untitled - imf elibrary

IMF Working Paper

© 1996 International Monetary Fund

This is a Working Paper and the authors) would welcomeany comments on the present text. Citations should referto a Working Paper of the International Monetary FundThe views expressed are those of the authors) and do notnecessarily represent those of the Fund.

WP/96/89 INTERNATIONAL MONETARY FUND

Research Department

How Accurate are the IMF's Short-Term Forecasts?Another Examination of the World Economic Outlook

Prepared by Michael J. Artis1

Authorized for distribution by Flemming Larsen

August 1996

Abstract

This paper analyzes the short-term forecasts for industrial and developing countriesproduced by the International Monetary Fund, and published twice a year in the World EconomicOutlook (WEO). For the industrial country group, the WEO forecasts for output growth andinflation 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 thebusiness cycle remains a weakness. For the developing countries, the task of forecastingmovements in economic activity is even more difficult and the conventional measures of forecastaccuracy are less satisfactory than for the industrial countries.

JEL Classification Numbers:E17,E37,F17,F47

Professor, European University Institute, Florence. The author is grateful for commentsand suggestions on an earlier draft by participants in a Research Department seminar. Hewould like to acknowledge the help of many, most notably including Flemming Larsen,who formulated many of the questions that are addressed in the study, Paula De Masiwho helped prepare the paper for publication, and Wenda Zhang who supplied researchassistance.

©International Monetary Fund. Not for Redistribution

• •-11 -

Contents

Summary

I. Introduction

II. Basic Definitions and Methods of Evaluation

III. Industrial Countries

1. Basic facts2. Further summary results3. Efficiency4. World variables5. MSE regression tests6. The WEO forecasts over time7. Directional accuracy8. Forecasting the cycle9. A comparison with private sector forecasts

10. Generality of forecast errors

IV. The Developing Countries

V. Conclusions

Text Tables

1. Test for Biasedness and Serial Correlation of Forecast Error in Industrial Countries2. WEO Forecast Accuracy: Real GDP Growth in Industrial Countries3 WEO Forecast Accuracy: Inflation in Industrial Countries4. WEO Forecast Accuracy: Balances of Payments on Current Account in Industrial

Countries5. WEO Forecast Accuracy: Growth of Export Volumes in Industrial Countries6. WEO Forecast Accuracy: Growth of Import Volumes in Industrial Countries7. WEO Forecast Accuracy: World Trade Volumes and Terms of Trade8. MSE Regression Test: Current Year Forecast9. MSE Regression Test: Year Ahead Forecast

10. A Comparison of Two Subperiods: Current Year Forecasts11. A Comparison of Two Subperiods: Year Ahead Forecasts12. 2x2 Contingency Table of Directional Forecast Accuracy: Current Year Forecasts13. 2x2 Contingency Table of Directional Forecast Accuracy: Year Ahead Forecasts14. Forecasts Made at Different Time Horizons15. Turning Point Errors: Systematic Under and Overestimation in Output Growth Forecasts16. Consensus Forecasts Through the Cycle17, Turning Point Errors: Systematic Under and Overstimation in Output Growth Forecasts1 8 . Cross-Correlation of Current Year Forecast Errors19. Cross-Correlation of Year Ahead Forecast Errors

Page

iv

1

2

5

68

1415151922252831

34

43

69

10

11121316171820212324262729303233


-iii-

Page

Text Tables (Concluded)

20.21.22.23.24.

25.26.27.

Charts

1.

2.

3.

4.

5.6.

7.

8.

9.

10.11.

12.

13.

14.

Cross-Correlation of Forecast Errors Between GDP Growth and InflationTest for Biasedness and Serial Correlation of Forecast Error in Developing CountriesWHO Forecast Accuracy: Real GDP Growth in Developing CountriesWEO Forecast Accuracy: Consumer Prices in Developing CountriesWEO Forecast Accuracy: Balances of Payments on Current Account in Developing

CountriesWEO Forecast Accuracy: Growth of Export Volumes in Developing CountriesWEO Forecast Accuracy: Growth of Import Volumes in Developing CountriesWEO Forecast Accuracy: Nonfuel Commodity Prices

WEO Forecast: Real GDP Growth in Industrial Countries, Current Year Forecastand First Available Out-Turn

WEO Forecast: Real GDP Growth in Industrial Countries, Year Ahead Forecastand First Settled Estimate

WEO Forecast: Inflation in Industrial Countries^ Current Year Forecast andFirst Available Out*Turn

WEO Forecast: Inflation in Industrial Countries, Year Ahead Forecast andFirst Settled Estimate

Forecasts Made at Different Time HorizonComparative WEO and Consensus Forecasts Prediction Errors, Output Growth:

Current Year ForecastsComparative WEO and Consensus Forecasts Prediction Errors, Output Growth:

Year Ahead ForecastsComparative WEO and Consensus Forecasts Prediction Errors, CPI Inflation:

Current Year ForecastsComparative WEO and Consensus Forecasts Prediction Errors, CPI Inflation:

Year Ahead ForecastsForecast Errors in GDP Growth and Inflation, Current Year ForecastWEO Forecast: Real GDP Growth in Developing Countries, Current Year

Forecast and First Available Out-TurnWEO Forecast: Real GDP Growth in Developing Countries, Year Ahead Forecast

and First Settled EstimateWEO Forecast: Inflation in Developing Countries, Current Year Forecast and

First Available Out-TurnWEO Forecast: Inflation in Developing Countries, Year Ahead Forecast and

First Settled Estimate

Appendix Tables

References

35363738

39404144

6a

6b

6c

6d26a

30a

30b

30c

30d34a

34b

34c

34d

34e

46-72

73

Contents


- iv -

Summary

This paper analyzes the short-term forecasts for the industrial and developing countries,produced by the International Monetary Fund and published twice a year in the WorldEconomic Outlook.

For the industrial country group as a whole, World Economic Outlook forecasts foroutput have on average over- or under-predicted growth by about 1 percentage point, and havegenerally been unbiased, serially uncorrelated, and efficient. Inflation has been over- or under-predicted on average by about % of 1 percent a year; although these forecasts have beenunbiased and efficient, they appear to suffer from serial correlation.

Forecasting output and inflation for the industrial countries has not become any easierover time. An analysis of two subsamples of the forecast errors (pre- and post-1983) showsthat there is not a great deal of difference in the accuracy of forecasts between the two periodsrelative to forecasts based on "naive" methods, such as the assumption of a random walk. Thestudy also confirms that the greatest area of weakness in forecasts for the industrial countries ispredicting turning points in the business cycle. In addition, a comparison with the private sectorConsensus Forecast reveals that over the last business cycle (1990-94), the forecasting errorsfor output growth and inflation were generally about the same.

For the developing country group, forecasting movements in economic activity is evenmore difficult. Many of these economies experience relatively greater volatility; and the data onwhich the forecasts are based tend to be poor in quality and lack timeliness. The averageforecast errors for the period 1977-94 for output growth and inflation differ considerably acrossregions of the developing world, but are relatively large in comparison with their actual averageabsolute values. Overall, by comparison with the results for the industrial countries, theseresults suggest that it is much more difficult to forecast both output and inflation for thedeveloping countries.


I. Introduction

This paper reports the results of an examination of the short term forecasts producedby the International Monetary Fund and published twice a year in its World EconomicOutlook.

The forecasts concerned are comprehensive in their coverage, both of countries and ofeconomic variables and only a part of the whole is examined here. The evaluation is thusdirected at the accuracy of short term forecasts for key economic variables for the principalindustrialized ("G-7") countries and for regional aggregates of developing countries. Thisconcentration on the value of the forecasts contained in the World Economic Outlook (WEO)follows the precedent of an earlier examination by the present author (Artis, 1988), whichitself built on a previous analysis by Kenen and Schwarz (1986), and was subsequentlyupdated and supplemented by Barrionuevo (1993).

Two cautionary notes are enjoined by this concentration on forecast post-mortemanalysis. First, for many commentators the principal value of the WEO may lie in its analysisof the conjuncture, its diagnosis of the situation reached by the world economy and itsevaluation of the options available to the world's policymakers—rather than in the fine detailof its short-run forecasts. Second, from the perspective of strengthening global economicpolicy making and performance in the longer run, the IMF's medium-term projections andscenario analyses are arguably more relevant than the short-term forecasts. However, it mustremain true that the quality of the IMF's analysis should be reflected in its forecast of thenear-term evolution of the world economy and, as these forecasts are reported withconsiderable precision and detail, they offer the most accessible and feasible means ofbringing quantitative analysis to bear on the quality of the IMF's conjunctural analysis.

The choice of evaluation methods employed is dictated in part by the forecastingmethods employed by the IMF. These forecasts are not produced in the framework of anoverall econometric model, so that forecast post mortem methods applicable to model-basedforecasting (see e.g. Osborn and Teal, 1979; Artis 1982; Wallis et al. 1984) are notappropriate. Rather than relying upon a global model (and forecaster intervention on themodel) to produce forecasts, IMF procedures rely heavily on the provision of forecastinformation from individual country desk officers. This optimizes on the availability ofcountry-specific information at the desks. Overall economic consistency is provided in twostages - first, via the setting of common global assumptions to which the country desks workand, second, via the aggregation and resultant check for consistency of the individual countryoutput, trade and balance of payments projections provided by the country desks.Inconsistencies revealed by the aggregation result in iterations on the original countryforecasts until an acceptable set of forecasts is arrived at. The global assumptions specifiedto the country desk officers in a WEO forecasting round will typically include the values tobe assumed for oil prices and assumptions to be made regarding key monetary and fiscalpolicy variables and sensitive market variables such as exchange rates. In general, policy


- 2 -

variables are taken to be given at current values or at publicly projected values if firmcommitments have been made by the governments concerned. Thus, in principle, and likemost official forecasts, those in the WEO are formally presented as projections based on"unchanged policy" assumptions; however, it is certainly difficult for forecasters to maintainsuch an assumption strictly, for much of the market information "in the air" at any point oftime (including such relevant indicators as interest rates, exchange rates, and businessexpectations) will reflect, inter alia, anticipation that the values of policy variables may bechanged in the future. Such anticipations will also, of course, be reflected in forward-lookingmarket variables. It can be argued, in fact, that much the greater part of any genuine policyinnovations will in general not be felt until some time after the horizon of the forecast. Forthese reasons the general practice of treating "unchanged policy" projections as "total" or"unconditional" forecasts is followed in this study.1

The plan of the paper is as follows. In the next section, we lay out the principaldefinitions of forecast and out-turn used in the study and comment on the selection ofvariables to be examined. In the same section we go on to discuss the evaluation methods tobe used. The main results are then presented and discussed in the following two sections -the first dealing with the industrial and the second with the developing group of countries.There is a fined section of conclusions. Appendix A lays out the sources for forecast andout-turn data in detail and provides a statistical characterization of the corresponding datadistributions. Appendices B and C provide the full data listing used.

II. Basic Definitions and Methods of Evaluation

As in the previous studies mentioned earlier we employ two definitions of forecasthorizon with corresponding out-turn in this study. The "paradigm" WEO timetable providesfor publication twice a year, in May and October; the forecasts themselves are finalized inApril and September. Correspondingly, we define as a "current year forecast" the forecastfor year x appearing in the May issue of the WEO for year x. The out-turn data, which wedescribe as "first available estimates" are taken from the issue of the WEO appearing in Mayof year x + L Thus, the "current year" forecast corresponds to a near-term forecast, made ata time when some data for the first quarter of the year in question are already on hand formost, though not all, countries; and the realization for the year as a whole is identified withthe data available in the first publication of the following year. We then also define a "yearahead" forecast, which is of longer term. Thus the "year ahead" forecast for year x is foundin the issue of the WEO for October of year x - 7; the realization for this forecast is identified

*In Artis (1988), an attempt was made to explain forecast errors by relating those errors todeviations in policy and environmental variables from the values set for the forecast.This was quite a difficult procedure and produced no positive results that were notalready obvious. It has not been repeated in this study.


