Chancellors don't respond symmetrically to good and bad economic news. This represents a risk to the accuracy of official borrowing forecasts.
Governments make policy choices on the basis of forecasts. This is particularly true of fiscal policy, where governments often target borrowing or debt at some future date. These forecasts are subject to considerable uncertainty, however, and are subject to frequent revisions - especially when the economy gets buffeted by shocks.
This is illustrated in Figure 5.1, which shows the range of forecasts for public sector net borrowing (PSNB) made for each year since 1982-83. In periods of unexpected economic turmoil, the range of forecasts is particularly wide. For the 2009-10 financial year, for example, the level of forecast borrowing ranged from 1.5% of GDP (in the March 2005 Budget) to 12.6% of GDP (in the December 2009 Pre-Budget Report); in the event, borrowing amounted to 10.2% of GDP. Even in less turbulent times, forecasts can still be revised by a per cent or more of GDP: the average forecast error over the past 40 years was 1.8% of GDP.1
Figure 5.1. Public sector net borrowing forecasts and out-turn since 1982
Note: Each line represents the range between the highest and lowest forecast for PSNB. From 1982-83, there are at least five forecasts for each fiscal year. Forecasts produced prior to 2010 were made by the Treasury. Out-turn data are the latest available data published in March 2023.
Source: Office for Budget Responsibility, ‘Historical official forecasts database', https://obr.uk/data/; authors' calculations.
Figure 5.1 also shows that borrowing tends to come in towards the top end of the forecast range. In other words, there is a tendency for most borrowing forecasts to be overly optimistic. Since 1982-83, borrowing has turned out higher than the median forecast on three-quarters of occasions. The early 1990s, the 2000s and the 2010s stand out as periods during which forecasts were particularly optimistic. The tendency for forecasts made during the 2010s to underestimate borrowing, and subsequently to be revised upwards, is illustrated in Figure 5.2.2
Figure 5.2. Successive Office for Budget Responsibility borrowing forecasts since 2009-10 and the subsequent out-turn
Source: Chart 1.2 of Atkins and Lanskey (2023).
As Chancellors prepare ahead of each fiscal event, they are provided with a new set of forecasts, which contain information about how the outlook has changed since the last fiscal event. These changes can be thought of representing ‘good' or ‘bad' economic news. Chancellors often adjust their tax and spending plans in response to this news. For example, if the economic and fiscal outlook improves, it could be that the Chancellor is able to lower taxes and/or increase spending and still be on track to meet his or her stated objectives for borrowing or debt. Conversely, if an adverse event occurs and the outlook deteriorates, the Chancellor might choose to raise taxes and/or cut spending to get (the forecast level of) borrowing back towards his or her desired level.
It matters whether or not Chancellors respond symmetrically to good and bad news. If Chancellors respond asymmetrically to underlying changes in borrowing forecasts - for example, by spending windfall gains in the case of good news, but allowing borrowing to increase when bad news comes along - then over time, borrowing will systematically diverge from the forecast.3 This represents a non-trivial risk to the accuracy of OBR borrowing forecasts, and potentially to fiscal sustainability.
A particularly blatant example of this pattern of asymmetric behaviour - previously highlighted in the 2018 IFS Green Budget - came during Philip Hammond's period as Chancellor. His statements indicated that he would view forecast improvements and deteriorations rather differently.
In the Autumn 2017 Budget, he cited a ‘balanced approach' when responding to a deterioration in the forecast:
‘I reaffirm our pledge of fiscal responsibility and our commitment to the fiscal rules I set out last Autumn. But now I choose to use some of the headroom I established then. So that as well as reducing debt, we can also invest in Britain's future. Support our key public services. Keep taxes low. And provide a little help to families and businesses under pressure.'
Philip Hammond's Autumn Budget speech, November 2017
That is, he said that he would allow borrowing to rise following a forecast deterioration (‘use some of the headroom'). Then, in the following Spring Statement (2018), he said:
‘And if, in the Autumn, the public finances continue to reflect the improvements that today's report hints at, then, in accordance with our balanced approach, and using the flexibility provided by the fiscal rules, I would have capacity to enable further increases in public spending and investment in the years ahead.'
Philip Hammond's Spring Statement speech, March 2018
In other words, he indicated that if the outlook for the public finances improved, he would be minded to spend any such improvement. So, in one instance, the Chancellor is saying that he will allow borrowing to increase following a forecast deterioration, rather than offset the increase through policy measures. Yet in the subsequent instance, he is promising to spend the windfall should the public finances improve. In this chapter, we examine how successive Conservative Chancellors have reacted to underlying changes in public sector net borrowing forecasts since 2010.
