When the facts change, I change my mind
Next Monday (29th October), the Chancellor of the Exchequer will present his budget to the House of Commons. He is expected to present spending plans for the coming five years, based on the Office for Budgetary Responsibility (OBR)’s forecasts for GDP growth. As with any projection, these forecasts are inherently uncertain, particularly as a result of substantial lags from preliminary measures of economic activity to the final estimate. For instance, since 1993, there has been only one occasion when GDP figures were not revised a quarter after the first release. Considerable uncertainty also exists as to the exact state and prospects for the UK economy. Figure 1 depicts the uncertainty around the spending plans (in terms of Total Managed Expenditure (TME)) announced at the 2017 budget by constructing upper and lower bounds based on previous (historic) revisions to spending plans. We suggest that as a matter of course, the Treasury should construct error bands around the Chancellor’s expenditure projections, and provide a clearer narrative as to why expenditure might end up being higher or lower than published at the Budget.
New NIESR research (Chadha, Hantzsche, Lazarowicz, Pabst and Young) released yesterday presents results on the size of revisions to published TME plans and the factors that determine these revisions. Figures 2 and 3 plot each spending plan announced in the House of Commons across the period of interest: blue lines denote Conservative led Governments, while red lines denote Labour Governments. As is immediately clear at first inspection, public expenditure plans are not fixed.
We build on this insight by constructing a time series of revisions to TME using NIESR’s own real time macroeconomic database. This series plots the cumulative absolute value of revisions announced by a Chancellor at a Budget or pre-Budget report. We term this series as Fiscal Event revisions. To further illustrate the scale of revisions to any particular plan, the series is plotted in figure 4. This series confirms the insight from plotting the spending plans, namely that spending plans are subject to revision, particularly following economic shocks, and changes of governing party. The former suggests that expenditure responds to the state of the economy and the latter to changing preferences of voters (and parties).
To analyse how TME plans respond to uncertainty about the economy, we assess whether this series of revisions to plans can be explained by changes in the economic outlook. Using standard econometric methods, we find that the most significant factor in determining changes to TME is changes to expectations about the UK economy. Specifically, our analysis suggests that if the Government believes growth will be 1% higher in the coming fiscal year, the planned level of TME will be reduced by 0.9%. For longer horizons, we find that the response can be up to twice as large, which suggests that automatic stabilisers may be in place to respond to changes in the state and prospects of the economy. At this stage we cannot say if the extent of these stabilisers is socially optimal, only that they seem to be in place.
We extend this result by separating the Cameron and May Governments from the rest of the sample, and find on a preliminary basis that since 2010, spending plans have been less counter-cyclical than previously. Equivalently, that the Major, Blair and Brown Governments reacted more to news of the economy when setting fiscal policy than have the Conservative Governments since 2010. We cautiously suggest that this finding may reflect a change of fiscal priorities across the two groups, from an active stabilisation role for fiscal policy prior to 2010, to an overriding aim for debt stabilisation policy in the later period following the financial crisis. More work is required to firm up this explanation.
We also find some evidence that Bank of England policy rate is an important factor in explaining revisions to spending plans. In short, the Bank of England rates tend to be higher when TME is also higher and this indicates some tensions between monetary and fiscal policy. This result suggests the importance of considering the interaction between monetary and fiscal policymakers when any spending or interest rate plan is set.
Throughout the paper we draw on interviews with leading policymakers in the past 25 years about how fiscal expenditure was managed and uncertainty confronted (anonymised summaries of the interviews are available in the annex of the paper). A number of interviewees argued that there is an important difference between two levels of uncertainty: (1) ‘predictable’ uncertainty, which includes ‘normal order’ events (political or economic events such as changes in government or interest rate changes), what we might think of as identifiable risks; and (2) ‘unpredictable’ uncertainty, which relates to unidentifiable risks, for example the implications of events such as the collapse of Long-Term Capital Management in 1998, the bursting of the dot.com bubble in 2001, the 2008 financial crash (Chadha et al., 2016) or the near collapse of the Eurozone in 2010-11.
The second finding concerns fiscal rules. All the interviewees emphasised that fiscal rules matter but that they have a limited duration. One of the main reasons why fiscal rules and their usefulness are limited in time is the trade-off between credibility and flexibility. At some point in the economic and the political cycle, sticking to a set of rules will be at odds with having room for manoeuvre.
Third, fiscal rules tend to reduce uncertainty in the sense of unpredictable behaviour by government, but they can introduce new types of complexity that exacerbate an already uncertain horizon. For example, new rules can lead to even greater departmental under-spend than previous fiscal frameworks, as the sanctions for over-spend may become more severe.
Fourth and in light of the above, what ‘rules of thumb’ did HMT use in order to manage uncertainty? ‘Rules of thumb’ ranged from formal fiscal rules via estimates of Annual Managed Expenditure and forecasts for GDP and tax revenue to pension expenditure, public sector pay and other big ‘fixed costs’ (e.g. monthly meetings of pay boards in the period 1997-2010). Arguably, this process can be described as the ‘rule of big numbers’, which is the result of an asymmetry of information in favour of spending departments. Finally, our research found that the commitment to prudence, which successive governments have invoked, raises questions about the purpose of prudence (economic and/or political considerations) and the ways in which prudent planning of public expenditure translates into economic policy. HMT tends to have built in margins of errors: (i) cutting capital expenditure rather than current spending; (ii) ‘back pockets’ to offset forecast errors. Problems arise when cutting capital expenditure hits national output in times of sluggish economic growth or when new spending pledges drain ‘back pockets’.
This work forms part of a new initiative on policy making under uncertainty at NIESR, and the results of our analysis and the findings from our interviews with key policymakers will be used to guide our future work in the area.The early results from this pilot project are made available but are not final. You can read the full Discussion Paper here.
This project was funded by the Nuffield Foundation, but the views expressed are those of the authors and not necessarily those of the Foundation.