Economic Forecasting

This chapter provides an overview of the main themes covered in the book. It revolves around three basic questions:

Measuring and interpreting economic developments requires a set of statistical standards as well as a coherent overall framework. By and large, forecasters around the world use a common language and refer to the same key macroeconomic variables. However, there are differences across time and space regarding the specific empirical content of some of the concepts used. For most practical purposes, at least in forecasting, these differences are relatively minor, but they ought to be borne in mind when laying side by side national forecasts or comparing performance across countries.

The first step in economic forecasting is to establish where the economy actually is. This constitutes a prerequisite for a proper forecast of where it is heading (Section 3.1). The starting point is therefore the collection of a great variety of economic information with a view to assessing the state of the economy before national accounts data become available, weeks or months down the road (Section 3.2). In this regard, survey data can provide precious insights, almost in ‘real time’, and usefully complement the ‘hard data’ due to be released later on (Section 3.3). Once reasonably comprehensive information has been gathered, the challenge is to make sense of it, by adjusting the raw data in various ways and combining some of them to build summary indicators (Section 3.4). In this context, bridge models are especially useful for near-term forecasting purposes (Section 3.5). Even so, macroeconomic monitoring involves difficult trade-offs between abundant but often seemingly inconsistent bits of information (Section 3.6).

A macroeconomic (or macroeconometric) model is a quantitative representation of an economy, or of several interdependent countries. It assembles a number of equations and allows study of the behaviour of the economy(ies) when all the various relationships between variables are operative simultaneously. In addition, a model synthesises data and knowledge with a view to explain economic history better and to forecast future developments.

Interest in medium- and long-run projections is on the rise, not least due to concerns about the economic implications of population ageing, but also against the background of the ongoing rebalancing of the global economy towards some large emerging economies, especially in Asia. Medium- and long-run projections differ from the short-run forecasts discussed in previous chapters. Beyond the next few quarters, it is indeed pointless to try and forecast the business cycle. However, it is possible and instructive to look at the underlying trends further out. To differentiate the two types of exercises, the term ‘projection’ tends to replace that of ‘forecasts’ for longer-run horizons. Even so, projections are and should be quantified rather than purely qualitative. In this context, the numbers are not meant to convey what would be an illusory sense of precision. Rather, they serve to illustrate broad trends and to highlight constraints, notably through recourse to alternative scenarios.

Financial and commodity prices are not forecasted in the same way as national accounts aggregates. These prices are volatile and very much driven by market participants’ expectations. A great variety of tools are used to analyse and forecast them, ranging from traditional models of interest rate determination to ‘heterodox’ methods such as chartism, to country-risk analysis.

Budget forecasts try to anticipate the evolution of the fiscal accounts based on an economic scenario. They are therefore one specific aspect of overall economic forecasting. They range from the rather global to the very detailed, depending on the purpose of the exercise (Section 9.1). Over the short and medium run, budget forecasting mainly involves assessing the sensitivity of receipts and spending to changes in macroeconomic conditions and estimating the impact of new measures (Section 9.2). Over the long run, the focus is more on the evolution of public spending and on the underlying demographic trends (Section 9.3). Budget scenarios can be put together in several ways (Section 9.4), following a fairly general top-down approach or a more detailed bottom-up one (Section 9.5). In practice, these two approaches complement each other. Budget forecasts come with substantial error margins, which is a cause for concern given the large size of the public sector in the overall economy (Section 9.6). One way to analyse and prevent fiscal surprises is to look at the budget from a more analytical angle, bringing in the concepts of structural balance, debt sustainability and rules (Section 9.7).

This book mostly deals with forecasting developments at the macroeconomic level, even when discussing specific aspects such as financial markets (Chapter 8) or fiscal policy (Chapter 9). In contrast, sectoral forecasts follow a microeconomic approach. Here, the focus is on sectors, or even on one particular sector. This is a relevant perspective for a firm concerned with the outlook its sector faces, or for a local government wondering how much needs to be spent on infrastructure to ensure that firms in the area can grow without encountering local bottlenecks (such as road congestion).

Though demand for economic forecasts is strong, their reliability is frequently questioned and scepticism is widespread among the general public regarding the accuracy of any forecast. Granted, forecasters often get it wrong, not least when it comes to foreseeing turning points or crises. But uncertainty is part and parcel of advanced economies. In addition, forecasters themselves consider that they tend to be unduly criticised. They argue that their track record is more respectable than is often alleged, noting that professional forecasts have been proved to be more reliant than the so-called ‘naïve forecasts’ produced by elementary methods. Moreover, accuracy is perhaps not the best criterion to judge the value of forecasts. Indeed, their usefulness may have more to do with the associated diagnoses and policy implications (see Chapter 12).

Demand for forecasts obviously arises from a need to form a view on the future.¹ A government, for example, when preparing next year’s budget, has to rely on a forecast of activity in order to quantify the foreseeable tax receipts. Social partners have to refer to some forecast of inflation when negotiating wage increases. Firms contemplating investment in new factories try to anticipate demand for the corresponding output. In fact, in virtually all walks of economic life, agents regularly use forecasts as inputs into their decisions. That said, forecast accuracy is far from perfect, as discussed in Chapter 11, and preferences and constraints along with forecasts matter in framing decisions.

Forecasts need to be explained both to policymakers and to other clients, but also to a wider public. This raises communication issues. First, some of the technical subtleties are difficult to explain in a simple way, not least as regards the uncertainties surrounding the forecast (Section 13.1). Second, there is the question of how much transparency is desirable: to what extent should governments and central banks publish their forecasts or keep them confidential (Section 13.2)? Lastly, in the case of official forecasts, it is important to acknowledge their ambivalent status: they are both a technical and a political exercise, and this raises tensions that need to be addressed (Section 13.3).

Myriad public, semi-public and private bodies are engaged in regular or occasional forecasting. Users are also very diverse. On the supply side of the market for forecasts, some producers are driven by profit motives, while others offer a public good. The demand side includes policymakers – especially finance ministers and central bank governors – but also a broader public, say, the readers of the financial press. The product itself is far from homogeneous: many forecasts cover only the short run, but some extend over longer horizons, and quality is very uneven. It is also a market with fads, as some forecasters’ reputation rides high when they surf on a run of successful predictions, but can suddenly collapse when they fail to foresee some major turning point. Lastly, price-setting is rather opaque in this market, since cross-subsidisation is rife, including in the private sector.

A British Chancellor of the Exchequer who suffered more than others from the major forecasting errors made under his watch decided to take revenge on the technicians and declared in his Budget statement:

Like long-term weather forecasts, they [economic forecasts] are better than nothing …. but their origin lies in the extrapolation from a partially known past, through an unknown present, to an unknowable future according to theories about the causal relationships between certain economic variables which are hotly disputed by academic economists and may in fact change from country to country or from decade to decade (Healy, 1990).