No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
How Accurate Are Private Sector Forecasts: Cross-Country Evidence From Consensus Forecasts of Output Growth
Abstract
Forecasts are an inherent part of economic science and the quest for perfect foresight occupies economists and researchers in multiple fields. The release of economic forecasts (and its revisions) is a popular and often publicized event, with a multitude of institutions and think-tanks devoted almost exclusively to that task. The European Central Bank (ECB) also publishes its forecasts for the euro area, however ECB’s forecast accuracy is not a deeply researched theme. The ECB forecasts’ accuracy is the main point developed in this paper, which tries to contribute to understand the nature of the errors committed by the ECB forecasts and its main differences compared to other projections. What we try to infer is whether the ECB is accurate in its projections, making less errors than the others, maybe due to some informational advantage. We conclude that the ECB seems to consistently underestimate the HICP inflation rate and overestimate GDP growth. Comparing it with the others, the ECB shows a superior performance, committing almost always fewer errors. So, this signals a possible informational advantage from the ECB. Since the forecasting errors could jeopardize ECB’s credibility public criticism could be avoided if the ECB simply let forecasts for the others. Naturally, this change should be weighted against the benefits of publishing forecasts.
... The professional macroeconomic forecasting literature of unbiasedness and rationality focuses on the United States of America (USA) and Group of Seven (G7) countries. Also Loungani (2001) finds differences in forecast performance of professionals between developed and emerging economies, where emerging economies tend to have larger forecasting errors. Only recently there are articles about emerging countries, as we can cite Issler and Lima (2009), Gaglianone et al. (2017) and Gaglianone and Issler (2021) for Brazil, Bentancor and Pincheira (2010) and Pincheira andÁlvarez (2012) for Chile, Deschamps and Bianchi (2012) for China, Capistrán and López-Moctezuma (2014) for Mexico, Chen et al. (2016) for a group of six emerging Asian countries (out of 10 analyzed). ...
... For emerging economies, professionals tend to have larger forecasting errors compared to those in advanced countries based on Loungani (2001). Deschamps and Bianchi (2012) study the SPF for China inflation and use the individual forecasts. ...
Our objective is to test the rationality of the forecasters for Brazilian inflation and we analyze the relationship of rationality with the macroeconomic and electoral variables. We use Survey of Professional Forecasters (SPF) for next month's inflation with monthly data. We consider times series and panel data traditional tests as Mincer and Zarnowitz (1969) and West and McCracken (1998) to verify if forecast errors have zero mean and are uncorrelated with variables available at the time the forecast is made. We use Rossi and Sekhposyan (2016) test with their asymptotic critical values and finite sample adjusted distribution critical values of El-Shagi (2019) to consider the possibility of instability. Also we consider the possibility of the forecasters loss function is asymmetric following Elliott et al. (2005) and whether forecasters adjust manually their forecasts based on their personal expertise according to Franses (2021). We reject forecast rationality with panel data or time series (consensus) using traditional tests. We do not reject the null hypothesis of rationality for the consensus inflation forecast if we use fluctuation rationality test. We obtain that forecasters have bias in inflation forecasts in the easing and tightening periods of monetary policy or election periods with panel data. But we have that economic cycle, monetary policy or election do not affect the rationality test with panel data. The consensus forecast seems to neutralize the bias of individual forecasts comparing with panel data and it reduces irrationality only for periods of recession, monetary policy tightening and without election. We obtain an asymmetric loss function mainly for institutions and we can reject forecast rationality by incorrectly considering the loss function as symmetric. Also we obtain that if the forecasters use an econometric model, they all adjust their model forecasts.
... The professional macroeconomic forecasting literature of unbiasedness and rationality focuses on the United States of America (USA) and Group of Seven (G7) countries. Also Loungani (2001) finds differences in forecast performance of professionals between developed and emerging economies, where emerging economies tend to have larger forecasting errors. Only recently there are articles about emerging countries, as we can cite Issler and Lima (2009), Gaglianone et al. (2017) and Gaglianone and Issler (2021) for Brazil, Bentancor and Pincheira (2010) and Pincheira andÁlvarez (2012) for Chile, Deschamps and Bianchi (2012) for China, Capistrán and López-Moctezuma (2014) for Mexico, Chen et al. (2016) for a group of six emerging Asian countries (out of 10 analyzed). ...
... For emerging economies, professionals tend to have larger forecasting errors compared to those in advanced countries based on Loungani (2001). Deschamps and Bianchi (2012) study the SPF for China inflation and use the individual forecasts. ...
Our objective is to test the rationality of the forecasters for Brazilian inflation and we analyze the relationship of rationality with the macroeconomic and electoral variables. We use Survey of Professional Forecasters (SPF) for next month's inflation with monthly data. We consider times series and panel data traditional tests as Mincer and Zarnowitz (1969) and West and McCracken (1998) to verify if forecast errors have zero mean and are uncorrelated with variables available at the time the forecast is made. We use Rossi and Sekhposyan (2016) test with their asymptotic critical values and finite sample adjusted distribution critical values of El-Shagi (2019) to consider the possibility of instability. Also we consider the possibility of the forecasters loss function is asymmetric following Elliott et al. (2005) and whether forecasters adjust manually their forecasts based on their personal expertise according to Franses (2021). We reject forecast rationality with panel data or time series (consensus) using traditional tests. We do not reject the null hypothesis of rationality for the consensus inflation forecast if we use fluctuation rationality test. We obtain that forecasters have bias in inflation forecasts in the easing and tightening periods of monetary policy or election periods with panel data. But we have that economic cycle, monetary policy or election do not affect the rationality test with panel data. The consensus forecast seems to neutralize the bias of individual forecasts comparing with panel data and it reduces irrationality only for periods of recession, monetary policy tightening and without election. We obtain an asymmetric loss function mainly for institutions and we can reject forecast rationality by incorrectly considering the loss function as symmetric. Also we obtain that if the forecasters use an econometric model, they all adjust their model forecasts.
... Furthermore, researchers report persistent systematic biases which are especially visible in the forecasts with longer horizons (Ager, Kappler, Osterloh 2009). There is also strong evidence of failure to predict severe downturns (Loungani 2001) or effects of structural changes, e.g. those related to fiscal policy (Blanchard, Leigh 2013). ...
... We follow an approach used previously by Ashiya (2003, 2009), Loungani (2001, and Lahiri and Sheng (2010). This test is also widely adopted in different contexts, for example, with fiscal forecasts (Artis, Marcellino 2001;Brück, Stephan 2006;Pina, Venes 2011). ...
The aim of this paper is to evaluate gross domestic product (GDP) forecast errors of Polish professional forecasters based on the individual data from the Rzeczpospolita daily newspaper. This dataset contains predictions obtained from forecasting competitions during the years 2013-2019 in Poland. Our analysis shows a lack of statistical efficiency of these predictions. First, there is a systemic negative bias, which is especially strong during the years 2016-2019. Second, the forecasters failed to correctly predict the effects of major changes in fiscal policy. Third, there is evidence of strategic behaviour; for example, the forecasters tended to revise their estimates too frequently and too excessively. We also document herding behaviour, i.e. an alignment of the most extreme forecasts towards market consensus with time, and an overly strong reliance on forecasts from NBP inflation projections in cases of estimates for longer horizons.
... Over our sample period, WEO forecasts for real GDP growth have shown a tendency towards over-optimism: on average, the prediction for next year's growth rate has been 0.58 percentage points higher than the subsequent realization. Such an upward bias is also documented in IEO (2014), while Loungani (2001) shows that it is present in private sector forecasts as well. This observation rhymes with Stylized Facts #2 and #3, which suggest that there is a reluctance to predict recessions. ...