- 3 -

with the data published in the issue of the WEO for October of year x + 1. These data aretermed "first settled estimates".

These definitions were first suggested by Kenen and Schwarz (1986) and wereemployed in our earlier study. They provide for a test of the sensitivity of the forecast to itshorizon. There is no clearly agreed definition of the "correct" vintage of realization data toemploy. These data are continuously revised and the forecast post mortems are dependent, indetail, on the choice of data vintage made. A more common practice than that employed hereis to use the latest available data—a mixed set of revision vintages. This reflects anunderstanding of the objective of forecasts which is that they aim to "forecast the truth"whilst the nearest to the revealed truth on hand at any time is the latest available set of data.2

But this may not be the way in which the forecasts are evaluated by their constituency, wherea higher premium on immediate predictive accuracy may be found. It is arguable thatconfronting the forecaster with the latest available set of realizations obliges him to forecastthe data revision process as well as to predict the immediate evolution of the data he hasavailable. In practice, there is probably little of general significance in the results thatdepends on the vintage of realization data employed.3 The definitions of forecast andout-turn given above apply to "paradigm" WEO publication schedules. In practice, andespecially in the period before the forecasts were made public, the intervals betweenreporting are sometimes erratic and the interpretation of "current year" and "year ahead"forecasts with their associated out-turn has to be adjusted correspondingly. Appendix TableAl lists in detail the precise sources of forecast and out-turn data.

The WEO forecasts are rich in detail. It would be excessively burdensome to processall the series for which forecasts are made. Those that we concentrate upon here are theprojections for GDP, inflation, the balance of payments and the growth of imports andexports. These choices coincide with those made in the previous study. It is forecasts for theindustrial countries group—specifically the individual G-7 member countries—which are themost detailed and the larger part of the study is devoted to them. When the record for thedeveloping countries is examined, the analysis is confined to regional aggregates.

2 Leitch and Tanner (1995) quote the judgement of McNees and Ries (1983) to the effectthat "it is crucial to use the most accurate estimate of the actual data in order to avoidpenalizing the best prediction of what actually happened as opposed to the best predictionof what initially was mistakenly thought to have happened."

3 In Artis (1988) the principal calculations were all replicated on latest available data inorder to check the sensitivity of the general results to the choice of realization data.Whilst the results were somewhat weaker, no significant qualitative difference wasdiscovered.


-4-

The study examines the whole WEQ forecasting record from its inception in 1971 to1994. The length of the series now available enables us also to examine whether anysignificant change has occurred in the IMF's record over time, particularly in the intervalsince the previous study.

The literature is replete with a large number of suggested forecasting evaluationtechniques (see Wallis (1989) for a survey). Rationality considerations suggest that a "good"(rational) forecast should produce errors which are unbiased and display an absence of serialcorrelation: evidence to the contrary would suggest immediately that an improving correctioncould be made to the forecast process. In addition, it ought not to be possible to show thatthe forecast errors could be explained (hence potentially reduced) by taking account of anyinformation available at the time the forecast was made (such as, for example, informationprovided by alternative forecasting procedures). The first two desiderata of a rationalforecast can be tested for directly by applying the appropriate econometric procedures to theseries of forecast errors. To test for the efficiency of the forecast procedure in the broadersense involves the evaluator in determining what might be critical information and testing tosee whether indeed forecast error can be explained by it. An immediate difficulty is that theset of possibly relevant information is huge. Evaluators have generally concentrated on aneasily available subset, stressing in particular, the possible relevance of the forecast valuesthemselves, and the forecasts that could have been produced by alternative naive - or not sonaive - time series forecasts. The first set of information is exploited exclusively by the"realization-forecast" regression introduced by Mincer and Zarnowitz (1969); this regressionhas the attractive property that it is clear what parameter restrictions would correspond to theperfect forecast. Results for this regression featured extensively in our previous study and doso again in the current one.

Forecast evaluation traditionally looks to some alternative forecasting procedure toprovide a benchmark against which to appraise the performance of the procedures underexamination. One set of alternatives is provided by simple time series models. Traditionally,the potential contribution of alternative, naive models has been filtered through the Theil(1966) statistic which is computed as the ratio of the RMSE of the forecast in question to theRMSE of the naive alternative (in Theirs original exposition the 'no change* forecast). Inpractice the naive alternative may be represented by a "not-so-naive" model, such as, forexample, a BVAR (see Artis and Zhang 1990 for such an application). In this study wepresent Theil statistic computations both for the original naive interpretation and for a lessnaive alternative based on a knowledge of the trend. However, these computations simplyprovide point estimates without any accompanying significance level. A more recentextension of this form of testing against an alternative has been formulated so as to providesignificance tests. Tests of this type, in the form of the "MSB regression", are alsointroduced in the present study.

On the occasion of the previous study (Artis 1988), extensive comparisons were madebetween the forecasts produced in the WEO with those produced by the OECD, and byindividual national official forecasters. This extensive comparison of official forecasts was


- 5 -

notable chiefly for the finding that the major forecasting errors were widely shared across theofficial forecasting community and for emphasizing the importance of timeliness to goodforecasting.4 This time, we have sought to make a comparison with private sector forecasts.The extent to which it is possible to do this is limited, however, since the ConsensusForecasts which are used are only available from the latter part of 1989. The comparison isthus confined to the part of the study which investigates the forecast record through the lastcycle.

In addition to testing the quantitative forecast it is well-recognized that an addeddimension of a forecast is the directional information it contains. Leitch and Tanner (1991,1995) have shown that accurate directional information is important for business users offorecasts; for policy makers, correctly predicting the turning point in the business cycle isalso of separate and significant importance to quantitative accuracy. For this reason we alsoinclude tests of directional accuracy and discuss some aspects of turning point forecasting inthe latest business cycle.

Finally, we also examine how general the prediction errors are across the economiesof the world. Interdependence between economies might be expected to result in asynchronization of the business cycle, leading individual national forecasters to commitforecasting errors of similar sign. This indeed was a finding of the earlier study. The IMF,by reason of its position, should in principle be better placed to 'internalize' internationalinterdependence in its forecasting procedures.

III. Industrial Countries

1. Basic facts

The summary table (Table 1) and the four Figures (Charts 1-4) give an immediateimpression of the quality of the IMF forecasts for output growth and inflation, both for the"current year" and the "year ahead" forecasts. Table 1 provides evidence on the questions ofbias and persistence.

Bias may be identified with the significance of the mean forecast error, as indicatedby a simple regression of the error on a constant term (see Holden and Peel, 1990). In the

4 The study employed a form of encompassing test with respect to the OECD forecasts.Thus the question was posed whether OECD forecasts could explain IMF forecast errors(the question was also posed in reverse). Almost no evidence was discovered then thatIMF forecast error could be explained by OECD forecasts (the reverse question was alsoanswered in the negative).


- 6 -

Table 1. Test for Biasedness and Serial Correlation of Forecast Error in Industrial Countries

Current year (1971-1994)GDP growth

PoSignificance levelInflation

PoSignificance levelYear ahead (1973-1994)GDP growth


PoSignificance level

Current year (1971-1994)GDP growthSignificance Ievel-Q(l)Significance level-Q(2)Significance level-Q(3)InflationSignificance level-Q(l)Significance Jeve]-Q(2)Significance level-Q(3)Year ahead (1973-1994)GDP growthSignificance level-Q(l)Significance level-Q(2)Significance leve!-Q(3)InflationSignificance level-Q(l)Significance Ievel-Q(2)Significance level-Q(3)

UnitedStates Japan Germany France

UnitedItaly Kingdom Canada

Test for biasedness

.08

.67

.01

.92

.27.49

-.05.86

.10

.76

.47

.26

.76.20

.44

.54

,18.51

.04

.80

.60.17

.02,88

,09.71

-.2625

.52

.15

-.50.14

.13

.65

-.54.11

.33.47

.15

.54

-.4624

.4521

-.81 -1.01.19 .07

21.42

-.14.64

.79.07

-.38.44

Test for serial correlation (Ljung-Box Q-statistic)

.28

.10

.18

.49,41.55

.64

.74

.88

.08

.22

.35

.81

.71

.30

.42

.03

.06

.43

.63

.81

.78

.53

.73

.11

.12

.22

.46

.33

.53

.92

.98

.99

.57

.75

.53

.46

.75

.50

.852636

.59

.85

.86

.02

.06

.13

.78

.31

.45

.39

.42

.58

.53

.25

.43

.06

.15

.05

.76

.60

.59

.03

.00

.00

.04

.11

.14

.11

.25

.32

.33

.46

.46

.102035

.56

.43

.63

.03

.09

.16

Notes: The test for biasedness f s based on the regression expressed as e^fr,^ where et is the forecast error, andthe significance level of the t-statistic for P0=0 is reported. The Ljung-Box Q statistic is used to measure serialcorrelation and the Q statistic up to M lags may be expressed as Q(M) = T(T+2) S jiM fy2 /(T-j). Under a nullhypothesis of no serial correlation, Q is asymptotically distributed as a x2.


- 6a -

Chart 1. World Economic Outlook Forecast: Real GDP Growth in Industrial Countries

Current Year Forecast and First Available Out-Turn


- 6b -

Chart 2. World Economic Outlook Forecast: Real GDP Growth in Industrial Countries

Year Ahead Forecast and First Settled Estimate


- 6c -

Chart 3. World Economic Outlook Forecast: Inflation in Industrial Countries

Current Year Forecast and First Available Out-Tarn


- 6d -

Chart 4. World Economic Outlook Forecast: Inflation in Industrial Countries



- 7 -

table, the value of the mean forecast error, p0, is shown both for output growth and forinflation for each member country of the G-7 for each of the types of forecast distinguished.In parentheses are shown the significance levels or probability values at which the null (meanequal to zero) might be rejected. As indicated, these values are generally far in excess of thesignificance levels which it is customary to employ in this type of situation (0.01, 0.05 or0.10). Generally, then, the evidence is that these forecasts are not, on a country-by-countrybasis, biased. This evidence seems especially strong for the current year forecasts of outputgrowth - stronger than for the corresponding year ahead forecasts, for example, and a similaraccount holds true for the inflation forecasts. It is worth noting, however, that thequalification 'country-by-country' may be a little misleading. The fact is that all of the pointestimates of bias in the GDP growth rate forecasts are positive - suggesting that there may bea widespread error of output growth optimism. Indeed, when the individual countryobservations are pooled, the result is a finding that there is significant positive bias in theyear ahead forecasts of just over 0.5 percent per year; but when the period is divided into two(the first sub-period terminating in 1982), it appears that this bias is overwhelmingly due toexperience in the first sub-period; bias is not significant in the later period. For the currentyear forecasts the pooling did not reveal any significant bias for the period as a whole (aproduct of some positive bias in the first sub-period and some negative bias in the second).5

Serial correlation in the time series of the forecast errors itself is tested by theLjung-Box Q-statistic, significance levels for the null (no serial correlation) being shown inparentheses. Test statistics are reported for up to three orders of autocorrelation. Theforecasts for inflation appear to suffer from serial correlation in the errors far more than theoutput growth forecasts do. In the current-year forecasts for inflation, serial correlation isdetected for both Japan and the U.K.; in the year ahead forecasts serial correlation affects theerrors for France, Italy, and Canada. In the corresponding chart (Chart 4), it seems clear thatserial correlation affects the errors for the G-7 as a whole. By contrast, the output growthforecasts are almost entirely free of serial correlation in the errors, even at the 10 percentlevel, with the single exception of the year ahead forecasts for the UK, where serialcorrelation of the first order is detectable.

