In this little exercise I have gone through commonwealth government budgets from 1999‒2000 to 2013‒14 to study changes in the figures.
The figures for each budget can vary quite significantly. For quite a few years now we have had the Mid-Year Economic and Fiscal Outlook (MYEFO) which updates and revises the figures used in the budget as presented on budget night. Even before MYEFO, it was not uncommon for budget figures to be revised during the year and this often became apparent during the process known as ‘Additional Estimates’. Then there is the next budget: while it provides the figures for the coming financial year, it also provides another set of figures for the current financial year — which still has about seven weeks to run at the time of the budget. And each budget also contains projections for the next three years (the forward estimates). To complete the picture, in about September each year, the ‘Final Outcome’ is released which presents the actual figures, the actual revenue and spending, that occurred in the previous financial year (and it compares those final figures not to the original budget but the figures used on the more recent budget night). So, before the Final Outcome, there could have been at least six different sets of figures relating to any single budget.
I will focus on two sets of figures: the budget figures as presented by the treasurer on budget night and the final outcome (which is why I stop at 2013‒14). That will give us an indication of how accurate (or otherwise) the treasury and treasurer estimates of income and spending actually are.
There is also an issue about accrual accounting and cash accounting, and there often seems to be a few billion dollars difference between the two: for example, in 2007‒08 the ‘final outcome’ revenue for the government in accrual terms was $303 billion but $295 billion in cash receipts. (One reason I have commenced in 1999‒2000 is because that was the year that accrual accounting was introduced.)
Accrual accounting assigns income and expenditure when a commitment is made whereas cash accounting only enters money when it is actually received or spent: for example, if the government promises an organisation $100,000 on 10 June this year but doesn’t pay the money until 10 July, it would appear in the 2014‒15 budget under accrual accounting but the 2015‒16 budget under cash accounting. While the budget papers do contain both cash and accrual tables, I have stuck with the figures from accrual accounting as they now seem to provide the headline figures. (Although from my reading it appears that governments may sometimes pick the figure which best suits them at the time.)
In the eight years from 1999‒2000 to 2006‒07, actual revenue exceeded the budget estimate in every year by an average of $6.7 billion (or about $9 billion in constant 2013‒14 prices), varying from $2.8 billion to $13.1 billion, or from 101.5% to 106.8% of the budget night estimate. We might argue that a difference of 1‒2% isn’t too bad when one is trying to predict the future but it was only that close in two of those eight years (and it has been that close only once since). Also, because in those years the mistake was to under-estimate revenue, it wasn’t a problem for the government. (It also helps explain how Costello managed to achieve his surpluses — the cumulative under-estimates for those years amounted to $54 billion.)
The year 2007‒08 was a turning point but initially for a different reason. It was the year the GST was counted in the government’s revenue — before that it was discussed in a separate budget attachment. For that reason, government revenue jumped by almost $57 billion above the original budget night estimate but $44.4 billion of that was the GST revenue. Without the GST, the final outcome for revenue was still $12.6 billion above the budget night estimate. The influence of the GST on the budget can basically be ignored because although it adds to revenue it is simply offset in spending as a transfer to the states. It does mean, however, that the headline revenue figure in the budget is a little misleading: for example, in 2013‒14 the government’s final revenue was $374.0 billion but that included $55.5 billion in GST, which left $318.4 billion for spending on commonwealth government programs (including additional support to the states for hospitals and schools). Despite that I will use the full budget figure as it becomes too time-consuming to re-arrange every budget to remove the GST.
From 2008‒09, the budget night estimates have tended to over-estimate government revenue, with only 2009‒10 under-estimating revenue (by $2.2 billion).
||Budget estimate (billion)
||Final outcome (billion)
In dollars terms that amounts to an average ‒$17.7 billion each year. The cumulative total of budget night forecasts for those six years was $2,080 billion but the actual revenue was only $1,974 billion (95%), meaning the Rudd and Gillard governments had available for spending $106 billion less than they were originally told. That also suggests that when treasury over-estimates revenue it does so by more than twice as much as when it under-estimates revenue.