... But given that private sector forecasts correlate highly with, and display similar biases to, forecasts produced by governments and the IMF (cf. Loungani (2001)), there is no strong evidence that this is the case. In addition, the downward bias described in footnote 22 remains present. ...
... Despite their relevance, medium term forecasts have not received as much attention as short term forecasts. More specifically, most of the literature has analyzed the properties of forecasts with prediction horizons of up to 24 months (Pons 2000, Loungani 2001, Ager et al. 2009, Dovern & Weisser 2011, Dovern et al. 2015. While the evidence found in this literature is informative about likely properties of medium term forecasts, it is not known the extent to which they share similar properties. ...
... Social learning and the likely use of common set of prediction tools can lead to similar forecasts for a broad set of economic actors. Additionally, previous literature shows different experts tend to generate similar forecasts (Loungani 2001, Timmermann 2007 To evaluate medium term forecasts released by other set of experts, a database of forecasts from a private consulting firm has been built. More specifically, the medium term forecasts correspond to those released by the Economist Intelligence Unit (EIU), the research and analysis division of The Economist Group. ...
We compare the medium-term GDP growth forecasts generated by experts to those generated by simple models. This study analyzes a large set of forecasts that covers 48 countries from 1997 to 2016. Out-of-sample exercises indicate that no noticeable difference in performance is observed for advanced economies. In contrast, in the case of emerging economies, model forecasts perform better than expert forecasts. In addition, similar patterns are found for a collection of forecasts from a different set of experts, which suggests that the reported regularity is prevalent. Further analyses suggest that the documented difference in performance can be explained by an optimism bias, excessive reactions to innovations in growth trajectories, and insufficient responses to the information contained in the current account balance.
... Previous research has also demonstrated that that persistent over-forecasting is not a unique to the IMF. Its widespread occurrence has been documented across forecasts prepared by national governments (Frankel (2011) and Frankel and Schreger (2012)), central banks (Alessi et al. (2014) and Lansing and Pyle (2015)), multilateral institutions (Genberg and Martinez, 2014) and those of the private sector such as the Consensus Forecast (Loungani, 2001). Studies have also found similar levels of efficiency forecasts and a near-perfect collinearity between these sources (Loungani (2001) and Timmermann and Granger (2004)). ...
... Its widespread occurrence has been documented across forecasts prepared by national governments (Frankel (2011) and Frankel and Schreger (2012)), central banks (Alessi et al. (2014) and Lansing and Pyle (2015)), multilateral institutions (Genberg and Martinez, 2014) and those of the private sector such as the Consensus Forecast (Loungani, 2001). Studies have also found similar levels of efficiency forecasts and a near-perfect collinearity between these sources (Loungani (2001) and Timmermann and Granger (2004)). ...
Oil discoveries can constitute a major positive and exogenous shock to economic activity, but the resource curse hypothesis would suggest they might also be detrimental to growth over the long run. This paper utilizes a new methodology for estimating growth underperformance to examine the extent to which discoveries depress the growth path of a country following a discovery and prior to production starting. The study finds causal evidence of a significant negative effect on short-run growth and growth relative to counterfactual forecast growth in countries with weak institutions, creating growth disappointments prior to private and public resource windfalls. This effect is termed the presource curse. For a giant oil or gas discovery in 1988-2010, the study estimates an average growth disappointment effect of 0.83 percentage points, measured as the average annual gap between forecast and actual growth over the five years following a discovery. Further, the estimated effect varies by the size of the discovery, increasing to a 1.77 percentage points gap in the case of super giant discoveries. The estimated effect is inversely related to the quality of political institutions, and driven by countries with lower institutional quality at the time of the discovery, consistent with the similar long-run results documented in the resource curse literature. For countries with below-threshold institutional quality, the growth disappointment effect is larger, measured as 1.35 percentage points in annual terms. There is no measured growth disappointment effect for countries with strong institutions. Using the synthetic control method, we confirm our findings for a selection of countries above and below the institutional quality threshold. The findings suggest that studies of the resource curse that focus only on the effects of resource exploitation or examine only long-run growth effects may overlook important short-run growth disappointments following discoveries, and the way countries respond to news shocks.
... Moreover,Loungani (2001) finds a high degree of similarity between the private sector (consensus) forecasts of output growth and the corresponding predictions of international organizations such as the IMF, the OECD, and the World Bank: Their forecast errors have a correlation of 0.7 or better. ...
This paper examines whether the current level of debt in the country, given the national government’s fiscal policy and plans, remains on a sustainable path. By end-2021, a year after the peak of the public health and economic crisis brought about by the COVID-19 pandemic, the country’s debt-to-GDP ratio had already climbed to 60.5 percent, over 20 percentage points above pre-pandemic levels and slightly above the government’s indicative cap. Several empirical exercises were performed in this paper to investigate the country’s fiscal solvency, namely by: (1) providing a historical decomposition of public
debt, (2) tracking the evolution of the debt-to-GDP ratio in the next half-decade through standard debt sustainability analysis (DSA), (3) computation of the fiscal gap to shed light on the fiscal adjustment needed to bring the country to more comfortable debt levels, and (4) estimation of fiscal reaction functions for the Philippines and developing ASEAN-5 economies to see how fiscal policy will likely respond to debt and other relevant macroeconomic conditions. Results suggest that the country’s debt position today is less worrisome than it had been during previous debt crises, and that the debt-to-GDP
ratio will remain manageable despite peaking above 65 percent over the next couple of years. Given the need to spend to prevent possible scarring from the pandemic and give the economy time as well as room to recover from the pandemic crisis, it may not be feasible to immediately return to pre-COVID-19 debt ratios, based on fiscal gap computations. This underscores the need for a sound medium- to long-term fiscal consolidation plan to anchor sentiments. Fiscal reaction functions for the Philippines and similar economies in the region meanwhile indicate responsible fiscal policy that guarantees fiscal solvency. This presupposes however the absence of major fiscal policy reversals, especially of hard-won fiscal reforms since the mid-1980s.
... Hence, for the investment effect to drive the results, individuals should be able to forecast economic growth correctly within a 30-to-50-year horizon; and they should also be able to discriminate in their forecast between the next 20 years and the next 30 to 50 years. This seems unlikely, given that even short-term economic forecasts by international and private organizations are not always accurate in predicting the rate of economic growth (Lewis and Pain, 2015;Loungani, 2001;Zarnowitz, 1991). ...
... Tables Table 1. Macroeconomic forecasters have an abysmal track record in predicting future recessions as borne out by numerous surveys of different forecasters over various periods for many countries (for example : Loungani, 2001;Abreu, 2011;Fildes and Steckler, 2002;Pain and Lewis, 2014). OECD forecasts published in the OECD Economic Outlook are typical; since 1991, less than 10% of forecasts for OECD countries published in May/June correctly identified a recession in the following year ( Figure 1). 2 OECD forecasts are also typical in being biased towards over-prediction, at least for forecasts beyond the current year, with the largest outliers corresponding to downturn periods, so that the distribution of forecast errors cannot be characterised by a normal distribution for forecast horizons beyond the current year ( Figure 2). ...
This paper describes a method for parameterising fan charts around GDP growth forecasts of the major OECD economies as well as the aggregate OECD. The degree of uncertainty – reflecting the overall spread of the fan chart – is based on past forecast errors, but the skew – reflecting whether risks are tilted to the downside – is derived from a probit model-based assessment of the probability of a future downturn.
This approach is applied to each of the G7 countries separately, with combinations of variables found to be useful in predicting future downturns at different horizons up to 8 quarters: at short horizons of 2-4 quarters, a flattening or inverted yield curve slope, recent sharp falls in house prices, share prices or credit; at longer horizons of 6-8 quarters, sustained strong growth in house prices, share prices and credit; and at all horizons, a tight labour market and rapid growth in OECD-wide (or in some cases euro-wide) house prices, share prices or credit. The in-sample fit of the probit models appears reasonably good for all G7 countries.