The overall conclusion is that on a country-by-country basis, looking at the period asa whole, there is little evidence of bias in the forecasts; when the data are pooled, whereevidence of significant bias emerges, this is due entirely to early experience. The record inrespect of an absence of serial correlation is somewhat less reassuring, especially in relationto the longer-term forecasts of inflation. The rather more favorable impression given by thecurrent year forecasts than by the year-ahead projection is borne out by the graphicalevidence of Charts 1-4. Charts 1 and 3 for the current year forecasts give a very strongimpression that these forecasts are highly accurate and that errors are soon canceled.

5Barrionuevo (1993) came to similar conclusions. In the earlier study (Artis, 1998), wenoted that the bias finding in the earlier period was itself essentially due to particularlylarge errors in 1974.


-8 -

Charts 2 and 4, for the year-ahead forecasts, indicate much greater variability in the accuracyof the projections.

2. Further summary results

Further summary statistics are reported in Tables 2-6. These report the mean(average) absolute error, the mean absolute actual value (for comparison), the root meansquare forecast error (RMSE) and two Theil statistics. Each of these is constructed as theratio of the RMSE of the WEO forecast to the RMSE of a 'naive' alternative. 'Naive 1' issimply the original Theil 'no change' forecast (here meaning "the same rate of growth(inflation, etc.) as last year") and 'Naive 2' which is a value equal to the trend. Whilst Naive1 corresponds to a random walk with no drift, Naive 2 is the opposite extreme of instantmean reversion. By construction, values of the Theil statistics in excess of unity indicate thatthe WEO forecast is inferior to a forecast built on one of these two alternative extremeassumptions.6

For output growth forecasts and for inflation forecasts the statistics reported in Tables2 and 3 support two general propositions: first, these WEO forecasts are superior to the naivealternatives posed; second, the performance of the current year forecasts is notably betterthan that of the year ahead forecasts: RMSEs are some 50 percent bigger in the year aheadforecasts than in the case of the current year forecasts; and the size of the mean absolute erroris also generally larger by a similar margin.

The balance of payments forecasts (Table 4) are much less satisfactory. Whilstcurrent year forecasts are again generally superior to the year ahead projections, even in theformer case the Theil statistic exceeds unity in the case of Canada: in the year-ahead forecaststhose for Italy and for all industrial countries exceed unity (Naive 1) or are very close to unity(Naive 2).

Tables 5 and 6 report results for export and import growth. These are comparablewith those obtained for output growth.

The summary statistics clearly support the propositions that current year forecasts arebetter than the longer term year-ahead projections; and that the balance of payments forecastsare markedly weaker than those for output growth, inflation and the growth of export andimport volumes. These findings are much in line with those arrived at in the earlier study ona smaller data set, as will be amplified further below.

6Despite the typification of the random walk assumption as 'naive', it is worth recallingthat in some circumstances a random walk is about the best forecast assumptionavailable-e,g., this is so for a variety of asset prices, exchange rates etc.


- 9 -

Table 2. World Economic Outlook Forecast Accuracy: Real GDP Growth in Industrial Countries

(In percent)

UnitedStates Japan Germany France Italy

UnitedKingdom Canada

AllIndustrial

SummitSeven Europe

Current year (197 1-1994)Mean absolute actual valueAverage absolute errorRMSEThcil statistic

Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

3.16.77.94

.29

.34

.10(.72).93

(.38).62.86

2.33

4.421.141.53

.43

.45

.35(-58).90

(-42).68.68

1.88

2.72.99

1.32

.46

.57

.06(-89).90

(.48).62.62

1.39

2.72.71

1.12

.46

.52

.00(1.00)

,97(-81).91.67

1.70

2.621.011.31

.39

.50

-.53(.29)1.18(32).55.64

L80

2.23.93

1.15

.47

.48

.01(.97).90

(.41).59.73

1.72

3.38.91

1.25

.42

.48

-.26(.57)1.02(.89).72.74

2.34

2.84.60.72

2933

-22(.44)1.03C71).64.86

1.50

2.95.58.72

.2731

-.12(.68)LOO

(LOO).74.87

1.69

236.65.96

.42

.51

-29(-46)1.05(-75).63.68

L52

Year ahead (1973-1994)Mean absolute actual valueAverage absolute errorRMSEThcil statistic

Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

3.171.241.78

.36

.50

-.05(.94).92

(.70).74.51

2.14

4.101.732.75

.46

.59

1.88(-19).44

(-05).07.07

1.76

2.741.542.02

.60

.76

-.92(.49)1.12(-80).38.20

1.96

2.411.181.66

.74

.80

.36(.69).68

(.29).20.17

2.15

2.661.582.06

.61

.74

24(.80).77

(.50).61.16

1.78

2.461.471.84

.50

.62

-.87(31)1.20(.57).46.35

1.08

3.121.692.07

.60

.74

-1.18(.31)1.12(.71).2035

2.21

2.761.041.46

38.50

-.54(.53)1.03(.91)37.42

1.70

2.86L051.49

36.47

-,56(.50)1.03(.89)37,45

1.78

230L171.59

.62

.73

-.41(.71).95

(.89).28.18

1.87

Notes: The regression is expressed as R^Po+p^+jLit, where R< is the realization in year t (first available out-turnor first settled estimate) and Ft is the forecast for year t. Figures in parentheses are the significance level of the t-statistic for p0=0 or P^L The significance level of the F-statistic for the test of the joint hypothesis: P0=0 and pt-l,is reported. T^aive 1" means a no-change forecast and "Naive 2" means a forecast which is set at the trend (averagevalue) for the period.


- 10 -

Table 3. World Economic Outlook Forecast Accuracy: Inflation in Industrial Countries

(In percent)

UnitedStates Japan Germany Italy

United AllFrance Kingdom Canada Industrial

SummitSeven Europe

Current year (1971-1994)Mean absolute actual valueAverage absolute error

RMSETheil statistic

Naive 1Naive 2

Regressionintercept

slope

joint test

R2

D-W

5.11.42.61

.35

.24

-.05(-87)

1.01(-90).99.93

1.74

3.78

1.232.01

.53

.41

-.26(.66).95

C61).48.81

1.64

4.08

.59

.76

.50

.38

-J29(-54)1.06058).83.81

2.10

6.75.77

1.09

.72

.29

.33(-44).98

(-78)

.51

.911.99

11.151.031.65

.59

.29

].J9(.11).94

(.31).17.92

1.62

9.121.391.88

.35

.28

-JO(-89)1.06(39).35.90

2.80

5.831.051.39

.6537

-.55(39)1.12023).43.85

1.52

5.62

.36

.52

.33

.18

,02094)1.00(-96).95.96

1.14

5.47

.42

.58

.36

.20

-,02(.95)1.00(.98)

.98

.95127

6.90

.62

.84

.45

.26

,40(.38)1.00(.94)

.10

.921.45

Year ahead (1973-1994)

Mean absolute actual valueAverage absolute error

RMSE

Theil statisticNaive 1Naive 2

Regressionintercept

slope

joint test

R2

EMV

5.30.96

1.38

.35

.29

-.40(-65)1.09(.58).84.70

1.29

3.672.07

3.28

.66

.44

-.03(.98)

.90(-61).73.50

1.79

3.82.59.70

.39

.38

-.06(-89)1.01(.93).99.80

1.70

7.15

1.201.60

.43

.26

.01(.99)1.07(-44)

.26

.861.08

11.462.15

2.84

.64

.38

1.91(-18).90

(.38)

.29

.741.09

9.141.84

2.62

.36

.34

-.98(.34)

1.25(-03).02.86

1.40

6.021.65

2.26

.53

.37

-.87(-48)1.22(.27)

.40

.651.33

5.72.83

1.30

.3121

-.19081)1.06(.64).80.77

1.14

5.57

.90138

3322

-21(.73)

1.06(-64).87.76

1.19

6.97.94

1.24

.4131

22(.75)

1.05(.62)

.11

.84

.79

Note: For definitions etc., see notes to Table 2.


- 11 -

Table 4. World Economic Outlook Forecast Accuracy: Balances of Payments onCurrent Account in Industrial Countries

(In billions of U.S. dollars)


UnitedItaly Kingdom Canada

AllIndustrial

SummitSeven

Current year (1973-1994)Mean absolute actual valueAverage absolute errorRMSETheil statistic

Naive 1Naive 2

Regressioninterceptslope

joint testR2

D-W

59.5912.8415.91

.47

.25

.50(-92)1.03(-67).87.93

1.37

44.079.33

12.69

.68

.272.71(.48).98

(-73).76.92

1.61

18.406.929.62

.60

.40-1.27(.59).99

C93).82.83

1.10

4.772.773.88

.74

.66-.31(-73).85

(-42).71.50

1.33

7.404.497.01

.74

.74-1.17(.45).69

(-06).17.47

2.27

8.815.047.27

.79

.60-.24(.89).84

029).51.62

1.51

7.953.05428

1.13.53

-.86(.53).98

(-91).73.70

2.58

-

--

-

_

—

---

282314.501721

.63

.67-5.81

(.31).80

(-25).49.52

221Year ahead (1973-1994)

Mean absolute actual valueAverage absolute errorRMSETheil statistic

Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

57.5221.0530.24

.61

.46

-6.03(.50).87

(.26).52.75

1.05

44.2115.5521.73

.70

.44

4.71(.48).97

(-77).75.76

1.30

19.6111.5117.17

.68

.69

.32(-95).85

(-48).71.43.86

4.554.385.57

.79

.98

-.61(-62).40

(.06).15.04

1.14

7.708.13

12.86

.921.24

-2.83(.26).03

(.00).00

-.051.06

9.425.978.61

.75

.63

-.20(.92).84

(36).63.51

1.11

8.262.453.50

.68

.38

-.10(-92)1.19(.08).04.87

1.36

27.9028.6834.64

.951.01

-9.58(.23).44

(.01).02.18

1.39

24.6220.8824.55

.87

.83

.̂89(-38).58

(.01).02.41

1.78



- 12 -

Table 5. World Economic Outlook Forecast Accuracy: Growth of Export Volumes in Industrial Countries

(In percent)


UnitedItaly Kingdom

All SummitCanada Industrial Seven


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

8.712.303.06

36.34

-.30(-72)1.15(.11).18.87

2.59

6.113.904.99

.54

.70

1.46(30).83

(.38).56.44

2.15

6.563.264.10

.44

.57

-2.11C15)1.43(-09).22.61

1.86

5.912.503.13

.46

.54

-.69(.59)1.12(-57).84.54

2.52

6.033.54436

.69

.84

2.45(.18).53

(.10).25.11

2.00

4.652.803.59

.61

.74

1.63(.18).62

(-13)2921

2.11

7.053.614.98

.59

.68

125(.37)1.16(.51).15.51

1.56

5.641.972.51

.43

.54

-.43(.69)1.14(.47).70.61

2.03

5.811.992.59

.43

.53

-.51(-63)1.16(39).60.64

2.17Year ahead (1973-1994)


Naive 1Naive 2

Regressionintercept

slope

joint testR5

D-W

8.673.154.18

.33

.36

-.07C95)125(-09).09.78

2.36

6.325356.25

.59

.90

2.43C23).67

(.30).47.15

2.40

6.094.185.31

.58

.70

-30(.91)1.02(.97).98.12

2.26

5.573.104.08

.69

.85

.69(.76).73

(-48).54.11

2.45

5.914.134.91

.71

.94

4.02(.04)22

(-02).06

-.021.92

4.702.963.48

.58

.73

1.64(-23).62

(.18).40.17

2.13

7.023.835.13

.58

.82

2.06(-28)1.03(.94).1421

2.04

5.692.823.57

.58

.73

.95(-65).82

(.64).89.15

2.21

4.912.653.21

.58

.78

1.51(.47).74

C59).74.10

1.80

Notes: For definitions etc., see notes to Table 2. Year-ahead data for Summit Seven cover the period 1980-94.


