On October 5, 2012, the federal Department of Finance released the audited financial results for 2011-12 in its Annual Financial Report (AFR).  As has been the case since 2006-07, this year’s report contained little meaningful analyses on what impacted on the financial results for 2011-12.  For example, although the report includes a table showing the differences between the final results for 2011-12 and the March 2012 Budget forecast for 2011-12, explanations of these changes amounted to less than four lines and were questionable. In another example, a major reclassification of expenses from Crown corporations to other transfers is explained in a footnote to Table 9 in the Independent Auditor’s section.

Without a detailed explanation of why the final deficit outcome turned out different from the initial budget forecast, it is difficult to hold the government to account for its forecasting errors. Were the differences due to errors in the economic forecasts or inherent bias in forecasting the revenues and expense components? Given the nature of the errors, what impact will they have on the government’s  forecasts for future years? Large and persistent forecast errors, which are not explained, undermine the credibility of the government’s budget plans and stifle public and Parliamentary debate over the main budget choices. 
Fiscal forecasting is more of an art  than a science. It requires considerable professional judgement and is subject to substantial uncertainty and risk. It is, nevertheless, critical to serious policymaking.

The budgetary balance (deficit/surplus) is the difference between two very large numbers (budgetary revenues - $245.2 billion in 2011-12 and total expenses - $271.4 billion in 2011-12), such that a small error in forecasting will result in a large error in the residual - the deficit or surplus. It is because of this that a detailed reconciliation between the initial forecast of the deficit and final outcome of the deficit is required.

In a review of the Government’s forecasting accuracy, Tim O’Neill stated that the most important goal of improving transparency is that full and detailed disclosure should be the norm for all budget-related documents. He also strongly recommended that the Government provide a longer-term perspective on its fiscal forecasting record. In other words, the government should provide reconciliation between what it forecasts in the budget and the final audited budget outcome. A reconciliation was provided in the November 2005 Economic and Fiscal Update, covering the period 1994-95 to 2004-05 but there have been no detailed reconciliations provided since the Conservative Government was elected.

This note provides such a longer-term perspective on the Finance Department’s forecasting record for the period 2006-07 to 2011-12 – the government’s first six budgets.
 

We compare the final audited financial results to the one-year ahead budget forecast (first estimate), usually tabled prior to the start of the fiscal year (April 1) or shortly thereafter. The final audited financial results are provided in the Annual Financial Report usually released about eight months after the end of the fiscal year (March 31), with detailed information tabled in the Public Accounts of Canada shortly thereafter.
 

Table 1 presents the evolution of the fiscal forecasts for the period 2006-07 to 2011-12.  The second column indicates the Budget date for the first estimate of the surplus/deficit, which is provided in Column 3.  The final audited outcome for the budgetary balance is shown in Column 4 and the difference in Column 5.

The differences between the first budgetary balance estimate and the final outcome range from an underestimation of a surplus by $6.0 billion in 2010-11 to an underestimation of a deficit by $21.9 billion in 2009-10. These annual differences are substantially larger than those witnessed during the 1994-95 to 2004-05 period.
 

Table 1: Evolution of Fiscal Forecasts - First Estimate to Final Outcome

1. A negative sign indicates a deterioration in the budgetary balance.

There are two factors that can be easily identified in explaining some of these differences: the impact of the “risk adjustment factor” and “policy initiatives”.
 

A “risk adjustment” factor was included by Minister Flaherty in the January 2009 Budget, recognizing the inherent difficulties in forecasting economic developments and translating these into fiscal projections. It amounted to $4.5 billion in both 2009-10 and 2010-11 and $1.5 billion in 2011-12. This is shown in Column 6 in Table 1.
 

The inclusion of a risk adjustment factor is not new. The Liberal government first introduced them in the 1994 Budget in order to provide greater confidence that the announced budget targets would be met. Subsequently, a risk adjustment factor was included in various forms (economic prudence and a contingency reserve) in every budget until the first Conservative budget in 2006, at which time both the economic prudence and the contingency reserve were eliminated. 
 

In budget presentations, the value of the risk adjustment factor is currently not shown as a separate line item, but is instead “buried” in the component(s) of budgetary revenues  without indicating which components. Based on our analysis, it would appear that since 2009 most, if not all, of the risk adjustment factor was allocated to corporate income tax revenues, and as a result, for the purposes of this note, we have adjusted the forecast of corporate income tax revenues accordingly. In previous blogs, we have pleaded with the Government not to bury the risk adjustment factor, but to show it as a separate line item. This would allow a better assessment of the effective corporate tax rates that the Department of Finance is assuming. To date, this recommendation has fallen on deaf ears.
 

Column 7, entitled “policy initiatives”, provides estimates the fiscal impact of the various policy initiatives introduced during the course of the fiscal year.  Substantial new initiatives were introduced during fiscal years 2006-07 ($4.9 billion), 2007-08 ($7.6 billion) and 2009-10 ($14.5 billion), thereby either reducing the surplus (2006-07 and 2007-08) or increasing the deficit (2009-10). 
 

Column 8 is the residual, after subtracting the impact of the risk adjustment factor and policy initiatives from the difference between the final outcome and first estimate.  For example, in 2009-10, the deficit was underestimated by $21.9 billion (Column 5).  Without the inclusion of the $4.5 billion risk adjustment factor in the first estimate, the underlying first estimate would have been $29.2 billion and the difference between this underlying first estimate deficit and the final outcome would have been $26.4 billion. If no new policy initiatives had been introduced during the course of the year, the difference would have been only $11.9 billion.  As a consequence, the impact of the risk adjustment and the new policy initiatives accounted for about half the difference (Column 5) between the first estimate and final outcome. 
 

The differences in Column 8 are still higher than those witnessed over the 1994-95 to 2004-05 period.  So what accounts for these differences? In the November 2005 Economic and Fiscal Update Annex 1, Risks and Uncertainties in Fiscal Projections, the Department of Finance listed a number of reasons:

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

To these, one could add:
• Unexpected developments during the course of the year (response to natural disasters, valuation adjustments, etc.)
• Delays in implementing new policy initiatives resulting in a higher-than-expected lapse;
• Errors in converting monthly cash receipts into accrual revenues; and
• Differences in concepts and coverage between the Estimates and the Budget forecast for total expenses.
 

Table 2 breaks out the residual errors (column 8 Table 1) into the three major components of the budgetary balances: budgetary revenues, program expenses and public debt charges. The forecast errors for the years 2006-07 to 2009-10 resulted primarily from errors in budgetary revenues. Budgetary revenues were underestimated by $9.1 billion in 2006-07 and by $10.5 billion in 2007-08.  They were overestimated by $10.7 billion in both 2008-09 and 2009-10. 

Table 2: Residual by Major Componen: $ billionst

The forecast errors in program expenses largely explain the difference in both 2010-11 and 2011-12.  Program expenses were overestimated by $8.4 billion in 2010-11 and $8.6 billion in2011-12. The impact of the overestimation of program expenses in 2011-12 was partially offset by lower-than-expected budgetary revenues. In all years under review, public debt charges were overestimated, but not by significant amounts with the possible exception of 2011-12.
 

What could account for these errors in forecasting budgetary revenues and program expenses?
 

Table 3 shows the changes in the average private sector economic forecasts for nominal GDP (the most applicable tax base for budgetary revenues), and for short- and long-term interest rates, from the first estimate of the deficit to the final outcome. Unfortunately, budget forecasts do not provide a breakdown of the various components of nominal GDP, such as wages and salaries, corporate profits, interest income, etc., so it is difficult to properly assess the impact of changes in the economic forecast to changes in the major components of budgetary revenues.  We have recommended in the past that the Government provide these details to enhance the credibility of its economic and fiscal forecasts, but to no avail.
 

Table 3: Evolut ion of Economic Forecasts: First Estimate to Final O utcome

However, using the “sensitivity” analyses presented in the budgets, one can derive a rough estimate of the impact of changes in nominal GDP and interest rates on the major components of the budgetary balance. These are presented in Table 4. 

As expected, economic errors affected budgetary revenues the most, with only a minor impact on program expenses.  However, only in 2009-10 do the errors in the economic forecast account for a significant portion of the error in the residual deficit forecast.  Economic errors account for about one-third of the residual error in 2007-08.  In 2006-07, 2008-09, and 2011-12, the unexplained amount of the residual is actually increased.  There are, therefore, many other factors contributing to the residual error.

Table 4: Impact of Changes in Economic Forecasts onthe Major Components:

                                       Impact of Changes in Economic Forecasts on

Note: A positive sign indicates an improvement in the budgetary balance.

         A negative sign indicates a deterioration in the fiscal balance.

         Total may not add due to rounding.

Within budgetary revenues, there appears to be a systematic bias in the forecasting of personal and corporate income tax revenues and in other revenues.  The former two could be related to the accrual adjustment factors used to convert monthly cash receipts into accrual estimates.  It is only after the close of the fiscal year that the audited accrual numbers become available. Other revenues consists of profits from enterprise Crown corporations, revenues from consolidated Crown corporations, Exchange Fund losses/gains, investment income, Bank of Canada profits, sales of goods and services, etc. Some of these components can be very volatile and difficult to forecast.
 

Within program expenses, the largest consistent source of error is in direct program expenses. In our assessment of previous budget fiscal forecasts, we have argued that the Department of Finance has consistently overstated this component in the budget.  Part of the reason, we believe, is the disconnect between the budget estimate of spending and that included in the Estimates prepared by the Treasury Board Secretariat. Until there is a detailed reconciliation between these two sources of government spending, there will continue to be a bias in the budget estimates.
 

Public debt charges are affected by changes in interest rates, changes in the stock of interest-bearing debt and changes in the composition of that stock of interest-bearing debt.  Some of these errors can be explained by forecast errors in the average private sector interest rates. However, higher-than-expected financing requirements and changes in the composition of debt were also likely significant contributors.
 

Much still needs to be explained in analyzing the long-term fiscal track.  We would encourage the Minister of Finance to provide more detailed explanations on the reasons for the forecast errors.  In doing so, more details are required with respect to the economic forecast.       .

------------------------

1. Update to incorporate Public Accounts data for 2011-12.