The predicted probabilities from the probit models provide a graduated assessment of downturn risk, which is reflected in the degree of skew in the fan chart. Fan charts computed on an out-of-sample basis around pre-crisis OECD forecasts published in June 2008 encompass the extreme outturns associated with the Global Financial Crisis for five of the G7 countries. A weakness of the approach is that, although it predicts a clear majority of past downturns, it will not predict atypical downturns. For example, in the current conjuncture, it is unlikely that current concerns about risks associated with Brexit, an escalation of trade tensions or spillovers from emerging markets would be picked up by the models. At the same time, a severe downturn triggered by such atypical events might be more severe if more typical risk factors are also high.
... better data availability or a superior performance of computers and programs) nor possible improvements in economic theories and methods led to a significant decline of forecast errors. This claim is further supported by findings that the anticipation of economic recessions failed in almost all cases (Loungani, 2001;Zidong An and Loungani, 2018). ...
Based on a panel of annual data for 17 growth and inflation forecasts from 14 institutions for Germany, we analyse forecast accuracy for the periods before and after the Great Recession, including measures of directional change accuracy based on Receiver Operating Curves (ROC). We find only small differences on forecast accuracy between both time periods. We test whether the conditions for forecast rationality hold in both time periods. We document an increased cross-section variance of forecasts and a changed correlation between inflation and growth forecast errors after the crisis, which might hint to a changed forecaster behaviour. This is also supported by estimated loss functions before and after the crisis, which suggest a stronger incentive to avoid overestimations (growth) and underestimations (inflation) after the crisis. Estimating loss functions for a 10-year rolling window also reveal shifts in the level and direction of loss asymmetry and strengthens the impression of a changed forecaster behaviour after the Great Recession.
... 4 Over our sample period , WEO forecasts for real GDP growth have shown a tendency toward over-optimism: on average, the prediction for next year's growth rate has been 0.58 percentage points higher than the subsequent realization. Such an upward bias is also documented in IEO (2014), while Loungani (2001) shows that it is present in private sector forecasts as well. ...
Analyzing International Monetary Fund (IMF) data, we find that overly optimistic growth expectations for a country induce economic contractions a few years later. To isolate the causal effect, we take an instrumental variable approach—exploiting randomness in the country allocation of IMF mission chiefs. We first document that IMF mission chiefs differ in their individual degrees of forecast optimism, yielding quasi-experimental variation in the degree of forecast optimism at the country level. The mechanism appears to run through excessive accumulation of debt (public and private). Our findings illustrate the potency of unjustified optimism and underline the importance of basing economic forecasts upon realistic medium-term prospects. (JEL C53, E23, E27, E32, F33, H63)
... Loungani (2002) states "the record of failure to predict recessions is virtually unblemished." This conclusion was based on the finding that only two of the 60 recessions that occurred around the world during the 1990s were predicted by forecasters a year in advance. ...
We test whether analysts display multiple biases in forecasting the Institute for Supply Management’s manufacturing purchasing manager’s index (PMI). We adopt a test that does not require knowledge of the forecaster’s prior information set and is robust to rational clustering, correlated forecast errors and outliers. We find that analysts forecast the PMI poorly and display multiple biases when forecasting. In particular, forecasters anti-herd and anti-anchor. Anti-herding supports a reputation-based notion that forecasters are rewarded not only for forecast accuracy but also for being the best forecast at a single point in time. Anti-anchoring is consistent with forecasters overreacting to private information. The two biases show a strong positive correlation suggesting that the incentives that elicit anti-herding also elicit anti-anchoring behavior. Both biases result in larger absolute errors, although the effect is stronger for anti-herding.
... Across all G7 countries, there are 13 occurrences when the absolute forecast error exceeds twice the RMSE of all forecast errors for that country, but all these occurrences are over-predictions rather than under-predictions and all are associated with negative growth outturns (Annex A, Table A3). This confirms a finding of previous evaluations of both OECD and other macroeconomic forecasts, namely that they are invariably poor at predicting turning points, particularly downturns (Fildes and Steckler, 2002;Loungani, 2001;Abreu, 2011;Pain and Lewis, 2016). ...
RÉSUMÉ Designing Fan Charts for GDP Growth Forecasts to Better Reflect Downturn Risks Forecasts of GDP growth are typically over-optimistic for horizons beyond the current year, particularly because they fail to predict the occurrence or severity of future downturns. Macroeconomic forecasters have also long been under pressure to convey the uncertainty surrounding their forecasts, particularly since the financial crisis. The current paper proposes a method to address both these issues simultaneously by constructing fan charts which are parameterised on the basis of the historical forecasting track record, but distinguish between a "safe" regime and a "downturn-risk" regime. To identify the two regimes, use is made of recent OECD work on early warning indicators of a prospective downturn, relating to housing market or credit developments. Thus, when an early warning indicator is "flashing", the associated fan chart is not only wider to reflect increased uncertainty, but is also skewed to reflect greater downside risks using a two-piece normal distribution of the form used by central banks to provide fan charts around inflation forecasts. Conversely, in a safe regime, when the early warning indicators are not flashing, as well as being symmetric, the fan chart is narrower both relative to the downturn-risk regime and relative to what the fan chart would be if the dispersion was calculated with respect to the entire forecast track record with no distinction between regimes. The method is illustrated by reference to OECD GDP forecasts for the major seven economies made just prior to the global financial crisis, with fan charts calibrated using the track record of forecasts published in the OECD Economic Outlook. Fan charts which take account of early warning indicators in this way are much better at encapsulating the outturns associated with a downturn than a symmetrical fan chart calibrated indiscriminately on all forecast errors. JEL codes: E58, E17, E65, E66, GO1 Concevoir des graphiques en éventail des prévisions de croissance du PIB pour mieux refléter les risques à la baisse Les prévisions de croissance du PIB sont typiquement trop optimistes à des horizons dépassant l'année courante, particulièrement car elles échouent à prédire l'arrivée ou la sévérité des crises futures. Les macro-économistes sont depuis longtemps sous pression pour communiquer l'incertitude entourant leurs prévisions, surtout depuis la crise financière. Ce papier propose une méthode pour traiter simultanément ces deux problèmes, en construisant des graphiques en éventail, paramétrés par l'historique des prévisions passées, qui font la différence entre un régime « sûr » et « risqué ». Pour identifier ces deux régimes, les travaux récents de l'OCDE sur les indicateurs avancés des crises potentielles, liés au marché du logement ou à l'évolution du crédit, sont utilisés. Ainsi, lorsqu'un indicateur « clignote », le graphique en éventail associé est non seulement plus large pour refléter l'incertitude, mais aussi déformé pour montrer les risques plus importants à la baisse, en utilisant une distribution normale à deux-pièces similaire à celle utilisée par les banques centrales pour produire les graphiques en éventail des prévisions d'inflation. Inversement, dans un régime sûr, lorsque les indicateurs avancés ne clignotent pas, les graphiques en éventail sont non seulement symétriques mais également plus étroits comparés à ceux du régime risqué et à ce que seraient les graphiques si la dispersion était calculée sur tout l'historique des prévisions sans distinguer les régimes. Cette méthode est illustrée par les prévisions du PIB de l'OCDE pour les sept économies majeures effectuées juste avant la crise financière, avec des graphiques en éventail calibrés par l'historique des prévisions publiées dans les Perspectives économiques de l'OCDE. Les graphiques en éventail qui prennent en compte les indicateurs avancés sont beaucoup plus performants pour contenir les résultats associés à une crise que ceux qui sont symétriques et calibrés sur toutes les erreurs de prévisions. Codes JEL: E58, E17, E65, E66, GO1 Mots clés : Graphique en éventail, prévisions économiques, incertitude, risque, crise ECO/WKP(2017)60
... This is acknowledged byArtis (1996), who makes a visual comparison of IMF and Consensus Economics forecasts for real GDP and inflation, and concludes that there is "little difference between WEO and Consensus errors". In a similar vein,Loungani (2001) plots IMF and Consensus Economics real GDP forecasts and notes that "the evidence points to near-perfect collinearity between private and official (multilateral) forecasts . . ." 10 Country list includes: Argentina, Brazil, Bolivia, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Mexico, Panama, Paraguay, Peru, Uruguay, and Venezuela. ...