- 13 -

Table 6. World Economic Outlook Forecast Accuracy: Growth of Import Volumes in Industrial Countries

(In percent)

UnitedStates Japan Germany France Italy


AllIndustrial

SummitSeven


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

8.834.195.15

.44

.52

.54(.70)1.20(-26).18.67

1.96

8.023.654.48

.39

.46

-2.36CIO)1.27(.10)21.75

2.27

6.853.374.73

.62

.73

-.35(-85)1.03(.92).98.37

1.86

6.512.863.56

.38

.49

-2.04(-36)1.23(.19).41.70

1.63

7.393.975.13

.48

.61

-2.48(.16)1.36CIS).33.54

1.82

6.472.863.45

.49

.57

-.34(.76)1.24C25).35.62

2.22

8.425.085.95

.59

.69

.89(.60)1.28(-32).16.49

1.94

6242.642.97

.40

.50

-1.34C22)1.34(.07).16.72

1.78

6.722.863.16

.41

.50

-1.02(37)128013)26.70

1.93Year ahead (1973-1994)


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

8.844.925.82

.44.58

-1.11(-53)1.51(-06).08.62

2.18

7.254.966.55

.47

.56

-1.64(.50)1.05(.86).66.35

1.73

6.373.954.83

.53

.77

.55(-80).86

(-71).92.18

1.83

6.243.995.22

.64

.79

1.00(.64).78

(-54).83.16

2.12

7.434.616.35

.50

.64

-1.26(-59)1.08(.83).8121

i.88

6203.804.57

.50

.66

-.34(-86)1.13(-73).92.28

1.92

8.165256.42

.60

.78

127(-55)1.15(-70).3721

1.97

6.023.574.40

.56

.72

-122(.66)123(.65)0.900201.93

5.772.993.39

.65

.60

-1.04C61)1.42(33).44.44

1.96

Notes: For definitions e/c., see notes to Table 2. Year-ahead data for Summit Seven cover the period 1980-94.


-14-

3. Efficiency

A test of weak efficiency is represented by the realization-forecast equation

Ri = Po + PiFt + mwhere R^ is the realization, Ft the forecast and ^t an error term.

Since R< = Ft + et where e is the forecast error, the estimate of p] in the equationwould significantly differ from unity if in fact Ft and et are correlated. But if they are, theforecast could be improved. It is in this sense that the realization-forecast equation can bethought of as a weak efficiency test. An efficient forecast would yield an estimate of p, thatis not significantly different from unity, and an estimate of p0 that is not significantlydifferent from zero. Otherwise, again, there would be a simple way of improving theforecast. Since estimates of p0 and p j are generally likely to be correlated, the appropriatetest of whether these desirable restrictions (p0 = 0;pj = 1) hold is a joint one (Wallis, 1989).7Tables 2-6 report estimates of realization-forecast regressions for output growth, inflation,the balance of payments and export and import growth and show the significance level of theF-test for the joint restriction.

The results are reasonably reassuring regarding the efficiency of these WEO forecasts.Certainly, in Table 2 (output growth) the evidence in favor of efficiency is very strong: withthe exception of Japan the significance levels reported exceed the customary value (0.05) bya substantial margin: albeit this margin is bigger for the current year than it is for the yearahead forecasts. The results reported for forecasts of inflation are also generally reassuring:the exception is the year ahead forecast for the UK. Turning to the balance of payments(Table 4), there is again evidence of a much weaker performance. The year ahead balance ofpayments forecasts for Italy, Canada, the G-7 as a group and for individual countries as awhole all fail the weak efficiency test. The forecasts for the growth of exports and importsare on the other hand all highly satisfactory from this point of view.

In summary, the WEO forecasts generally perform well in relation to the test for weakefficiency. It is, however, entirely possible for a forecast to be efficient in this sense, yet tobe poor in some other key respects. A forecast may satisfy the tests for bias and serialcorrelation in its errors and those for weak efficiency without being the minimum varianceforecast and without being good enough for its purpose.

7 This test is sometimes interpreted as a test for bias, but this is misleading. It is true thatif Po= 0»Pi = 1 there wiN be no bias (e will be zero), but it is possible for e to be zeroeven while P0 * and p, # 1. The appropriate test for bias per se is the one reportedearlier; the realization-forecast regression is an efficiency test of a weak type.Barrionuevo (1993) provides an instructive discussion of these issues.


-15-

4. World variables

Table 7 reports the summary statistics and estimates of the realization-forecastregressions discussed above for two key 'world' variables: the growth of world trade andindustrial countries' terms of trade. WEO estimates of world trade are widely used bynational forecasting agencies in their own forecasts in which world variables are 'exogenous'.The evidence of Table 7 is reassuring in this respect, since the data reported strongly supportthe efficiency of the corresponding forecasts and they appear to be superior by a margin tothe two naive alternatives. For the terms of trade forecasts, the results are less reassuring.Whilst superior to naive forecasts in RMSE terms, they are strikingly inefficient.

5. MSE Regression Tests

Ashley et al. (1980) suggest a procedure for examining the statistical significance ofthe difference between the mean square errors of pairs of forecasts. Originating in thecontext of a causality study the test is directly applicable to an evaluation of alternativeforecasts and has been used as such by, among others, Stekler (1991) and Kolb and Stekler(1993). Where, as in these studies and the present one, the alternative forecast is the originalTheil (1966) "naive" random walk model, the test can be regarded as supplying significancelevels in a context in which forecast comparison is otherwise carried out by simple inspectionof the point value of the Theil statistic. In the present case, this supplementary examinationconfirms the handful of particularly weak Theil statistic performances already noted above(Tables 2-6).

The basis for the test is the 'MSE regression'

where 6 is the difference (in our case) between the error of the naive forecast and the error ofthe WEO forecast and a is the sum of these errors (o its mean). The null - in our case thatthe WEO cannot improve on the naive forecast - can be rejected, in the case that both p, andp2 are non-negative, if ajoint F-test for $} = p2 = 0 is satisfied or, either p t or P2 beingnegative (but not significantly so) a t-test on the other coefficient shows it to be notsignificantly different from zero. If either p, or p2 is negative, the null cannot be rejected.

These tests can be shown to be equivalent to appropriate tests on an expression whichdefines the difference in mean square error of each of the two forecasts (see, e.g., Ashley etal. 1980).

The results of this regression test are shown in Tables 8 and 9. Nearly all theforecasts are shown as superior to the naive (in the sense that the naive does not improve onthe WEO forecast). Exceptions arise for the balance of payments forecasts (France andCanada in the case of the current year forecasts; France and Italy in the case of the year-aheadforecasts). It has already been shown that the balance of payments is the most poorly forecast


- 16 -

Table 7. World Economic Outlook Forecast Accuracy: World Trade Volumes and Terms of Trade

(In percent)

World TradeIndustrial Countries*

Terms of Trade World TradeIndustrial Countries'

Terms of TradeCurrent year (1972- 1994)

Mean absolute actual valueAverage absolute errorRMSETheil statistic:

Naive 1Naive 2

Regression!intercept

slope

joint testR2

D-W

5.441.852.17

.40

.48

-.87(-35)1.18(-28).55.71

2.09

2.531.001.35

.25

.26

.37(-10)1.32(.00).00.94

2.23

Year ahead (1973-1994)5.522.993.66

.60

.75

-1 26C66)1.12(.81).75.16

2.04

2.432.133.11

.51

.66

.19(.75)2.69(-01).03.48

1.89

Notes: For definitions ete., see notes to Table 2. Current-year data for industrial countries* terms of trade cover the period1974-1994.


- 17 -

Notes: Figures in parentheses are two-sided significance values of the t-statistic for ${=Q or P2=0."Reject" denotes that the null hypothesis (Pi = P2=0) is rejected at the 5 percent significance level and"no" means no rejection of the null at the 5 percent significance level.

Table 8. MSE Regression

GDP Growth

P,

P*

F-test:Hn:P,=P,=0Inflation

p,p.F-test:H»:P,-P,=0

UnitedStates

-.135(.662).665

(.000)

reject

.096(.483).523

(.000).000reject

Japan

.283(.418).468

(.000).000

reject

-.278(.523).361

(.000)

reject

Germany

-.148(.737).509

(.000)

reject

.143(.482).426

(.001).002

reject

Test: Current Year Forecast

France

.030(.919).447

(.000).000

reject

-.404(-187).242

(.090)

reject

Italy

.065(.817).493

(.000).000

reject

-.648(.115).344

(.002)

reject

UnitedKingdom

-.257(.573).615

(.001)

reject

.717(.222).618

(.000).000

reject

Canada

-.191(.616).524

(.000)

reject

-.274(.109).225

(.000)

rejectBalances of Payments on Current Account

P,

P:

F-test:H.:3,=P,=0

6.062(.059).378

(.000).000reject

3.986(.026).179

(.004).003

reject

-.319(-831).273

(.000)

reject

.490(.504).170

(.052).119no

.738(.466).156

(.020).051

reject

-.700(.434).130

(.025)

reject

.048(.941)-.073(.389)

noGrowth of Export Volumes

P,

P2

F-test:H«: P,=P,=0

-.500(.490).532

(.000)

reject

-.959(.445).362

(-001)

reject

.286(.780).451

(.000).000

reject

.055(.944).450

(.000).000

reject

-.286(-727).213

(.020)

reject

.141(.890).330

(.019).059

reject

-1.586(.165).344

(.001)

rejectGrowth of Import Volumes

P,

P2

F-test:H.:P,=P,=0

-1.245(.401).502

(.000)

reject

-.741(.521).511

(.000)

reject

-.200(.849).279

(.005)

reject

.177(.858).532

(.000).000

reject

-.968(.270).387

(.000)

reject

-.823(.430).454

(.000)

reject

-2.027(.111).323

(.001)

reject


- 18 -

Table 9. MSE Regression Test: Year Ahead Forecast

GDP GrowthP.

P:

F-testH.: 3,-P,-0InflationP.

P:

F-testH,,: P,=P,=0

UnitedStates

.200(.716).609

(.000).000reject

.290(281).601

(.000).000reject

Japan

.135(.763).447

(.000).001

reject

.365(.639).359

(.053).134no

Germany

-.325(.473).313

(.002)

reject

.325(.062).476

(.000).000reject

France

-.105(-739).189

(-037)

reject

.985(.007).455

(.000).000

reject

Italy

.090(.843)285

(-005).017

reject

1340(-019)248

(.017).006

reject

UnitedKingdom

-.345(.344)377

(.000)

reject

-1.560C010).533

(.000)

no

Canada

-.655(-283)374

(.007)

reject

.900C036)338

(.003).003

rejectBalances of Payments on Current Account

P.