Key findings
- Forecasts for government borrowing are uncertain and subject to frequent revision. Over the last four decades, borrowing has turned out higher than the median forecast for that year on three-quarters of occasions. In other words, forecasts have tended to underestimate the future level of borrowing. This was particularly the case just prior to periods of economic distress, such as the early 1990s, late 2000s and the pandemic, but is true generally.
- Forecast revisions often reflect economic ‘news' since the previous forecast and unfortunately, since 2010, there has been more bad news than good. Across the 26 fiscal events since 2010, there have been just 8 occasions on which economic news has meaningfully improved the borrowing outlook, versus 12 where bad news has materially worsened the outlook. On 6 occasions, there was no meaningful change.
- Chancellors often adjust their tax and spending plans in response to these forecast changes. For example, if the economic and fiscal outlook improves, it could be that the Chancellor is able to lower taxes and/or increase spending and still be on track to meet his or her stated objectives for borrowing or debt. Conversely, if the outlook deteriorates, the Chancellor might decide to raise taxes and/or cut spending to return forecast borrowing back towards the desired level.
- It matters whether or not Chancellors respond symmetrically to good and bad news. If Chancellors respond asymmetrically to underlying changes in borrowing forecasts - for example, by spending windfall gains in the case of good news, but accommodating increased borrowing when bad news comes along - then over time, borrowing will systematically diverge from that forecast. This represents a non-trivial risk to the accuracy of borrowing forecasts, and potentially to fiscal sustainability.
- Chancellors have not responded symmetrically to good and bad economic news since 2010. On the 12 occasions when economic conditions deteriorated meaningfully between fiscal events, Chancellors have planned to offset just over a quarter (27%) of the medium-term borrowing increase, on average, by reducing the planned level of spending and/or announcing tax rises for implementation by the final year of the forecast period. Meanwhile, when economic conditions improved, Chancellors have planned to offset an average of 60% of the windfall through higher spending and/or lower taxes.
- This tendency for Chancellors to loosen more than they tighten in response to economic news led to tens, and possibly hundreds, of billions of additional borrowing over the 2010s. Public sector net debt at the eve of the pandemic could have been between 3% and 11% of GDP lower - with a central estimate of 7% - had Chancellors responded symmetrically to underlying forecast changes over the preceding decade.
- Asymmetric policy responses mean that the Office for Budget Responsibility (OBR)'s central forecast is not actually ‘central'. Based on Chancellors' past responses to shocks, and assuming good shocks are as likely to come along as bad ones, we estimate that forecastgovernment borrowing in 2027-28 should be 1.4% of GDP higher than under the OBR's central forecast. In 100,000 simulations of future shocks and subsequent policy responses, we estimate that there is just a one-in-ten chance that borrowing turns out lower than the OBR forecast. This is symptomatic of a wider issue facing the OBR: the requirement to take government policy as stated, rather than exercise its judgement based on past government behaviour, can make it more likely that the forecast underestimates borrowing.
- When economic conditions improve, Chancellors since 2010 have tended to top up their spending plans, rather than use any ‘windfall' to cut taxes. When conditions worsen, they have tended to cut back their spending plans and raise taxes. Combined, this tendency has acted to increase the size of the state over time. We estimate that if future Chancellors respond to economic news in the same way as their predecessors, the forecast for total government spending should be 1.6% of GDP higher in 2027-28 than under the OBR's central forecast. In contrast, forecast government revenues would be just 0.2% of GDP higher in our simulations.
- In the short term, Chancellors tend to announce a policy loosening (i.e. higher spending and/or lower taxes) regardless of whether there has been an underlying economic improvement or deterioration. This may be appropriate - depending on the nature of the economic news - but risks a further ‘ratcheting' effect if short-term loosenings are implemented but medium-term tightenings are ever-postponed.
Endnotes
- 1.
This refers to the absolute mean forecast error (which was 5.6% of GDP for 2009-10, the example year given in the text). The average range (between the lowest and highest forecasts for a financial year) was 4.3% of GDP (and 11.1% of GDP in 2009-10).
- 2.
The Office for Budget Responsibility (OBR) recently concluded that its tendency to underestimate government borrowing largely reflects underestimates of the future level of government spending by departments (Atkins and Lanskey, 2023).
- 3.
The direction of the divergence will depend on the direction of the asymmetry.