This paper provides a full characterization of inflation rate forecasts using the mean values from Consensus Economics for a sample of 14 Latin American countries between 1989 and 2014. It also assesses the performance of inflation rate forecasts around business cycles' turning points. Results show that inflation forecasts in the region display the standard property that as the forecast horizon shortens accuracy improves. On average, forecasters underpredict inflation, but this masks very different country experiences. We find evidence point to biasedness of inflation forecasts for year-ahead forecasts but not for current year. Tests' results point to forecast inefficiency which is also evidenced by a tendency to smooth them between revisions. Focusing on business cycle turning points, forecasters tend to slightly underpredict the inflation rate and the extent of underprediction increases during recessions. The hypothesis of forecast efficiency is overwhelmingly rejected both during recessions and recoveries.
... As Juhn and Loungani (2002) find in case of industrialized countries, Root Mean Square Error of 9 months ahead forecast is 1.37 by both International Monetary Fund (IMF) and Consensus forecasts, the same for developing countries are 2.82 by Consensus Economics and 3.24 by IMF. The Theil's U statistics for these forecasts indicate that not only the magnitude, even the directions are not right in many cases (Loungani, 2001). ...
This paper evaluates the accuracy of a forecast based on the properties of the forecast error. To measure how close, the predictions of GDP growth are to the actual outcome in India, we have calculated three measures of forecast accuracy: MAE, RMSE and Theil’s U statistic. To evaluate the performance of the forecasts, we have compared them with naive forecast and common rules of thumb, using moving averages as rules of thumb. The results are
inconclusive regarding biasedness and clearly inefficient. Further, the forecasts have a high degree of correlation amongst themselves. The findings of forecast errors suggest that the performance of RBI forecasts are favorable compared to the other organizations, as well as with respect to the general international standard.
... There is evidence in the economic literature that an optimistic bias exists in private expert forecasts of real GDP growth rates, especially near cyclical peaks 3 ( Loungani (2001) (a t+0 , a t+1 , a t+2 , a t+3 , and a t+4 ) also taken from the PhilFed historical database. As an actual we used the first (or advance) GDP estimate published by the BEA. ...
... Initially, we assumed that 26 Full regression parameters are available upon request. 27 Until the end of the 1980s averages were usually less than medians. It means that there were some respondents at that time who forecasted much lower GDP growth rates than the majority of the expert society. ...
There is evidence in the economic literature that professional forecasters are unsuccessful in predicting recessions, but the reasons for these failures are still not clear. Meanwhile, this phenomenon has been little studied on the basis of quarterly estimates for various target horizons. We analysed quarterly consensus forecasts of real GDP growth rates and probabilities of recession taken from the Survey of Professional Forecasters (SPF) conducted by PhilFed and established several stylized facts including: “alarm signals” usually appear only after cyclical peaks; consensus forecasts of recessions for distant target horizons (more than two quarters) have never met except several quarters after the second oil shock; as a rule, pre-recession probabilities of recessions are much less than 50%; the expected durations of recession are less than actual ones; the Mincer-Zarnowitz test with a dummy for recessions reveals that SPF give biased forecasts of real GDP growth rates for almost all target horizons; experts regularly overestimate growth rates during recessions; adding a dummy for recessions significantly increases adjusted R2s; consensus forecasts clearly signal a recession only after a black swan; the majority of experts avoid predicting declines in real GDP before a recession; depth of contraction is even more underestimated for quarters after black swans. None of these stylized facts proves the unwillingness of professional forecasters to predict recessions (especially prolonged ones) in a direct way. However, in our view, together they constitute indirect evidence for the existence of a wishful bias against predicting recessions. If this bias exists, customers of SPF forecasts should take this into account in their decision-making processes.
... Aunque los economistas han mostrado con cierta claridad las causas y consecuencias de la reciente recesión global y la recuperación subsecuente, nuestra comprensión de por qué tienen lugar estas situaciones aún sigue siendo limitada (Kose y Terrones, 2015). Loungani (2001) recoge un estudio sobre la exactitud de las previsiones económicas de crecimiento del PIB real en un número amplio de países industrializados en la década de los noventa. Concluye respecto al fracaso generalizado de predecir recesiones y señala que existe un alto grado de similitud entre las previsiones de crecimiento del sector privado y de las organizaciones internacionales (FMI, la OCDE y el Banco Mundial). ...
La era actual y el desarrollo de internet han propiciado la accesibilidad a diferentes fuentes de información estadística. Disponemos de más datos cuantitativos, más información, aunque no siempre significan más conocimiento. El gran reto es convertir los datos disponibles en información y ésta en conocimiento. Una parcela interesante del conoci- miento en economía es la que nos proporciona la posibilidad de ahondar en el ámbito de la predicción, favorecida por la disponibilidad de datos y de programas informáticos específicos. La principal magnitud que aún se sigue em- pleando para calificar la prosperidad económica de un país o área territorial es el Producto Interior Bruto (PIB). Su tasa de variación precisa ser interpretada correctamente para extraer conclusiones relevantes en la toma de decisiones económicas, en la generación y difusión de conocimiento. Revisamos los principales aspectos a considerar en el análisis de predicciones del PIB regional en España, con la finalidad de contribuir a esclarecer el interés por la combinación de predicciones alternativas o la apuesta por una predicción en particular por su mayor fiabilidad.
The current age and the internet development give raise accessibility to different sources of statistical information. We have more information, though not always it means an increase of knowledge. The great challenge is to turn the available economic data into information and this one into economic knowledge. The possibility of going more deeply into the area of economic forecasting is an interesting focus of knowledge in economic science, supported by the availability of information and specific software. The gross domestic product (GDP) still continues being used as the main magnitude to qualify the economic prosperity of a country or geographic area. It is necessary to interpret correctly the GDP variation in order to extract relevant conclusions for the economic decisions, for the generation and dissemination of knowledge. We review some relevant aspects to consider in the process of regional economic forecasting: forecast combination or to choose the most accurate prediction.
... Interestingly, there is much recent literature documenting that at the individual level experts fail to reach this objective, in the sense that expert-adjusted forecasts are often not better than the original model-based forecasts; see Franses (2014) and references cited therein. By contrast, Ang, Bekaert, and Wei (2007), Dovern andWeisser (2011), andLoungani (2001), among others, show that "consensus" (i.e., combined) survey-or expert-based forecasts do outperform model-based forecasts. This also transpires in Armstrong's (2001) review of the literature on forecast combination, which includes forecasts ranging from model-based forecasts to expert forecasts and expert-adjusted model-based forecasts. ...
It is well known that a combination of model‐based forecasts can improve upon each of the individual constituent forecasts. Most forecasts available in practice are however not purely based on econometric models but entail adjustments, where experts with domain‐specific knowledge modify the original model forecasts. There is much evidence that expert‐adjusted forecasts do not necessarily improve the pure model‐based forecasts. In this paper we show however that combined expert‐adjusted model forecasts can improve on combined model forecasts, even in the case when the individual expert‐adjusted forecasts are not better than their associated modelbased forecasts. We discuss various implications of this finding.