P:

F-testH«: P,-P,-0

13.890(.007)244

(.000).000reject

9.045(.001).143

(.006)COOO)

reject

.545(.767).201

(.000).001reject

.690(.510).128

(-147)279no

-260(.853).044

(.404)

no

-1290(.436).163

(.064)

reject

.500(.471)220

(.020).052

rejectGrowth of Export VolumesP,

P>

F-testIfc P,-P,=0

-1.810(.154).627

(.000)

reject

-1.815C254)307

(.006)

reject

-385.764320.004

reject

-.055(.953)223

(.032)

reject

-355(.731)215

(.054)

reject

-.530(.549)320

(.005)

reject

-1.925(.109).444

(.000)

reject

Growth of Import Volumes

Pr

P:

F-test:It: P,-P,-0

-.730(.675).488

(.000)

reject

-.320.838.427.000

reject

-.645.407333.000

reject

1.130278245.006.015

reject

285.806370.000.000

reject

-.590.639.401.001

reject

-1.810287325.006

reject

Note: See notes to Table 8.


-19-

variable and it is for the balance of payments that the Theil statistics ("Naive I11) appearedleast satisfactory. According to the MSE regression test, however, the year ahead forecastsfor inflation (Table 9) for Japan and the UK are also unsatisfactory. Whilst Japan (Table 3)had the highest Theil statistic, that for the UK was quite low: but it may be recalled that therealization-forecast regression for the UK was adverse in this case and, more relevant, thatthis was one of the few cases where bias was shown to be significant.

6. The WEO forecasts over time

The availability of data over a comparatively long period of time as in the full samplelends strength to the statistical verdicts it is possible to deduce from the record. However,interest attaches to the question whether the forecast record has improved over time. At onelevel, this question may be answered by simply inspecting the error statistics and looking forreduced values; this does not allow, however, for the possibility that the economy may havebecome ft easier' to forecast. To allow for this it seems natural to make a comparison with analternative forecast. In addition it might be hoped that bias, if initially present woulddisappear: we have already noted the extent to which this is the case. In addition to the issueof the underlying forecastibility of the economy and summary error statistics, there is also theissue of timely recognition of cyclical turning points, which we deal with below. Here,Tables 10 and 11 provide summary statistics for two sub-samples, where the main sample isapproximately halved by breaking it at 1983. This means that the first sub-sample containsboth of the major oil price increases and the forecasting errors associated with them.Nevertheless, it does not appear to be the case that the subsequent environment provednotably easier to forecast.

There is not a great deal of difference between the current year and year aheadforecasts in respect of their relative performance in the two sub-samples. For output growth,the mean absolute actual value fell nearly everywhere~as did the actual absolute error and theRMSE. Nevertheless, the Theil statistic values (computed for the naive 1 or no changeassumption) tended to rise. This perhaps indicates that, with a less volatile economy, therandom walk forecast itself improves. For inflation, there are quite large declines in thevalues of the mean absolute actual value, with similarly quite large declines in the averageabsolute error and RMSE. The Theil statistic values, however, display little systematicchange.

For the balance of payments, actual absolute values have increased considerably andwith them the error statistics: in this case the Theil statistic values also tend to increaseoverall (more clearly in the current year forecasts than in the year ahead forecasts). Perhapscuriously, it is in respect of the forecasts of export and import volume growth that forecastingerror, judged by reference to the behavior of the Theil statistic values, has risen between thetwo periods most noticeably. The Theil statistics have increased in value in nearly everycase, although average absolute forecasting error is not systematically greater.


• 20 -

Table 10. A Comparison of Two Subperiods: Current Year Forecasts

United

GDP GrowthMean absolute actual value

Actual absolute error

RMSE

Theil statistic

InflationMean absolute actual value

Actuaf absolute error

RMSE

Theil statistic

Period

1971-821983-941971-821983-941971-821983-941971-821983-94

1971-821983-941971-821983-941971-821983-941971-821983-94

Stales

3.362.86

.83

.711.02.8421.33

6.923.52.60.30.80.41.35.42

Japan

5.273.531.38.87

1.891.03

.38

.64

6.421-20l.$iO.S42.77O.G5O.S30.75

Germany

2.842.481.21.88

1.571.13.44.52

5.253.05.69.48.89.59.47.55

France

3.481.94.74.67

1.33.83.43.58

9.484.64U6.47

1.48.62.87.34

Italy

3.082.011.22.94

1.631.12.36.53

14.888.19L55.51

2.22.71.66.35

UnitedKingdom

1.832.471.04.77

1.30.94.43.55

13315.162.13

.762.54

.9334.47

Canada

3.693.20

.981.121.461.53.4536

8.563.672.10.98

1.581.15.65.64

Balances of Payments on Current AccountMean absolute actual value


RMSE

Theil statistic

Growth of Export VolumesMean absolute actual value


RMSE

Theil statistic

Growth of Import VolumesMean absolute actual value


RMSE

Theil statistic

1973-821983-941973-821983-941973-821983-941973-821983-94

1972-821983-941972-821983-941972-821983-941972-821983-94

1972-821983-941972-821983-941972-821983-941972-821983-94

7.4595.736.65

17.247.41

19.78.59.47

8.059.521.952.682.713.37

.30

.41

8.009.203.145.173.936.00

.25

.67

6.7869.89

5.0712.286.50

15.65-59.70

8.184.124.034.025.165.04

.43-87

8,976.603.423.734.114.64

.28

.62

6.5626.34

3.978,715,16

11.67.65,59

8.304.722.953.623.644.49

.32

.61

6.017.092.693.783.395.47

.51

.67

4.955.182.293.283.074.44

.56

.87

7.494.423.002.423.703.07

.45

.48

7.735.222.603.053.033.88

.24,63

5.158.982.625.693.148.70.47.79

6.545.164.352.985.273.50

.62

.97

8.076.404.983.075.844.32

.43

.60

53711323.116304.368.67

.67

.81

4.734.282.772.853343.76

.42

.94

6346.473.102.823.633.43

.40

.71

33311.071.924232.955.31

.94121

6.157292274.572.776.12

.34

.70

7.309.854.346225.406.%

.55

.62

Notes: For definitions etc., see notes to Table 2. The Theil statistic is the ratio of the RMSE of the WEO forecast tothat of the "Naive 1" forecast.


- 21 -

Table 11. A Comparison of Two Subperiods: Year Ahead Forecasts

GDP GrowthMean absolute actual value


RMSE

Theil statistic

InflationMean absolute actual value


RMSE

Theil statistic

Period

1973-821983-941973-821983-941973-821983-941973-821983-94

1973-82] 983-941973-821983-941973-821983-941973-821983-94

UnitedStates

3.143.101351.262.101.58.31.40

7.703.521.36.68

1.85.83.30.56

Japan

4.773.492.061.553.621.81.30.70

6.71J.223381.024.701.17.64.74

Germany

2.822.541.981322.501.68.56.69

4.973.0239.72.55.7721.46

France

2.891.981.161.151.771.51.58

1.11

10.914.6B1.85.64

2.18.83.5026

Italy

3242.042.031.282.441.72.51.61

16.43BJJ3.101.423.741.79.72.51

UnitedKingdom

2.192.56L511371.941.69.42.66

13.995.322.511.263.551.42.34.61

Canada

3.143212.051.722.49225

.84

.43

9.613.582.041242.981.35.56.54

Balances of Payments on Current AccountMean absolute actual value


RMSE

Theil statistic

Growth of Eiport VolumesMean absolute actual value


RMSE

Theil statistic

Growth of Import VolumesMean absolute actual value


RMSE

Theil statistic

1973-821983-941973-821983-941973-821983-941973-821983-94

1973-821983-941973-821983-941973-821983-941973-821983-94

1973-821983-941973-821983-941973-821983-941973-821983-94

7.0092.40

7.6430.04

8.8138.58

.61

.60

8.169323.143.194.244.08

.23

.43

8318.914.545.255.815.80.35.57

6.8470.187.78

20.439.78

26.93.69.71

8.474.425.805.126.845.81

.52

.66

8.485.766353.688.164.66

.41

.70

7.5528.15

6.5315307.79

21.48.72.68

8.164.174.683.826.174.48

.52

.72

5.516.593.893.804.664.79

.51

.54

4.405.034.074.774.796.17

.73

.80

6.874.453.413.154.474.06

.71

.67

7.265.214313.625.794.55

.52

.96

5.139.475.099.976.25

15.81.71.97

6375.125.303326.153.71

.71

.68

8.096.515.963.497.584.99

.40

.84

5.%12.154.306.94536

10.17.73.76

5.044.153.192.814.012.99

.51

.73

5.886.354.733.175.413.85

.48

.63

33311.591393.452.09437

.90

.68

5.747.482.914.343.995.70

.41

.62

7.099.495.055.916.137.16

.74

.46

Note: See notes to Table 10.


-22-

We return to a judgement of the record later below. Provisionally, however, thesummary statistics reviewed here do not afford a basis for a strong verdict either way onwhether forecasting error has fallen over the period. Barrionuevo's (1993) conclusion thatforecast accuracy had improved through the period was based on a data sample whichomitted the most recent downturn and, more significant, did not attempt to control forchanges in the stochastic structure of the world economy and, thus, in the 'ease' or'difficulty' of forecasting.

7. Directional Accuracy

Timely prediction of turning points in the business cycle is of obvious importance tothe forecaster whose predictions are designed to support policy actions. The record of WEOforecasting through the recent cycles is discussed in the next subsection of this paper.Conventional methods of quantitative assessment overlook the significance of directionalaccuracy. The conventional alternative benchmark, the 'naive1 no-change or random walkforecast, for example, makes no effort to predict a direction of change at all. In this sectionwe present estimates of the directional accuracy of WEO forecasts over the whole sampleperiod of the study and offer a non-parametric method of assessment. Tables 12 and 13tabulate information on directional accuracy for forecasts of growth, inflation, the balance ofpayments and the growth of export and import volumes, respectively, for current year andyear ahead definitions.

The directional data can be arranged in a 2x2 contingency table (one for each variableand country) and a simple x2 test applied. Given a table in which two columns are formed forforecasts of positive and negative change and two rows for positive and negative realizations,it is clear that the desideratum is that the sum of entries in the two cells of the leadingdiagonal should be satisfactorily "large". Then, in a high proportion of cases, the signs of thedirection of forecast change and the realization are the same. A formal test of independencecan be supplied in this framework (the classic reference is Yates (1984): the null is thatforecasts and realizations are independent; non-rejection of this hypothesis ("n.r." in thetable) implies that the success rate of directional forecasting is too low; rejection, on the otherhand, implies that there is a significant association between the signs of forecasts andrealizations. In practice, with the values relevant to Tables 12 and 13, an accuracy rate(percentage of correctly signed forecasts) of above 70 percent is required; values below thisregion lead to a verdict of non-rejection at the 5 percent level.8

By this standard, the WEO record in near-term forecasting (the * current year'forecasts) is reassuring. Failures are at the rate of only 1 in 7 for output growth forecasting

8 The significance levels for a small number of observations are relevant in our case; theymay be found in Daniel (1978).


- 23 -

Table 12. 2*2 Contingency Table of Directional Forecast Accuracy Current Year Forecasts

GDP GrowthUnited StatesJapanGermanyFranceItalyUnited KingdomCanadaInflationUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

APX)andAR>0

98778

1110

10895897

AFX) andARsO

1041511

0423004

AFsO andAR>0

3533102

0123223

AFsO andAR<;0

10109

129

1110

1310101213129

Percentage ofcorrect forecasts

.83

.78

.70

.83

.74

.96

.87

1.00.78.83.74.91.91.70

Significancelevel

1%1%n.r.1%5%1%1%

1%1%1%n.r.1%1%n.r.

