In Defense of Forecasting: Its Importance in the Budget Process

Boule de crystal
Posted by Tom Josephs

Recent research by IMF economists Hites Ahir and Prakash Loungani shows that forecasters comprehensively failed to predict the recessions that followed the 2008 financial crisis.2 This was the case even for forecasts made in early 2008 — after the collapse of Bear Stearns bank and with the crisis well established. This prompted the economist Tim Harford to write recently in the Financial Times that: “The obvious conclusion is that forecasts should not be taken seriously….It is still a source of constant wonder to me that the demand for forecasts – in economics and elsewhere – remains undiminished.”3

Nevertheless, around the world the production of an economic and fiscal forecast is typically one of the first steps in the annual budget preparation process. Governments in most advanced economies present forecasts over a three- to five-year horizon alongside their budgets. A number of countries also produce projections of longer-term scenarios for periods of up to fifty years into the future.

All budget policy decisions have long-lasting effects. A public sector capital project might take several years to deliver and will generate ongoing expenditure for many years beyond that. More generally all public spending is likely to be more efficient if it is planned ahead. Expenditure plans need to be based on an expectation of how much government revenue will be raised in the future, which in turn is determined by expectations about the economic activities that are taxed. When the government borrows to fund expenditure its lenders will seek assurance that it has a credible plan for remaining solvent and that over time these debts will be repaid.

In short, governments will always need to consider the future implications of both their detailed tax and expenditure policies and their overall fiscal strategy. They will need to form a view on the likely path of the economy and the public finances in the absence of policy changes, and then consider how their proposed policies will affect that path. A forecast provides the necessary analytical framework against which these policy choices can be made and evaluated.

Forecasters should strive to learn from past mistakes and improve the accuracy of their predictions. However, no one possesses a crystal ball so policymakers would be unwise to make their plans on the expectation that forecasts will turn out to be exactly correct. Ahir and Loungani point out that forecasts have been particularly bad at anticipating major economic turning points. But even in normal times it is the case that budget forecasts, which need to cover the range of government expenditures and revenues, are highly unlikely ever to be fully accurate.

Against this background, what should policymakers expect from a budget forecast? Four important characteristics can be highlighted:

• A forecast should be based on a comprehensive analysis of available evidence on the current position of the economy and public finances and past historical trends. Indeed one of the most useful contributions that a forecast can make is to enhance understanding of the current state of the world, which is vital for good policymaking. Such analysis could be provided without producing a forecast. But being required to consider how things may change is an excellent discipline for forcing a detailed understanding of the present and past. This understanding typically relies on national accounts statistics which in low capacity countries can often be poor quality. In such cases efforts to improve forecasting need to go hand-in-hand with developing the quality of national statistics.

• On this foundation a forecast should provide a realistic central view of the possible future path of the economy and public finances, formed on the basis of professional judgment and analysis of past experience. The important point here is that the forecast should be a central unbiased view. This is because, faced with an uncertain future, the best basis on which to plan is a central forecast around which the risks are evenly balanced. This should allow policymakers to make plans that have resilience in the face of the upside and downside surprises that will always come along. While forecast errors are inevitable, if they are unbiased they should even out over time, meaning spending plans based on those forecasts should be sustainable in the medium term, even if borrowing has to fluctuate from year-to-year as surprises hit.

• This underlines that a forecast’s value is much enhanced if it is accompanied by a wider analysis of these risks. There are a number of techniques that can be used to do this. Forecasters can produce sensitivity analysis of the impact on the forecast of changing key assumptions, they can produce full alternative forecast scenarios, and/or they can create probabilistic distributions of outcomes around the central forecast. This will aid policymakers’ understanding of the range of possible outcomes they face.

• Forecasters should regularly look back and analyze why outcomes differed from their expectations, and then transparently publish their findings. This will build understanding of the economy and the public finances, and also potentially help improve future forecasts. Without a baseline forecast as a point of comparison it is more difficult for policymakers and the public to know whether to be surprised by actual developments and what the appropriate response should be. Determining whether unexpectedly high inflation was caused by an external or domestic surprise, or whether unexpectedly high borrowing is due to structural or cyclical factors are vital to deciding how policy should respond.

All this information should also be presented to the public, as an important role of a forecast is the external transparency and accountability which it brings to policy-making. It allows the public to judge policy choices based on an understanding of the government’s view of the future and of the impact of its policies. Those who disagree can present alternative views and compare them to the government’s. It is difficult to imagine how a constructive debate on policy could take place in the absence of the framework provided by a forecast and its accompanying analysis.

In recent years a number of countries have established independent fiscal councils tasked, amongst other things, with producing forecasts or publically scrutinizing the forecasts of the government. The hope is that this will help ensure budget forecasts are not influenced by political factors and will increase their transparency. It has been well documented that many governments have in the past based their budget plans on forecasts which proved to be serially over-optimistic. As a result deficits and debt consistently turned out to be larger than desired. There is no technical reason why fiscal councils should necessarily produce more accurate forecasts than governments — they have no informational advantage and still no crystal ball. But if well-designed they should not be tempted to produce biased forecasts for political reasons. In addition the all important forecast transparency can be easier for an independent institution, because it should not worry about potential political sensitivities around its underlying assumptions and its risk analysis. However, fiscal councils will not be right for all countries. In particular they need sufficient staff with technical expertise in the analysis of the public finances which may be in short supply in lower capacity countries.

To summarize, a forecast is central to a credible budget given the long-lasting effect of budget policy decisions and the need to plan ahead. But a forecast should be much more than simply a series of potential future outcomes for key variables. It should provide the policymaker with a deep analysis of the current state of the world, an unbiased view on its potential future path (with and without the impact of new policies), and an analysis of the full range of risks that could lead or have led outcomes to differ from the central view. All this information should be made available transparently to the public to allow for an informed debate on the government’s budget policy choices.

1 The author is grateful for comments from colleagues in FAD and from Andy King at the UK’s Office for Budget Responsibility.

2 Hites Ahir and Prakash Loungani, Fail Again? Fail Better? Forecasts by Economists during the Great Recession. Available at:

3 An astonishing record – of complete failure. Financial Times, May 30 2014. Available at:

 Note: The posts on the IMF PFM Blog should not be reported as representing the views of the IMF. The views expressed are those of the authors and do not necessarily represent those of the IMF or IMF policy.