... Hence, for the investment effect to drive the results, individuals should be able to forecast economic growth correctly within a 30-to-50-year horizon; and they should also be able to discriminate in their forecast between the next 20 years and the next 30 to 50 years. This seems unlikely, given that even short-term economic forecasts by international and private organizations are not very accurate in predicting the rate of economic growth (Loungani, 2001;Zarnowitz, 1991). ...
Using a data set for a panel of 118 countries, this paper shows that changes in the level of education of national populations ages 45 to 64 are positively associated with economic growth. An increase of one percentage point in the share of individuals in this age group who attended secondary education is associated with a 1.1% increase in GDP per capita, although the effect is stronger for developing countries. In contrast, variation in the level of education in younger cohorts is not positively associated with economic growth. These results suggest that investment in education benefits society, but only in the long-term. Several possible explanations for this finding are discussed.
... An analog but opposite phenomena has been observed in the case of GDP forecasts in times of recessions reported by the official and private organizations. Examples of that are found in Barrionuevo (1993) in the case of IMF projections 16 or Loungani (2000) in the case of Consensus Forecast. 17 However, some important future research avenues can be identified. ...
This article presents a comparative analysis of GDP growth forecasts for 19 Latin American and Caribbean (LAC) countries, reported by the IMF and ECLAC for the period 2003–2013. Contradicting the general result of the literature that asserts that economic predictions of official organizations tend to be optimistic, our results show that during that period, one year-ahead IMF and ECLAC GDP growth forecasts were downward biased. Furthermore, the analysis of usual goodness of fit measures and accuracy tests show that ECLAC's forecasts performed relatively better than that of the IMF during the period. Another interesting result is that no evidence was found regarding the influence of political type incentives on the bias of GDP forecasts as suggested by some authors. Instead, the results show that downward bias in projections has been strongly influenced by the underestimation of the impact of international economic factors, particularly the drastic increase in commodity export prices, on GDP growth of LAC economies.
... Much research on professional forecasters concerns their joint predictive accuracy and their behavior relative to each other. Important studies are Batchelor (2001Batchelor ( , 2007, Dovern and Weisser (2011), Frenkel et al. (2013), Isiklar et al. (2006), Lahiri and Sheng (2010), Loungani (2001), Capistrán and Timmermann (2009), Genre et al. (2013) and Laster et al. (1999), where the focus is on accuracy, on disagreement across forecasters, and their eventual herding behavior. ...
Professional forecasters can rely on an econometric model to create their forecasts. It is usually unknown to what extent they adjust an econometric model-based forecast. In this paper we show, while making just two simple assumptions, that it is possible to estimate the persistence and variance of the deviation of their forecasts from forecasts from an econometric model. A key feature of the data that facilitates our estimates is that we have forecast updates for the same forecast target. An illustration to consensus forecasters who give forecasts for GDP growth, inflation and unemployment for a range of countries and years suggests that the more a forecaster deviates from a prediction from an econometric model, the less accurate are the forecasts.
... However, very similar results were obtained by including only the following-year forecast, only the current-year forecast, or the following-year forecast together with the difference between the two. In these cases, the dispersion measures were seasonally adjusted to account for the smaller dispersion of forecasts-documented by Loungani (2001)-at the end of each year than at the beginning of each year. 16 Using the mean instead of the median forecast or using the standard deviation instead of the mean absolute median difference would provide essentially the same results. ...
This paper studies how uncertainty about fundamentals contributed to currency crises from both a theoretical and an empirical perspective. We find evidenceCbased on a monthly dataset of Consensus forecasts for six Asian countries in the period January 1995-May 2001Cconfirming the theoretical predictions (from both unique- and multiple-equilibria models) that: (i) speculative attacks depend not only on actual and expected fundamentals but also on the variance of speculators' expectations about them; and (ii) the sign of the effect of the variance depends on whether expected fundamentals are "good" or "bad." These results are robust to the definition of exchange rate pressure indices, the estimation sample (precrisis vs. full sample), the method chosen to avoid spurious correlations, and possible time-varying coefficients for the mean, the variance, and the threshold separating good from bad expected fundamentals.
... Arora and Smyth (1990) investigated IMF's forecasts for developing countries, and found that these forecasts are not significantly different from a naïve model, and do not improve over time. Loungani (2001) and Timmermann (2007) examined and compared the forecast performances for both advanced and developing countries provided by three international organisations and private forecasts, and found a high degree of similarity among their projections. Further, Krkoska andTeksoz (2007, 2009) examined the EBRD forecasts for 25 transition countries in Central and Eastern Europe between 1994 and the mid-2000s. ...
This study assesses the performance of the forecasts of the Asian Development Bank (ADB) for 45 countries between 1989 and 2019 with the following results: first, forecast accuracy and bias vary among countries and across variables; second, the Asian currency and global financial crises adversely affected the GDP and inflation forecast accuracy, whereas there were no such impacts on the current account forecast accuracy; third, the ADB's forecasts are broadly unbiased; fourth, inflation forecasts are efficient but the GDP and current account forecasts are not; finally, unbiasedness is broadly robust for the countries' regions. These results suggest that the ADB plays a grounding economic role in providing a reliable economic outlook for policymakers and businesses world wide.
... Hence, for the investment effect to drive the results, individuals should be able to forecast economic growth correctly within a 30-to-50-year horizon; and they should also be able to discriminate in their forecast between the next 20 years and the next 30 to 50 years. This seems unlikely, given that even short-term economic forecasts by international and private organizations are not always accurate in predicting the rate of economic growth (Lewis and Pain, 2015;Loungani, 2001;Zarnowitz, 1991). ...
... In this subsection, as a additional evaluation of the extent to which these systematic errors are widespread, a collection of private macroeconomic forecasts are evaluated and compared to WEO forecasts. While previous evidence suggests that there are important similarities between forecasts released by different analysts (Loungani 2001, Gavin & Mandal 2001, a specific evaluation of their performance following large current account deficits is valuable in the context of the current study. ...
Using a sample covering 50 advanced and emerging economies over 1990–2017, it is found that large current account deficits are followed by systematic negative surprises in economic growth. This regularity is verified both in the case of advanced economies and emerging economies. In addition, large current account deficits are reversed significantly faster than what forecasters anticipate and are followed by low asset returns and drops in sentiment. The findings are robust to changes in the specification and do not seem to be explained by efficient learning dynamics. This evidence indicates that analysts are unable to incorporate the negative information transmitted by large current account deficits and has implications for the understanding of past economic events and for the design of macro-prudential policies.
... The accuracy of our model's GDP estimates steadily increases (Figure 4) tently fall throughout the forecasting horizon. 21 These results mirror the ability of market participants to increase the accuracy of their forecasts as they increase their information set (e.g., Loungani, 2001). Additionally, these results contrast with the constant RMSFEs of the benchmark auto-regressive model (blue line). ...
We build a model for simultaneously now-casting economic conditions in the euro area and its three largest member countries|Germany, France, and Italy. The model formalizes how market participants and policymakers monitor the euro area by incorporating all market moving indicators in real time. We find that area wide and country-specific data provide informative signals to now-cast the economic conditions in the euro area and member countries. The model provides accurate predictions of economic conditions in real time over a period that covers the past three recessions.
... The accuracy of our model's GDP estimates steadily increases (Figure 4) tently fall throughout the forecasting horizon. 21 These results mirror the ability of market participants to increase the accuracy of their forecasts as they increase their information set (e.g., Loungani, 2001). Additionally, these results contrast with the constant RMSFEs of the benchmark auto-regressive model (blue line). ...
Now casting the euro area GDP is an arduous task but we learn from market participants that country-specific data and soft data are very important. In this paper we formalize those lessons!
... In fact, whether forecasts over-react or under-react to a particular piece of news will then depend 1 See e.g. Lovell (1986); Ehrbeck and Waldmann (1996); Loungani (2001); Fildes and Stekler (2002); Coibion and Gorodnichenko (2015). 2 The advantages of using the SPF include the long time span of data availability at a quarterly frequency and the fact that one can observe forecasts made at different time points for different forecasting horizons for the same individual forecaster. The latter is necessary to calculate revisions and to investigate stickiness in expectations. ...
We propose a simple model of expectation formation with three distinct deviations from fully rational expectations. In particular, forecasters’ expectations are sticky, extrapolate the most recent news about the current period, and depend on the lagged consensus forecast about the period being forecast. We find that all three biases are present in the Survey of Professional Forecaster as well as in the Livingston Survey, and that their magnitudes depend on the forecasting horizon. Moreover, in an over-identified econometric specification, we find that the restriction on coefficients implied by our model is always close to being satisfied and in most cases not rejected. We also stress the point that using the past consensus forecast to form expectations is a rather smart thing to do if cognitive limitations and biases cause any attempt to build an own rational forecast to fail.
... It is perhaps not surprising that the SEP participants failed to predict the Great Recession. In a cross-country study of professional economic forecasters from 1989 to 1998, Loungani (2001) finds that "the record of failure to predict recessions is virtually unblemished." ...
Growth forecasts by Federal Open Market Committee meeting participants were persistently too optimistic for 2008 through 2016. The typical forecast started out high but was revised down over time, often dramatically, as incoming data failed to meet expectations. In contrast, forecasts for 2017 through 2019 started low but were revised up over time. Cumulative forecast revisions for these years were much smaller on average than in the past. These observations suggest that participants have adjusted their forecast methodology, including lowering estimates of trend growth, to eliminate the prior optimistic bias.
... Forecasts also persistently fail to predict turning points, i.e. when GDP growth changes sign. In one study, in 60 cases of negative growth the consensus forecast was for negative growth on only three of those occasions (Loungani 2001). ...
Has the rise of data-intensive science, or ‘big data’, revolutionized our ability to predict? Does it imply a new priority for prediction over causal understanding, and a diminished role for theory and human experts? I examine four important cases where prediction is desirable: political elections, the weather, GDP, and the results of interventions suggested by economic experiments. These cases suggest caution. Although big data methods are indeed very useful sometimes, in this paper’s cases they improve predictions either limitedly or not at all, and their prospects of doing so in the future are limited too.
... For our uncertainty measure, we now use current year output growth errors from Consensus Forecasts, rather than those from the IMF. However, Loungani (2001) shows a high degree of similarity between private sector growth forecasts and those of international organisations. 29 Consensus Forecasts produce output growth estimates at a monthly frequency, ensuring that we have a different observation for each quarter. ...
We examine an unexplored connection between loss aversion and international consumption smoothing. In the face of expected income declines, loss-averse behaviour implies that any adjustments in consumption are delayed until they are necessary. However, if the expected fall in income materialises, the eventual adjustments in consumption are larger. Therefore, loss aversion could partly explain weaker-than-expected international consumption smoothing. We test this mechanism using measures of economic sentiment. These provide a sense of the expected direction of future income changes. We demonstrate that prevailing confidence and uncertainty, our proxies for economic sentiment, reduce the degree of international consumption smoothing. We provide evidence that this effect arises from loss-averse behaviour, using a test of the asymmetric response of consumption to positive and negative output fluctuations that occur following periods of weak economic sentiment. Our findings have important implications for public institutions and private initiatives that aim to improve cross-country risk sharing.
... Other papers have found comparable results. On the first finding that recessions are difficult to forecast, Lewis and Pain (2014) also point to "a common failing to predict downturns and to predict their size" and add that "these difficulties have been found across forecasters, across countries and over longer periods of time (Zarnowitz, 1991;Loungani, 2001;Abreu, 2011;González Cabanillas and Terzi, 2012)." Dovern and Jannsen (2017) also analyze how the systematic growth forecast errors in advanced economies depend on the business cycle, and document the fact that "growth forecasts for recessions are subject to large negative systematic errors, while forecasts for recoveries are subject to small positive systematic errors." 1 On the second finding, Abreu (2011) studies a sample of nine advanced economies over the period from 1991 to 2009, and finds that "[…] the forecasting performance of the international organisations is broadly similar to that of the surveys of private analysts. ...
We describe the evolution of forecasts in the run-up to recessions. The GDP forecasts cover 63 countries for the years 1992 to 2014. The main finding is that, while forecasters are generally aware that recession years will be different from other years, they miss the magnitude of the recession by a wide margin until the year is almost over. Forecasts during non-recession years are revised slowly; in recession years, the pace of revision picks up but not sufficiently to avoid large forecast errors. Our second finding is that forecasts of the private sector and the official sector are virtually identical; thus, both are equally good at missing recessions. Strong booms are also missed, providing suggestive evidence for Nordhaus’ (1987) view that behavioral factors—the reluctance to absorb either good or bad news—play a role in the evolution of forecasts.
This paper provides a full characterization of inflation rate forecasts using the mean values from Consensus Economics for a sample of 78 advanced and emerging economies between 1989 and 2014. It also assesses the performance of inflation rate forecasts around business cycles’ turning points. As expected, that inflation forecasts start to mirror the actual data as the forecast horizon draws to a close, particularly in advanced economies. The mean forecast error is positive and larger than one point when we pool all countries, but this masks inter-group differences. Moreover, we find evidence for biasedness, inefficiency or information rigidities, with a clear tendency for “forecast smoothing”. Accounting for cross-country informational linkages is important: forecasters fail to adjust their inflation forecasts quick enough in response to domestic news and news from abroad. Finally, during recession episodes forecasts generally appear to be inefficient. The same holds true for recoveries.
This paper assesses the performance of the IMF’s unemployment forecasts for 84 countries, both advanced and emerging market economies, between 1990 and 2015. The forecasts are reported in the World Economic Outlook, a leading IMF publication. The forecasts display a small amount of bias—they tend to predict lower unemployment outcomes than occur—which arises because the forecasters fail to predict accurately the sharp increase in unemployment during downturns. Forecasts are characterized by inefficiency (errors of the past are repeated in the present) and rigidity (forecast revisions are serially correlated). There is little to choose between IMF and Consensus Forecasts, a source of private sector forecasts, for the small subset of 12 countries for which both sets of forecasts are available.
Neo-liberal policies have the same general effect in the universities as they do in society as a whole. In society, their tendency has been to form large inequalities in the distribution of income and wealth. Similarly, in the case of higher education and scientific research, more and more funding is going to a few privileged universities and their researchers at the expense of the others. This is justified on the grounds that these universities and researchers are better than the others, so that it more efficient to concentrate funding on them. To find out how to distribute funding regular research assessments are conducted. But how accurate are these research assessments in picking out the researchers who are better from those who are not so good? It is argued that research evaluation as well as the pressure to reward the top-excellence is enhancing conformism and thus it is stifling the diversification of research, the essential element to develop innovations and new technologies. Is competition increasing researchers productivity and differentiation of research projects? How to test that a method of organization and evaluation of science is more efficient than another? Historical examples of discoveries in physics and mathematics in the past thirty years, as the super-high-temperature conductivity, the scanning tunneling microscope, graphene, and others, allow us to clarify how scientific research proceeds and how the findings of basic research are transformed into technological innovations.
The upward phase of the South African business cycle that followed the global financial crisis gradually lost momentum from 2012 onwards, resulting in a peak in the business cycle being reached in November 2013. The subsequent downward phase has so far lasted for 48 months, up to November 2017. The gradual loss of momentum in domestic output growth was not widely recognised by many economists, judging by consensus forecasts of annual GDP growth. However, the South African Reserve Bank (SARB) composite leading business cycle indicator reached a peak in February 2011 before trending gradually lower, suggesting a continued deterioration in output growth. The leading indicator reached a trough in April 2016, suggesting an imminent end to the downward phase of the business cycle, in line with consensus forecasts. However, output growth slowed further with the economy experiencing a technical recession. This paper employs the indicator approach to business cycle analysis in order to establish whether the current downward phase in the South African business cycle could reasonably have been predicted, and also whether the recent strong increase in the composite leading business cycle indicator provided a clear signal that the current downward phase might have ended. The results show that closer analysis of the leading indicator and its subcomponents would have revealed that a broad slowdown in the South African economy was underway since 2012. Also, the recent upward trend in the leading indicator did initially signal the end of the current downward phase in the business cycle in an unambiguous manner. However, the anticipated business cycle recovery appears to have been postponed by idiosyncratic exogenous factors.
High-quality macroeconomic forecasts are crucial inputs for economic decisions and policy making. Both the accuracy of forecasts and the efficiency with which information is incorporated into forecasts are critically important. At the same time, these forecasts are persistently too optimistic.
To explain this phenomenon, the authors explore common drivers of macroeconomic forecast overoptimism in different countries using the principal component analysis. They find that most of the variability in optimistic next-year forecast errors can be explained by four common factors. The optimism or pessimism with respect to gross domestic product (GDP) targets exhibits a certain degree of consistency across countries. Uncertainty about U.S. macrofinancial developments and global demand are the key drivers of forecast overoptimism. These common factors matter most for advanced economies and G20 members. Moreover, the explicit link between uncertainty about U.S. macroeconomic developments and next-year forecast errors has implications for the trajectory of macroeconomic variables. A vector autoregression (VAR) analysis shows that upward surges in uncertainty about U.S. business conditions lead to a decline in the next-year GDP growth rate in advanced economies and emerging countries. This result supports the link between uncertainty and overoptimism in next-year forecast errors. It also implies that incorporating the common structure governing forecast errors across countries can help improve subsequent forecasts and future policy making.
Historically, forecasters have failed to predict cyclical turning points and the forecasting record in this regard has not improved. This suggests that we should focus on what should be an easier task, recognizing recessions as they occur. We present a new approach that will enable us to determine in real-time when there is a significant deviation from an economy’s dynamic growth path. This approach uses a CUSUM-like methodology and requires us to construct an index,that we call the Economic News Index, from real-time data that shows how the economy is functioning. We apply this approach to German data to nowcast the recessions that began in 2008 and 2012.
Using a panel of 54 countries between 1980 and 2013, we find empirical support for the view that changes in the fiscal policy stance (year-on-year change in the cyclically adjusted primary balance) have a significant positive correlation with inflation volatility. An increase in the volatility of discretionary fiscal policies by one standard deviation raises inflation volatility by about 6%. Moreover, results using alternative inflation volatility proxies confirm that an expansionary fiscal stance increases price volatility. Another relevant outcome is that in a context of economic expansions (recessions) the harmful impact of fiscal activism on price volatility is soft (heightened), while the negative impact of fiscal activism on price stability is higher when fiscal policy is expansionary. Finally, fiscal activism fuels inflation volatility much more pronouncedly in emerging market economies vis-à-vis advanced economies.
Free competition for truth, wealth and power is often cited as the historic contribution of Western civilisation. Based on the principle of market competition, capitalism has secured an unprecedented level of technological and economic advancement. The chapter argues that the principle of competition has morphed into a hyper-competitiveness that pervades all aspects of people’s lives, and not just the economic sphere, particularly in the neo-liberal Anglosphere. The authors question whether the ideologically presented condition for competition, the ‘free market’ and its alleged result, a state of ‘meritocracy,’ genuinely exist. The chapter discusses the social and psychological cost of competition and reviews arguments for more cooperative structures, including some real-life examples.
The Western consumer society, focused on shopping and material abundance, has been critiqued since the creation of the concept in the 1950s. The chapter discusses the idea of hyper-consumption driven by intense marketing focused on triggering people’s feeling of relative deprivation and the resulting desire to ‘improve’ and display one’s status through conspicuous consumption. Early modern examples of conspicuous consumption and their contribution to the fund of Western ‘high culture’ are discussed. The chapter juxtaposes the general hyper-consumption with the documented rise in inequality within Western countries since the 1970s, and especially in the Anglosphere where redistribution through the welfare state is limited. The case study of excessive remuneration in the finance industry is contrasted with new forms of poverty, including the working poor.
We describe the evolution of forecasts in the run‐up to recessions. The GDP forecasts cover 63 countries for the years 1992–2014. The main finding is that, while forecasters are generally aware that recession years will be different from other years, they miss the magnitude of the recession by a wide margin until the year is almost over. Forecasts during non‐recession years are revised slowly; in recession years, the pace of revision picks up but not sufficiently to avoid large forecast errors. Our second finding is that forecasts of the private sector and the official sector are virtually identical; thus, both are equally good at missing recessions. Strong booms are also missed, providing suggestive evidence for Nordhaus' view that behavioural factors—the reluctance to absorb either good or bad news—play a role in the evolution of forecasts.
The association between GDP growth forecasts and past information flows is evaluated for a sample of 49 countries during the period 1990–2014. The analysis exploits an extensive collection of forecasts available through IMF's historical database. The empirical results indicate a robust association between information arrival and subsequent mean forecast errors (the average difference between forecast and realization). Consistent with the overreaction hypothesis, more positive information is followed by higher mean forecast errors. The association is documented for multiple metrics of past information flows: growth performance, a novel metric of press sentiment, and lagged forecast errors. When advanced and emerging economies are differentiated, the regularity is detected for both groups but is stronger in the case of emerging economies.
Fan charts were pioneered by the Bank of England and Riksbank and provide a visually appealing means to convey the uncertainty surrounding a forecast. This paper describes a method for parameterising fan charts around GDP growth forecasts by which the degree of uncertainty is based on past forecast errors, but the skew is derived from a probit model-based assessment of the probability of a future downturn. The probit-based fan charts clearly out-perform the Bank of England and Riksbank approaches when applied to forecasts made immediately preceding the Global Financial Crisis. These examples also highlight weaknesses with the Bank of England and Riksbank approaches.
* The Riksbank approach implicitly assumes that forecast errors are normally distributed, but over a long track record this is unlikely to be the case because forecasters are generally poor at predicting downturns, which leads to bias and skew in the pattern of forecast errors. Thus, the Riksbank fan chart is neither an accurate representation of past forecast errors, nor is it a reflection of the risk assessment underlying the forecast.
* The Bank of England approach relies heavily on the judgment of the members of the Monetary Policy Committee to assess risks. However, even when they have correctly foreseen the nature of future risks, the quantitative translation of these risks into the fan chart skew has been too timid. Perhaps one reason for this is that the fan chart prediction intervals based on historical forecast errors already appear quite wide so that inflating them by adding skew may appear embarrassing (at least ex ante).
The approach advocated in this paper addresses these weaknesses by recognising that forecast errors are not symmetrical: firstly, this leads to more compressed prediction intervals in the upper part of the fan chart (representing the possibility of under-prediction); and secondly, using the large forecast errors from past downturns to calibrate downward skew clearly supports a more bold approach when there is a risk of a downturn. A weakness of the probit model-based approach is that it will not predict atypical downturns. For example, in the current conjuncture it would not pick up risks associated with a ‘no deal’ Brexit or a global trade war. However, a downturn triggered by atypical events may be more severe if risk factors describing a typical business-financial cycle are also high.
We analyze the properties of multiperiod forecasts which are formulated by a number of companies for a fixed horizon ahead which moves each month one period closer and are collected and diffused each month by some polling agency. Some descriptive evidence and a formal model suggest that knowing the views expressed by other forecasters the previous period is influencing individual current forecasts in the form of an attraction to conform to the mean forecast. There are two implications: one is that the forecasts polled in a multiperiod framework cannot be seen as independent from one another and hence the practice of using standard deviations from the forecasts' distribution as if they were standard errors of the estimated mean is not warranted. The second is that the forecasting performance of these groups may be severely affected by the detected imitation behavior and lead to convergence to a value which is not the "right" target (either the first available figure or some final values available at a later time).
This note demonstrates that the conventional test for unbiasedness of forecasts or expectations of estimating Y = "alpha"(subscript "1") + "beta"(subscript "1")F + u where Y is the outcome and F is the forecast, and testing whether "alpha"(subscript "1" = 0 and "beta"(subscript "1") = 1 is a sufficient, but not a necessary, condition for unbiasedness. A necessary and sufficient condition for unbiasedness, based on the conventional test, is presented. Copyright 1990 by Blackwell Publishers Ltd and The Victoria University of Manchester
The authors define "consensus" as the degree of agreement among point predictions aimed at the same target by different individuals and " "uncertainty" as the diffuseness of the corresponding probability distributions. This distinction is made operational with the aid of the NBERAASA survey data on matched point and probabilistic forecasts of inflation and the rate of change in gross national product. The means of the two sets of forecasts agree closely. Standard deviations of point forecasts tend to understate uncertainty as measured by standard deviations of the predictive probability distributions. However, these measures of consensus and uncertainty are on the whole positively correlated. Copyright 1987 by University of Chicago Press.
This article introduces the concept of forecast efficiency, in which the forecast contains all information available at the time of the forecast. Empirical tests investigate weak efficiency, where the information set is all past forecasts and where all forecast revisions and errors should be uncorrelated with past forecast revisions. Tests of macroeconomic, energy-consumption, and oil-price forecasts find a significant autocorrelation of forecast revisions, with fifty of fifty-one tests showing positive correlation of forecast revisions, as opposed to zero correlation consistent with forecast efficiency. Copyright 1987 by MIT Press.
This study compares the accuracy and information content of economic forecasts for G7 countries made in the 1990s by the OECD and IMF. The benchmarks for comparison are the average forecasts of private sector economists published by Consensus Economics. With few exceptions, the private sector forecasts are less biased and more accurate in terms of mean absolute error and root mean square error. Formal tests show these differences are statistically significant for forecasts of real growth and production, less so for forecasts of inflation and unemployment. Overall, there appears little information in the OECD and IMF forecasts that could be used to reduce significantly the error in the private sector forecasts. Copyright 2001 by Taylor and Francis Group
The answer to this question depends on the treatment of logically and empirically prior questions about (1) what the forecasts are and why they are needed, and (2) what can reasonably be expected of them. Further, what forecasters can and should do cannot be established without studying the record and assessing the probable future of their endeavors. Accordingly, the basic approach taken in this paper is to ask of the assembled data what professional standards have economists engaged in macro-forecasting been able to attain and maintain in competing with each other and alternative methods. There is much disenchantment with economic forecasting. The difficult question is how much of it is due to unacceptably poor performance and how much to unrealistically high prior expectations. My argument is that the latter is a major factor. In times of continuing expansion with restrained inflation, as in the 1960s, macro-forecasts looked good and economists were held in high repute. Later when inflation accelerated, serious recessions reappeared, and long-term growth of productivity and total output slackened, the errors of macroeconomic models and forecasts, and the old and new controversies among the economists, received increased public attention. The reputation of the profession suffered, and the interest of academic economists in forecasting, never very strong, weakened still more. Yet the performance of professional economic forecasters, when assessed proper relative terms, has been considerably better in recent times than in the earlier post-World War II period. What happened is that the improvements fell short of enabling the forecasters to cope with the new problems they faced.
Have macroeconomic forecasts grown more or less accurate over time? This paper assembles, examines, and interprets evidence bearing on this question. Contrary to some critics, there are no indications that U.S. forecasts have grown systematically worse, that is, less accurate, more biased, or both. Neither do any definite trends in a positive direction emerge from comparisons of annual and quarterly multiperiod forecasts and time-series projections for the principal aggregative variables. The argument is developed and to some extent documented that major failures of forecasting are related to the incidence of slowdowns and contractions in general economic activity. Not only the forecasts of real GNP growth and unemployment but also those of nominal GNP growth and inflation often go seriously wrong when such setbacks occur. Forecasters tend to rely heavily on the persistence of trends in spending, output, and the price level. More attention to data and techniques that are sensitive to business cycle movements and turning points could help improve their record.
This paper analyzes the short-term forecasts for industrial and developing countries produced by the International Monetary Fund, and published twice a year in the World Economic Outlook (WEO). For the industrial country group, the WEO forecasts for output growth and inflation are satisfactory and pass most conventional tests in forecasting economic developments, although forecast accuracy has not improved over time, and predicting the turning points of the business cycle remains a weakness. For the developing countries, the task of forecasting movements in economic activity is even more difficult and the conventional measures of forecast accuracy are less satisfactory than for the industrial countries.
This paper examines the forecasts that were prepared prior to and during the early stages of the recession that occurred in 1990 in the United States. It examines the characteristics of those forecasts, the data that were available and attempts to determine why the forecast errors occurred. Private sector and public sector predictions are compared and the possibility of rational forecast bias is investigated. We conclude that data problems might have contributed to the forecast errors and suggest that individuals might have been able to predict this recession.
This paper looks at unobserved components models and examines the implied weighting patterns for signal extraction. There are four main themes. The first concerns the implications of correlated disturbances driving the components, especially those cases in which the correlation is perfect. The second is about the way in which ARIMA-based methods for trend extraction relate to those based on unobserved components. The third explores the impact of heteroscedasticity and irregular spacing and shows how setting up models with t -distributed disturbances leads to weighting patterns which are robust to outliers and breaks. Finally, a comparison is made between implied weighting patterns with kernels used in non-parametric trend estimation and equivalent kernels used in spline smoothing. It is demonstrated that with irregularly spaced data, the weighting used by conventional spline smoothing techniques is not the same as that obtained from the time series model based approach.
An Assessment of Macroeconomic Forecasts in the International Monetary Fund's World Economic Outlook
- Peter Kenen
- Stephen B Schwartz
Kenen, Peter and Stephen B. Schwartz, "An Assessment of Macroeconomic Forecasts in the
International Monetary Fund's World Economic Outlook," Princeton University,
Working Papers in International Economics G-86-04, December 1986.
How Reliable are IMF Economic Forecasts?
- William W Beach
- B Aaron
- Isabel M Schavey
- Isidro
Beach, William W., Aaron B. Schavey, and Isabel M. Isidro, "How Reliable are IMF
Economic Forecasts?," Heritage Foundation Report 99-05, August 27, 1999.
Reserve Bank Forecasting: Should We Feel Guilty? An address to The New Zealand Society of Actuaries
- Donald T Brash
Brash, Donald T., 1998, "Reserve Bank Forecasting: Should We Feel Guilty?," An address to
The New Zealand Society of Actuaries
(http://www.rbnz.govt.nz/speeches/sp981021.htm).
A Load of Crystal Balls
- Jim Smalhout
Smalhout, Jim, "A Load of Crystal Balls," Euromoney, February 2000, Issue 370, 14-16.