Balances of Payments on Current AccountUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

6138

11896

2341125

3141123

10458

1187

.76

.81

.62

.91

.91

.81

.62

5%5%n.r.1%1%1%n.r.

Growth of Export VolumesUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

10557897

0532212

1343623

119

10106

1010

.96

.64

.68

.77

.64

.86

.77

5%n.r.n.r.5%n.r.1%5%

Growth of Import VolumesUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

61197974

2122200

3133168

1198

10109

10

.77

.91

.77

.77

.86

.73

.64

5%1%5%5%1%5%n.r.

Notes: F and R denote the current year forecast and the first available out-turn respectively, and AF=F,-Rtp|> AR=Rt-Rt.1.In the last column, "1 percent" indicates that the null hypothesis of independence can be rejected at the 1 percent significancelevel; "5 percent" at the 5 percent significance level and "n.r." indicates no rejection at 5 percent level.


. 24 -

Table 13. 2 X2 Contingency Table of Directional Forecast Accuracy: Year Ahead Forecasts

GDP GrowthUnited StatesJapanGermanyFranceItalyUnited KingdomCanadaInflationUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

AFX) andARX)

7678786

76

104345

AF>0 andARsO

2231322

2611003

AFsO andARX)

2233444

1003542

AFsOandAR*0

91078668

1089

12121210

Percentage ofcorrect forecasts

.80

.80

.70

.80

.65

.70

.70

.85

.70

.95

.80

.75

.80

.75

Significancelevel

5%5%n.r.1%n.r.n.r.n.r.

1%5%1%5%5%5%n.r.

Balances of Payments on Current AccountUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

71289874

1333024

2242422

1035689

10

.85

.75

.65

.75

.80

.80

.70

1%n.r.n.r.5%1%5%n.r.

Growth of Export VolumesUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

7774875

1344132

1202415

1189

1079$

.90,75.80.70.75.80.65

1%5%1%n.r.5%1%n.r.

Growth of Import VolumesUnited StatesJapanGermanyFranceItalyUnited KingdomCanada

8986785

2202211

2144233

888898

11

.80

.85

.80

.70

.50

.80

.80

5%1%1%n.r.n.r.1%5%

Notes: F and R denote the year ahead forecast and the first settled estime respectively, and AF^-R^,AR^Rf-R^. In the last column," 1 percent" indicates that the null hypothesis of independence can be rejected at the1 percent significance level; "5 percent" at the 5 percent significance level and "n.r." indicates no rejection at5 percent level.


-25-

and 2 in 7 for inflation forecasting. The record in longer-term forecasting is less good: 4 in 7country growth rate forecasts fail to maintain direction accuracy at an appropriate level and 2in 7 country inflation forecasts fail the test.

The overall verdict on directional accuracy is therefore somewhat mixed. Very fewof the forecasts are right about the sign of the change less than 50 percent of the time. Butnot enough are turning out with rates of directional accuracy clearly above 70 percent.

8. Forecasting the Cycle

Whilst the statistics of directional accuracy consider the relationship between the signof forecasts and realizations, Table 14 tabulates the WEO forecasts and realizations throughthe most recent cycles. In order to examine the process of recognition of the cycle andcorresponding revision of forecasts the table dispenses with the "current year" and "yearahead" distinction. Instead, it takes the successive forecasts for the out-turn in year x to befound, first, in the April issue of the WEO for year x-1, then in the October issue for year x-1,and subsequently the April and October issues for year x itself. The realization is identifiedwith the data in the WEO for October of year x+1.

Systematic turning point error taking the form of an initial under- or over-estimate ofoutput growth, followed by persistence in the same error with accompanying forecastrevisions in the same direction is uncomfortably pervasive in the data. The tabulation below(Table 15) identifies systematic underestimates as processes where the initial estimate isbelow the final realization and where the process of revision is systematic (with no more thanone change in direction of revision allowed). In addition the table shows the amount of thedifferences between the initial and final figure. Systematic overestimates are defined in asimilar manner. Chart 5 displays the data for the US and Germany.

It is clear from the data that 1988 - a peak year in the growth cycle everywhere exceptin Germany (where unification delayed the peak by two more years) - was a year for whichthe forecast process exhibited systematic underestimation for all the G-7 countries. InGermany and France the process of systematic underestimation was repeated in the followingyear and, for Germany, in 1990 also. Japan appears as a special case in that systematicunderestimation was a feature of the data in every year to 1991. The degree ofunderestimation on the other hand was rarely more than 2 percentage points and morecommonly around 1.5 points. This contrasts with the data shown for systematicoverestimation where figures in excess of 3 percentage points are not uncommon. Where theUS and Canada feature overestimation of the out-turns as early as 1990 and 1991, for theEuropean countries the experience of systematic overestimation appears to set in a little later,to be persistent through 1993 (except for the UK) and to involve some very large errors. Thesame could be said for Japan. The errors are notably smaller for the United States. Thedifference between the US and the other countries may partly be related to the dislocation ofthe cycle in the early 1990s. Synchronicity with the US cycle weakened, with thedevelopment of a European cycle based around Germany, producing a trough in 1993, two


- 26 -

Table 14. Forecasts Made at Different Time Horizons

GDP Growth

Realization1October/current

May2

/currentOctober

/previousMay3

/previous RealizationOctober/current

InflationMay2

/currentOctober

/previousMay2

/previousUnited States1988198919901991199219931994Japan1988198919901991199219931994Germany1988198919901991199219931994France1988198919901991199219931994Italy1988198919901991199219931994

4.42.5LO

-1.22.63.14.1

5.74.95.64.41.30.10.5

3.63.94.53.11.6

-1.72.9

3.53.62.81.21.4

-1.02.9

3.93.22.01.40.9

-0.72.2

4.02.91.3

431.92.73.7

5.84.95.14.52.0

-0.10.9

2.94.03.93.11.4

-2.21.8

2.93.43.11.32.2

-1.01.9

3.03.22.71.31.30.31.5

2.93.11.70.21.63.23.9

4.14.54.43.62.21.30.7

1.72.43.52.81.3

-2.00.5

1.62.83.12.11.80.01.2

2.53.43.01.71.60.31,1

2.72.82.11.73.03.12.6

3.44.24.73.73.43.82,0

2.11.93.03.32.01.90.8

1.82.43.03.02.42.71.1

2.32.42.92.72.51.51,7

3.12.72.52.32.73.53.2

3.33.84.44.23.93.93.5

2.01.72.92.71.92.21.2

2.11.72.83.12.72.62.3

2.32.33.02.92.52.41.9

3.34.14.14.12.92,22.1

0.41.51.91.91.81.00.2

1.52.63.44.54.43.22.3

3.13.52.82.82.32.31.3

6.06.37.57.34.74.43.6

3.24.54.34.02.72.82.3

0.91.81.52.51.71.30.5

1.82.52.93.94.33.82.4

2.63.23.43.12.92.21.6

5.16.66.56.75.63.73,8

3.24.74.13.72.42.62.2

1.21.41.92.62.11.51.2

2.02.52.93.94.33.92.7

2.53.23.33.32.52.01.9

5.06.16.56.15.24.73.5

3.84.14.64.23.72.92.7

1.71.41.32.12.61.91.3

2.22.22.53.63.73.72.8

3.02.22.83.22.92.82.2

535.05.15.75.54.74.2

3.43.54.54.14.02.92.9

2.61.51.21.52.61.91.6

2.62.02.53.03.53.82.3

2.62.52.52.82.72.32.5

5.25.05.04.95.75.24.7

United Kingdom1988198919901991199219931994Canada1988198919901991199219931994

4.22.20.8

-2.2-0,52.03.8

5.03.00.5

-1.70.72.24.6

4.03.01.4

-1.8-0.*1.83.3

4.22.6LI

-0.92.12.64.1

3.03.31.1

-2.10.81.42.5

3.42.91.6

-1.12.33.23.5

2.32.52.71.32A2.12.8

3.23.22.01.13.84.43.8

2.32.22.12.21.93.13.1

3.03.12.53.13.64.94.4

6.76.96.86.94A3.42.1

4.14.93.02.71.11.10.6

5.27.55.56.25.02.02.4

3.95.23.63.41.00.80.5

4.86.65.16.64.42.53.1

4,04.34.04.32.41.11.1

4.84.75.75.74.23.33.9

3.53.64.45.12.72.01.5

5.04.55.86.54.73.14.0

3.23.73.84.72.82.62.0

II From October issue of the WEO in the following year,2/ From April or May issue of the WEO.


- 26a -

Chart 5. Forecasts Made at Different Time Horizon


This page intentionally left blank


- 27 -

Table 15. Turning Point Errors: Systematic Under and Overestimation in Output Growth Forecasts

Underestimation Overestimation

United States

Japan

Germany

France

Italy

United Kingdom

Canada

1988

1988198919901991

198819891990

19881989

1988

1988

1988

(1.3)

(2.4)(1.1)(1.1)(0.2)

d-6)(2.2)(1.6)

(1.4)(1-9)

(1.6)

(1.9)

(2.0)

19901991

199219931994

19921993

199119921993

1990199119921993

19911992

1990199119921993

(1.5)(3.5)

(2-6)(3.8)(3.0)

(0.3)(3.9)

(1-9)(1.3)(3.6)

(1.0)(1.5)(1.6)(3-D

(4.4)(2.4)

(2.0)(4.8)(2.9)(2.7)

NotesrA systematic over (under-) estimation is defined as a process in which a) the initial forecast is above(below) the realization and b) the forecasts display constant downward (upward) revision with no more than onerevision in the opposite direction. The figure in parentheses is the difference between the initial forecast and therealization. The data are from Table 14.


-28-

years behind the US cycle (the UK being an exception) and, independently, a long drawn outdeflation in Japan. These new developments were poorly forecast.

The classification of systematic over- and under-estimation captures a particular typeof turning point error where the forecaster takes on board the evolution of the cycle tooslowly. The forecast process for 1994 reflects a different form of error - one in which, formost countries, an initially quite good forecast is pursued by a lack of confidence, withforecast performance falling away, only to be revived towards the end. The data for inflationreveal for all countries except Germany a pattern of systematic overestimation in 1994 and,to a lesser extent, in the preceding two years. This is particularly marked for Italy and theUK in 1994 and for Japan and Canada in 1992-1994. This may constitute evidence that theforecasters only gradually became convinced about the efficacy of the global policies ofdisinflation set in place since the early to mid-1980s.

9. A comparison with private sector forecasts

The availability of an alternative, private-sector forecast with which to compare theWEO is limited. However, Consensus Forecasts has made available, on a month-by monthbasis, a private sector consensus computed as the simple mean of a number of private sectorforecasts. The literature on optimal forecasts (see e.g. Diebold and Lopez, 1996) shows thatsuch simple means often perform well even compared to "optimal composites". The timeseries available for comparison is, however, relatively brief as the Consensus Forecasts arenot available until late in 1989. Defining Consensus Forecasts for output growth andinflation on the same basis as for the WEO yields just five data points per country for the"current year" forecasts and six data points for the "year ahead" forecasts.9 With so few datapoints it makes little sense to process the Consensus data in the same way as the WEO datahave been processed to this point in this study.

In the circumstances, the comparison is limited to the following exercises. First, inCharts 6-9 we present the WEO and Consensus Forecasts errors in the form of scatterdiagrams, pooling all the country observations. Second, we present the Consensus Forecaststhrough the recent cycle in Tables 16 and 17, which may be directly compared with Tables14 or 15 above for the WEO.

To be specific, current year forecasts are assumed to be those for May in the year inquestion whilst year ahead forecasts are those for October for the following year.Realization data are those used in the WEO evaluation in this study. The Consensusaverages, in any given month, a number of forecasts (how many depending on thecountry), many if not most of which will have been produced in preceding months. It isnot clear precisely how the 'center of gravity' of Consensus Forecasts compares with thatof WEO forecasts: but we think the comparison dates chosen are as fair as they can be.


- 29 -

Table 16. Consensus Forecasts Through the Cycle

GDP Growth

United States19901991199219931994

Japan19901991199219931994

Germany19901991199219931994

France19901991199219931994

Italy19901991199219931994

United Kingdom19901991199219931994

Canada19901991199219931994

Realization

1.0-1.22.63.14.1

5.64.41.30.10.5

4.53.11.6

-1.72,9

2.81.21.4

-1.02.9

2.01.40.9

-0.72.2

0.8-2.2-0.52.03.8

0.5-1.70.7224.6

October/current

0.9-0.31.82.73.7

5.54.22.00.20.7

3.93.01.1

-2.12.4

2.71.32.0

-1.32.1

2.71.11.2

-0.11.8

1.3-2.1-0.91.83,4

1.0-0.81.32.64.0

May/current

2.1-0.52.03.13.6

4.33.42.11.20.6

3.72.51.1

-1.70.8

3.31.71.9

1.5

3.01.41.50.31.5

1.1-1.50.91.52.7

1.41.32.03.23.5

October/previous

1.90.62.62.62.8

4.33.82.92.61.4

3.13.02.01.20.8

3.02.52.31.90.8

3.0232.1121.4

1.81.31.81.52.6

1.50.73.53.33.4

May/previous

2A2.63.03.1

3.93.83.93.1

3^222231.0

3.12.52.62.0

2!92.5221.5

2.*32.1232.5

1.73.23.93.8

Notes: Forecast data are from Consensus Forecasts, realizations from Table 14.


- 30 -

Table 17. Turning Point Errors: Systematic Under and Overestimation in Output Growth Forecasts

United States

Japan

Germany

France

Italy

United Kingdom

Canada

Underestimation

1994(1.0)

1994(0.9)

1994(0.7)

1994(1.3)

1994(0.8)

Overestimation

1991 (3.6)

1992 (2.5)1993 (3.8)1994(2.6)

1992(0.6)1993 (4.0)

1991 (1.9)1992 (1.1)1993 (3.6)

1991 (1.5)1992(1.6)1993 (2.9)1990(4.5)1992 (2.6)

1991 (3.4)1992(2.4)1993(1.7)

Notes: The table is constructed on the same basis as Table 15, with which it may be compared Data are fromTable 16.


- 30a -

Chart 6. Comparative WEO and Consensus Forecasts Prediction Errors

Output Growth: Current Year Forecasts

Source: World Economic Outlook; and Consensus Economics, Inc.1 Forecast errors are for the period 1990-94 and are defined as current year forecast value minus

actual realized value. Each observation shows the Consensus and World Economic Outlookforecast errors for one of the seven major industrial countries for forecasts constructed atapproximately the same time.


- 30b-

Chart 7, Comparative WEO and Consensus Forecasts Prediction Errors

Output Growth: Year Ahead Forecasts

Source: World Economic Outlook; and Consensus Economics, Inc.1 Forecast errors are for the period 1990-94 and are defined as year ahead forecast value minus



- 30c -


CPI Inflation: Current Year Forecasts

Source: World Economic Outlook and Consensus Economics, Inc.1 Forecast errors are for the period 1990-94 and are defined as current year forecast value minus



- 30d -


CPI inflation: Year Ahead Forecasts

Source: World Economic Outlook; and Consensus Economics, Inc.

Forecast errors are for the period 1990-94 and are defined as year-ahead forecast value minusactual realized value. Each observation shows Hie Consensus and World Economic Outlookforecast errors for one of the seven major industrial countries for forecasts constructed atapproximately the same time.


-31-

Turning now to discuss this evidence, Chart 6 shows the scatter of errors in outputgrowth forecasting on a "current year" basis for the period 1990-94; the errors, as should beexpected, are generally quite small and not obviously biased—most, but not all are confined tothe range +A 1 percentage point. The fact that the preponderance of observations falls on andclose to the 45-degree diagonal (not shown in the Chart) indicates that the two forecast errorrecords are very similar. Chart 7 plots the prediction errors for the WEO and ConsensusForecasts "year ahead" output growth forecasts. Here the preponderance of errors fall in thepositive quadrant, indicative of the propensity to overestimate growth in this period: bothforecasts are equally culpable for the bias, however, sharing the errors, even when these arelarge ones. Relative to a 45-degree diagonal there are few marked deviations (those few thatthere are, however, are WEO errors). Charts 8 and 9 provide scatters of the current year andyear ahead prediction error for CPI inflation. Once again, the current year errors appearrelatively small and unbiased; and the observations indicate that there was little differencebetween WEO and Consensus prediction errors. Chart 9 shows a slight positive bias in theforecasting of CPI inflation, with little between the WEO and Consensus error. (The outlierobservation pertains to the forecast for U.K. inflation in 1990, where the common large errorappears to be related to the introduction of the Community Charge ("poll tax").)

The data in Tables 16 and 17 may be directly compared with that in Tables 14 and 15,being constructed for the Consensus Forecasts in the same manner. Bearing in mind that thelimited availability of data removes 1988,1989, and 1990 from the comparison, the moststriking point is the qualitative similarity of the pattern of error. Both Consensus Forecastsand the WEO make the same type of error in the same years for the same country.

10. Generality of forecast errors

How general, across countries, has forecast error been? Tables 18 and 19 provideevidence, in the form of cross correlations of forecast error between countries. The crosscorrelations are, perhaps, lower (for output growth) than might have been expected: thelargest ones-between France and Germany and between the U.S., Canada and Japan-mightreflect the strong trading relationship within these groups of countries. The year-aheadforecast error correlations (Table 19) suggest a stronger role for Germany in respect of otherEuropean countries than the current year forecasts.

The prevalence of negative correlations between inflation forecast errors in thecurrent year forecasts (Table 18) is striking - especially as this is not so marked in theyear-ahead forecasts. Unforeseen exchange rate developments could be a reason for thenegative correlations, but it is not clear why this should not also be a feature of the yearahead forecasts. The prevalence of negative signs on the balance of payments forecasts is ofcourse to be expected on account of the closed nature of the world economy as a whole.Notably, however, nearly all the signs of the export and import growth volume forecast errorcorrelations are positive. This suggests that the forecast mistakes have a general character -an underestimation or overestimation of the buoyancy of trade as a whole is more importantthan idiosyncratic error.


- 32 -

Table 18. Cross-Correlation of Current Year Forecast Errors

UnitedStates Japan Germany

GDP GrowthUnited Slates LOO .47 ,00Japan 1.00 .41Germany 1.00FranceItalyUnited KingdomCanadaInflationUnited States 1.00 .43 -.34Japan 1.00 -.32Germany 1.00FranceItalyUnited KingdomCanadaBalances of Payments on Current AccountUnited States LOO -.28 -.45Japan LOO -.15Germany LOOFranceItalyUnited KingdomCanadaGrowth of Export VolumesUnited States LOO .27 .48Japan LOO .50Germany LOOFranceItalyUnited KingdomCanadaGrowth of Import VolumesUnited States LOO .48 .02Japan LOO .18Germany LOOFranceItalyUnited KingdomCanada

France

-.02.29.80

LOO

.10-.22.58

LOO

.14

.20-.18LOO

.18

.48

.65LOO

.15

.45

.62LOO

Italy

.01

.27

.23

.36LOO

.73.42

-24-.18LOO

-.11.0024.50

LOO

.03

.02

.16

.501.00

.36

.56

.60

.71LOO

UnitedKingdom

-J4-.16.33.43.03

LOO

-20-28.50.56

-.361.00

-.04.04

-27.07.05

LOO

.4121.46.12.10

1.00

.45

.59

.08

.40

.43LOO

Canada

.402124.062423

1.00

.4925

-25-.04.65

-.14LOO

-23-.55.37

-.39.00

-.181.00

.4225.16.10.46.07

LOO

.61

.402126.46.69

1.00


- 33 -

Table 19. Cross-Correlation of Year Ahead Forecast Errors

UnitedStates Japan

GDP GrowthUnited States 1.00 .53Japan 1.00GermanyFranceItalyUnited KingdomCanada

InflationUnited States 1.00 .67Japan 1.00GermanyFranceItalyUnited KingdomCanadaBalances of Payments on Current AccountUnited States 1.00 -.19Japan 1.00GermanyFranceItalyUnited KingdomCanadaGrowth of Export VolumesUnited States 1.00 .10Japan 1 .00GermanyFranceItalyUnited KingdomCanadaGrowth of Import VolumesUnited States 1.00 .62Japan 1.00GermanyFranceItalyUnited KingdomCanada

Germany France

.20 .17

.53 .381.00 .85

1.00

.09 .84-.05 .311.00 .25

1.00

-.61 -.12-.28 .391.00 -.29

1.00

.42 .28

.62 .62LOO .78

1.00

.18 .36

.48 .661.00 .67

1.00

Italy

.18

.43

.62

.771.00

.64

.33

.19

.681.00

-.50.15.17.56

1.00

.04

.222431

LOO

.41

.60

.70

.86LOO

UnitedKingdom

.44

.43

.42

.58

.41LOO

.35-.16.44.60.54

LOO

-.11.08

-29.1629

LOO

.21

.43

.63

.50

.44LOO

.61

.82,34.71.65

1.00

Canada

.67

.36

.40

.36

.27

.50LOO

.82

.49-.01.79,76.49

1.00

-.41-.46.50

•22.02.09

LOO

.43

.10-.10-.12.45.06

LOO

.66

.48

.152423.66

LOO


-34-

In the previous study (Artis, 1988) it was discovered that forecasts of nominal GDPgrowth outperformed those of real GDP and inflation taken separately. This must be true forthe extended sample used in the present study in that, with only one exception, thecorrelations between inflation and real output growth for the sample period as a whole arenegative (Table 20). It is notable, though, that when attention is given to the sampleseparation, the correlations in the second half of the sample are less supportive of theprevious conclusion. Whilst a balance of the reported correlations remains negative, they areon the whole lower in value than the nearly universal negative correlations of the firstsub-period and include quite a number of positive correlations. A straightforwardexplanation for the negative correlation of the first sub-period is that the innovations facingforecasters then were predominantly supply shocks; accordingly, this evidence suggests areduction in the incidence of this type of shock in the second sub-period. Chart 10 illustratesfor the G-3 countries - the US, Japan and Germany, where the negative relationship of outputgrowth and inflation forecast errors is indeed quite prominent for the years up to 1983 andwhere, through the 1990s, a positive relationship can be seen to emerge.

IV. The Developing Countries

Forecasts for the developing countries are analyzed for five regional groupings(Africa, Middle East, Asia, Western Hemisphere and Europe) and for one functional category- nonfuel exporters. Charts 11-14 display forecast and realization data for output growth andinflation for the current year and year ahead forecast definitions. These charts - especially ascompared with those for the G-7 (Charts 1-4) - demonstrate that WEO forecasts for thesegroups of developing countries are not particularly accurate.

Data for many of these countries are poor and tardy; their economies, in some cases,have been undergoing dramatic structural change; some of the forecasts incorporate data forcountries under IMF stabilization programmes, where the programme targets are taken as theforecast; and year-to-year growth and inflation rates can be extremely volatile.10 There areplenty of reasons why it is difficult to forecast growth and inflation for developing countriesand, hence, why the forecast performances for the developing countries will not lookparticularly impressive. This is what we found in the previous study.