The forecasts in the ‘forward estimates’ in each budget also provide a picture of how treasury thinking changes over time. (The figures are billions of dollars on an accrual basis and start at 2001‒02 because I would need to go back earlier than 1999‒2000 to get the full three-year forecasts for previous years and those earlier years are cash accounting).
||Forecast 3 years earlier
||Forecast 2 years earlier
||Forecast 1 year earlier
* First includes the GST, which was not included in revenue figures in previous years
If you wonder why Labor had problems in its last budget, check out that last line of figures. The final outcome was below all of the previous forecasts, the only time that has happened since 2000 (although it also appears likely to happen to Hockey’s 2014‒15 budget but we won’t know for certain until September). The previous two years were above only one forecast, the one made two or three years earlier. In contrast, the Howard government during the 2000s saw the final outcome above all previous forecasts.
You can also see that treasury became a little uncertain between 2001 and 2004 after the ‘tech bubble’ burst and adjusted forecasts up and down and that has happened again in recent years, although for 2013‒14, and now for 2014‒15, each subsequent forecast has been lower: for 2014‒15 the figures run:
The forecast three years out tends to be lower than the final outcome: that was the case in 11 of those 13 years. Leaving aside the change that the inclusion of the GST caused for the forward estimate for 2011-12, the only year in which that forcast three years out has been higher was in 2013‒14. That suggests the obvious: that revenue normally grows and it is difficult to predict by how much it will grow (which can also be influenced by inflation as these are ‘nominal’ dollars not adjusted to ‘constant’ or ‘real’ dollars). If you check the above table, you will see that revenue grew in all but two years (2008‒09 and 2009‒10) and, even with the problems since, has continued to grow although not as much as forecast.
- $425.8 billion forecast in 2011‒12
- $424.8 billion forecast in 2012‒13
- $411.6 billion forecast in 2013‒14
- $391.3 billion estimated on budget night (which was reduced to $385.9 billion in the MYEFO)
Individuals’ income tax and company tax are the two biggest sources of revenue for the commonwealth government. The final outcome for individual income tax revenue was around 102% to 104% of the budget night estimate from 1999‒2000 to 2006‒07, with the exception of 2002‒03 when it fell to 98.1%. The year 2007‒08 was a good one for the government with individual income tax coming in at 105.5% of the estimate but since then the budget night estimates have been very good to poor: for four years (2008‒09 to 2011‒12) the individual tax income was between 99.3% and 100.4% of the budget night estimate, but in 2012‒13 it fell to 98.5% and in 2013‒14 to only 95.8%.
Company tax is more volatile and the forecasts rarely get it right. In the nine years from 1999‒00 to 2007‒08, the company tax came in at 98‒102% of the budget night estimate on only three occasions but, otherwise, it was generally under-estimated (coming in at 117.5% of the estimate in 2002‒03). Since the GFC, however, the estimates have failed to foresee the slowness of the recovery. In 2007‒08 it had come in $210 million above the budget night estimate but collapsed to 82.6% of the estimate in 2008‒09 (‒$12.8 billion). The estimate was better in 2009‒10, although actual revenue was lower — only $53 billion compared to $61 billion in 2008‒09. In the last four years (2010‒11 to 2013‒14), however, the estimate has been wrong every year and has over-estimated company tax by a cumulative $29.1 billion. It has provided as much as 25% of government revenue but in 2013‒14 was down to 18.4% (since 1999 it has averaged just over 19%).
