Article - Errors in Growth Forecasts
International and national institutions, banks, governments and others provide growth forecasts, whose publication or announcement is impatiently expected and carefully analyzed. On June 2013, the World Bank forecast a 2,2% world growth for 2013 while IMF forecast it on October 2013 to reach 2,9% the same year, that is 2 pp. lower than its forecast two years earlier. Forecasts are often revised, upwards or downwards. But do forecasts really matter and why? How to explain that forecasts vary according to the different sources? Does the difference between forecasts and reality have economic consequences? Those questions will be addressed in the following article.
Forecasts do matter. They play a role in the implementation of monetary and fiscal policies. They also give a trend, and used to identify economic events.
Forecasts play a key role in the conduct of macroeconomic policy. A central bank chooses the interest rate to implement thanks to the Taylor Rule, according to the gap between inflation, the inflation target and the gap between real GDP and potential GDP.[1] However real GDP data are not available instantaneously, figures are published with delays, and potential GDP is estimated. Monetary authorities therefore rely on forecasting models—and, with an upward distortion, the interest rate implemented will be higher, leading to tighter credit conditions and and weaker growth. Forecasts also drive the conduct of fiscal policy, helping determine ex ante the stance of the budget.
Because forecasts are essential tools to predict economic turning points, adequate ones are paramount to prevent policy mistakes—such as overly long expansionary stances or too early exits. Discussions around the Fed’s tapering are a case in point. Hence the importance of knowing where the economy stands.
Now that it’s clear from the first part that forecasts matter and why they matter, let’s see whether forecasts are accurate or differ from actual growth.
Looking at actual examples, forecast errors are prevalent. The IMF forecast for pre-crisis Greece is illustrative.
The above graph [to see the graph, download the pdf version of the article] not only shows differences between forecasted and actual growth. It also stressed that there was no rule for mistakes: whereas in 2003 actual growth was 1 to 2 points higher than forecast, in 2005 it was 1 point lower. Forecasts errors are not specific to the 2008 crisis—and a pattern seems apparent of smoothing out real variations.
Where do forecasting errors come from? First from lags in data releases and data revisions.[2] The delay between real-time data and their first release imply forecasters to have to use estimated data in order to make their growth forecasts. This a technical, unavoidable component, in forecasting errors. Second, biases can explain why forecasts differ from effective growth and from one source to another, with typical patterns: governments tend to overestimate growth forecasts, to set lax budgets; institutional biases may also be at play, for example to support solvency in lending in IMF programs through overestimating future growth. In its annual report on the World Social Situation of 2011, the United Nations assessed that fiscal consolidation and austerity measures implemented in countries like Greece or Spain make the recovery more fragile and uncertain. They should instead be able to implement countercyclical policies. The UN thus called for « changing the fundamental orientation and nature of policy prescriptions that international organizations impose on countries as conditions for assistance ».[3] Third, macroeconomic tools may be simply wrong: suffice to recall the miscalculated fiscal multipliers during the eurozone crisis.[4] Blanchard and Leigh (2012) acknowledged that “for every additional percentage point of GDP of fiscal consolidation, GDP was about 1 percent lower than forecast”. Non-linearity is particularly difficult to take into account, and so are financial variables. Finally, forecasting errors may be used as signaling devices, or confidence builders—weak forecasts can lead to higher risk premiums and be self-fulfilling. Herd behaviors on financial markets and the increased interconnectedness among countries raise the probability of misestimating individual growth.
Forecasting errors may therefore turn into a policy tool, with a trade-off for decision makers: do they have to tell the truth or should they shift their prospects in order them to realize?
We illustrate this with the example of Greece and the IMF.
The following chart shows the continuous downwards revision of IMF on Greece forecasts from 2008 to 2013, similar to what is observed for the OECD and the European Commission (next charts).
What are the explanations for the Greek/IMF forecasting errors? A first, technical reason is that the Hellenic Statistical Authority has not provided seasonally adjusted figures since Q1 2011 because of “methodological issues.”[5] But program design[6] requires the country to be solvent in order to get exceptional access, and (to avoid contagion effects) the bar for debt sustainability was lowered thanks to higher growth forecasts.[7] Other institutions’ forecasts were similar, raising concerns about group think. The OECD (2013) published a post-mortem to explain why it under-estimated the fall of economic activity following the beginning of the crisis and why it recently over-estimated the pace of recovery. An important lesson is that forecasting models did not take well into account the interconnectedness between countries and so the existence of a systemic risk, induced by the deepening of globalization. The debt crisis, along with the link between banking and sovereign risks has been too neglected in models, leading to an overestimation of the recovery.
Improving models help make better forecasts, but in the end, predicting growth remains a political exercise.
Notes
[1] Politique économique, Chapter 4, A. Bénassy-Quéré et al, 2012
[2] Forecasting through the Rear-view Mirror: Data Revisions and Bond Return Predictability, E. Ghysels et al., staff report No. 581, FRBNY, 2012
[3] The Global Social Crisis, Report on the World Social Situation, United Nations, 2011
[4] Growth Forecast Errors and Fiscal Multipliers, O. Blanchard et D. Leigh, 2013
[5] Focus Economics, Greece - GDP
[6] IMF Completes Third Review Under Extended Fund Facility Arrangement for Greece, Concludes 2013 Article IV Consultation, Press Release No.13/195, 2013
[7] Greece: Ex-post Evaluation of Exceptional Access under the 2010 Stand-By Arrangement, IMF Country Report, 13/156, 2013
To read articles on similar topics, please click on the tags below:
