001. As there was no February 2001 budget, the forecast for 2001 is from the October 2000 Economic Statement and Budget Update (based on the September 2000 survey). [Return]

²

See note 1. [Return]

- Table of Contents - Previous - Next -