The Howard government claimed to be low taxing but it achieved the highest revenue as a proportion of GDP in 1999‒2000 when it took 26.4% of GDP from the economy (it’s budget night estimate had been a high 25.9%). It did try to lower its GDP take in the following years aiming for 22.5% to 23% of GDP but generally its final take continued to be a little higher (most often just over 23%). Based on the budget night estimates, Labor also aimed for about a 23% take of GDP (other than during the GFC when it estimated 25.5% but actually took 24.9%). Since then the government’s take has been lower, and lower than the budget night estimates, falling to 21.7% in 2010‒11 and even in 2013‒14 was 22.7% (after predicting 23.5% on budget night). The Hockey budget of 2014‒15 estimates a take of 23.6% of GDP. It does seem odd that in the last few years as revenue growth slowed it also became a lesser proportion of GDP: one would think that it would remain more consistent and that revenue would fall if GDP fell or grow at a similar rate to GDP but that doesn’t appear to be the case. (If anyone can explain this, please do in a comment.)
Now I turn to spending forecasts. They could be more difficult as they are subject to government decisions that are taken after budget night and in that sense treasury can’t always be blamed if they are wrong. For example, the Rudd government decisions during the GFC boosted ‘social security’ spending by $22.1 billion above the budget night estimate and education by $3.8 billion. That would not have been the problem it became if the original revenue forecast had been accurate. Even with the increased spending, the total actual expenditure was only $5.1 billion above the revenue estimate on budget night: it was the fact that revenue came in $20.5 billion below that estimate that turned what should have been about $5 billion added to the deficit to over $25 billion. (Although, obviously, that additional expenditure was made because it was becoming apparent that the economy was in danger and revenue was falling.)
Health and social security are two of the government’s bigger expenditure items. While spending has continued to increase over the years (the reasons for that are a separate issue), the estimates of spending have been relatively good. Health, in particular, achieved 98‒102% accuracy in 11 of the 15 years from 1999‒00 to 2013‒14 and social security still managed to achieve 10 years despite the extra spending in 2008‒09. That makes sense because the spending is actually estimated by each line department, not treasury, and they basically know what they will spend under prevailing policies — major changes generally occur only when policy is changed during the year. In this context, I am taking no account of whether or not a government is spending more than it receives in revenue, only that the forecasts of spending are reasonably accurate.
Estimates and forecasts of revenue are important. Governments base their policy and spending decisions on them. When governments introduce new spending proposals, they do so on the basis of forecasts over the forward estimates — they usually allow that the future costs can be met within those revenue forecasts. But when the forecasts are wrong, and over-estimate future revenue, everything goes wrong: the government is locked into expenditure that no longer fits under the available revenue and that leads to deficits. That is a reason that deficits are sometimes considered acceptable when revenue growth slows, as at present — as long as government can foresee better revenue growth in future years.
While I have suggested that a +/‒2% estimate isn’t too bad as an estimate, it can still make a difference to the government in dollar terms. Even with the smaller budgets back in 1999‒2000 a 2% variation in total revenue would have amounted to $3.3 billion and in 2013‒14 would have been $7.8 billion (about three-quarters of the amount raised by the Medicare levy, so it is significant). So getting revenue forecasts and estimates right is important — but it never happens. It demonstrates how little control the government actually has over the economy. The economy moves to its own pace, influenced by many factors, including overseas events, and even the ‘expert’ treasury officials can’t predict it accurately, not even one year ahead.
It is not a problem when the economy is going well because then the errors tend to be under-estimates and every government would be pleased to see more in its coffers than it anticipated. But when the economy is in the doldrums, as now, faulty estimates and forecasts put additional pressure on government programs that are meant to provide services to all of us.
I do wish treasury could get it right but, given the economic reality within which it operates, I won’t hold my breath. And I don’t think this week’s budget will be any different.
What do you think?
After reading Ken’s piece, do you still think we should always blame the government for the financial mess or should treasury share the blame for giving the government faulty advice? Should governments only spend 95% of what treasury tells them they have so as to allow for treasury’s mistakes? We look forward to your comments — and answers.
Next week 2353 looks at a different approach to homelessness in ‘Hope for the homeless’.