The data shown in Tables 21-26 confirm this verdict. Table 21 reports the results oftests for bias and serial correlation in the forecast errors. According to the former test, thereare several instances where bias cannot be rejected. Current year output growth forecasts forAfrica, for example, appear to display a significant positive bias whilst those for inflationexhibit bias more generally - for Africa, Asia and the Western Hemisphere. When it comes

10 This issue was investigated by Barrionuevo (1993) who discovered that, indeed, it wasa source of forecast error.


- 34a -

Chart 10. Forecast Errors in GDP Growth and Inflation

Current Year Forecast


Chart 11. World Economic Outlook Forecast: Real GDP Growth in Developing Countries

Current Year Forecast and First Available Out-Turn

- 34b -


Chart 12. World Economic Outlook Forecast: Real GDP Growth in Developing Countries


- 34c -


Chart 13. World Economic Outlook Forecast: Inflation in Developing Countries

Current Year Forecast and First Available Out-Turn1

- 34d -

1 Note different scales.


Chart 14. World Economic Outlook Forecast: Inflation in Developing Countries

Year Ahead Forecast and First Settled Estimate '

- 34e -

1 Note different scales.


This page intentionally left blank


- 35 -

Table 20. Cross-Correlation of Forecast Errors Between GDP Growth and Inflation

Current Year1971-19941971-19821983-1994

Year Ahead1973-19941973-19821983-1994

UnitedStates

-.34-.55.32

-.37-36-.26

Japan

-.22-.26-.23

-.44

-.51

-.07

Germany

-.35-.29-.39

-.27

-.69.02

France

-.55-.59-.11

-.24-.51.37

Italy

-.04-.02-.20

.08-.05.19

UnitedKingdom

-.33-.34-.19

-.58-.78-75

Canada

•21-.32.49

-.08.06.37


36 -

Table 21. Test for Biasedness and Serial Correlation of Forecast Error in Developing Countries

Current year (1977-1994)GDP growth


PoSignificance levelYear ahead (1979-1994)GDP growth

PoSignificance fevefInflation

PoSignificance levelCurrent year (1977-1994)GDP growthSignificance level-Q(l)Significance Ievel-Q(2)Significance Ievel-Q(3)Inflation

Significance leveKXl)Significance level-Q(2)Significance level-Q(3)Year ahead (1979-1994)GDP growthSignificance level-QO)Significance level-Q(2)Significance level-Q(3)InflationSignificance level-Q(l)Significance level-Q(2)Significance level<X3)

Africa Asia EuropeMiddleEast

WestHemisphere

NonfuelExports

Test for biasedness

LOS.00

-3.92.01

1.07.01

-5.85.01

-.31.36

-1.77.00

-.4832

-2.46.00

1.61.15

-21.23.05

1.51.05

-22.26.11

26.70

-.19.94

1.32.08

1.39.79

Test for serial correlation (Ljung

1.00.92.35

.74

.89

.96

.98

.38

.53

.03

.08

.13

.45

.63

.82

.46

.26

.30

.13

.29

.47

.20

.23

.19

.42

.67

.81

.17

.37

.55

.10

.24

.34

.96

.96

.99

.40

.66

.83

.78

.84

.93

24.45.64

2335.46

.56

.37

-84.52.03

1.51.03

-135.45.01

-BoxQ-statistic)

.17

.38

.40

.04

.1221

.06

.15

.02

.02

.05

.11

.11

.87

-8.78.42

25.63

-27.86.00

.65

.76

.74

.49

.472\

.04

.10

.07

.00

.01

.03

Notes: For definitions of the tests etc., see notes to Tabk I . Current-year data for Europe cover the period 1980-1991. Current-year data for Middle East cover the period 1977-1991. Current-year data for Western Hemispherecover the period 1980-1994. Year-ahead data for Europe cover the period 1980-1990.Year-ahead data for Middle East cover the period 1979-1990.


- 37 -

Table 22. World Economic Outlook Forecast Accuracy: Real GDP Growth in Developing Countries

(In percent)

Africa Asia Europe Middle EastWestern

HemisphereNonfuel

Exporters

Current year (1977-1994)Mean absolute actual value

Average absolute errorRMSETheii statistic

Naive 1Naive 2

Regressionintercept

slope

joint test

R*D-W

1.98

\21134

1.091.14

-1.35(.17)1.09(.78).00.41

2.03

6.18

1.141.39

.82

.84

.76C78).92

(.87).65.16

1.58

3.32

1.813.79

.94

.71

-4.69(.00)

2.74(.00).00.92

234

3.43

1.912.48

.97

.79

.79(-42)

.67M7)3435

1.91

2.54

1.772.34

.97

.86

.65C57)

.51(22)

31.05

123

4291.742.81

.92

.99

2.08C15).44

(-09)22.05

1.74

Year ahead (1979-1994)Mean absolute actual valueAverage absolute error

RMSETheft statistic

Naive 1Naive 2

Regressionintercept

slope

joint test

R2

D-W

2.03

1.431.67

.971.23

1.69(-23)

.11(.05).00

-.07

1.39

6.40

1.581.87

.85

.87

8.50(.06)

-35(.08).12

-.05.96

2.17

1.75

2.40

1211.15

1.41C74)

.00(.50).08

-.11

.62

3242.13

2.63

1.02.81

20(.89)

.65(27)

.1225

1.15

2.99

2.03

2.88

.79

.85

-.48(.85)

.73(-68).10.02

1.02

4301.64

2.01

.81

.96

3.87dl).10

(.09).19

-.07

.92

Notes: For definitions ere., see notes to Table 2. Current-year data for Europe cover the period 1980-1991.Current-year data for Middle East cover the period 1977-1991. Current-year data for Western Hemisphere coverthe period 1980-1994. Year-ahead data for Europe cover the period 1980-1990. Year-ahead data for Middle Eastcover the period 1979-1990.


-38 -

Table 23. World Economic Outlook Forecast Accuracy: Consumer Prices in Developing Countries

(In percent)

Africa Asia Europe Middle EastWest

HemisphereNonfuclExports


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

21.745.176.991.07.84

7.52(-17)

.80(.49).04.29

1.43

9.042.002.64

.98

.96

2.79CI6)

.86(-59).01.38

1.41

58.6121.2338.71

.92

.70

-.87C95)

L59(-09).04.68

2.32

22.445.268.97LOO.77

8.28CIS)

.64CIO).24.38

1.61

211.9684.52

157.82.85.80

-98.10(-10)

2.43(.00).00.74

135

57.3024.484430

1.431.10

39.78(.02)

36(.02).04.07.92

Year ahead (1979*1994)Mean absolute actual valueAverage absolute errorRMSETheil statistic

Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

22.247.29

10.07

1.181.20

18.06(.05)

.26(.15).02

-.05.76

9.352.953.69

.941.16

8.25(.03)

.16(.10).00

-.061.13

44.9723.6346.46

1.021.03

23.10(.63)

.96(.99).31

-.081.97

23.1810.5417.01

1.421.17

19.91(.02)

.13(-01).02

-.07.72

202.01135.45219.29

.891.06

7.74(.94)

2.92(-16).0122.80

56.60283040.89

.951.12

7.09(.81)

1.72(.46).01.13.75

Notes: For definitions etc., see notes to Table 2. Current-year data for Europe cover the period 1980-1991.Current-year data for Middle East cover the period 1977-1991. Current-year data for Western Hemisphere coverthe period 1980-1994. Year-ahead data for Europe cover the period 1980-1990. Year-ahead data for Middle Eastcover the period 1979-1990.


- 39 -

I Me 24. World Economic Outlook Forecast Accuracy: Balances of Payments onCurrent Account in Developing Countries

On billions of U.S. dollars)

Africa Asia Europe Middle EastWestern

HemisphereNonfuelExports


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D»W

8.572.183.35

.991.00

-5.26(-05).38

(.04).11.05

1.35

12.966.939.18

.93

.69

-.81(-77).73

(-10).14.55

1.51

4.831.832.27

.70

.47

.46(.66)1.05(.78).90.73

1.38

12.685.127.09

.40

.54

239(35).90

(.47).08.77

1.81

21.164.836.50

.74

.45

-.87(-78)1.04(-75).53,78

1.46

36.918.50

11.11

.7139

137(.76).96

(.69).47.83

1.72Year ahead (1978-1994)


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

8.382.493.44

.79

.86

-3.11(.43).61

(-38).65.06

1.79

11.499.83

1332

.96

.93

-1.40(-73).47

(-06).06.13

1.43

6235.068.89

.851.02

-4.28(.31)-.10(.19).40

-.121.27

9.065.52122

.89

.90

3.64(.49)1.02(.97)2421

2.73

21.779.76

1134

.67

.66

-3.56C59).94

(-84).7036.75

402320.7222.81

.69

.59

-2.58C81).83

(.45).55.47.86

Notes: For definitions etc., see notes to Table 2. Current-year data for Europe cover the period 1980-1991. Current-year datafor Middle East cover the period 1977-1991. Current-year data for nonfuel exports cover the period 1973-1994. Year-ahead datafor Europe cover the period 1981-1990. Year-ahead data for Middle East cover the period 1978-1990.


- 40 -

Table 25. World Economic Outlook Forecast Accuracy: Growth of Export Volumesin Developing Countries

(In percent)


HemisphereNonfuelExports


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

3.492533.09

.73

.64

-.40(.84).81

(.69)

.40

.14

.64

10.144.845.76

,891.15

12.84(-01)-.34(-03).03

-.052.03

6.995.407.05

1.06.79

-10.47(.10)3.23(-12)5434

1.65

5.336.988.02

.781.02

1.04(-69).15

(-15)33

-.101.92

5.453.764.54

.70

.80

.14(.95).75

C53)

.54

.171.45

6.%2.493.10

.67

.79

1.82(33).83

(.55).4358

1.92Year ahead (1981-1994)


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

3.342.293.32

.52

.52

36(.84).52

(-27).15.04

1.33

10.494.435.50

.831.06

1150(.05)-.09(.11).07

-.081.80

7.225.63859

.76

.84

-22.16(.08)5.16(.09).1835

159

5.547.108.14

1.041.00

.64(.90)54

(-41).41

-.112.16

5.803.344.52

.54

.78

-.55(-88).81

(.74)37.08

1.34

7.643.404.15

.66

.97

750(.16).03

(.18)35

-.061.65

Notes: For definitions etc., see notes to Table 2. Current-year data for Europe and Middle East cover the period1981 -1991. Current-year data for nonfuel exports cover the period 1973-1994. Year-ahead data for Europe andMiddle East cover the period 1981 -1990. Year-ahead data for nonfuel exports cover the period1977-94.


. 41 -

Table 26. World Economic Outlook Forecast Accuracy: Growth of Import Volumesin Developing Countries

(In percent)


HemisphereNonfuclExports

Current year (1979-1994)Mean absolute actual valueAverage absolute errorRMSETheil statistic:

Naive 1Naive 2

Regressionintercept

slope

joint testR7

D-W

3.313.464.67

1.071.10

-1.97(.07).37

C13).06

-.011.40

9212.943.89

.74

.76

1.50(-61).96

(.90).52.31

2.06

5.055.939.95

1.521.47

5.98(.03)-.95(-00).00.31.91

6.515.767.19

.91

.83

-335(.09).73

(27).08.42

1.37

9.038.048.83

.94

.74

-.71(.80)128(.57).8429.69

6.813.003.81

.69

.67

25C«8).96

(.88).99.43

130Year ahead (1980-1994)


Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

3.834.094.95

.871.03

-1.72(.19).31

(-25).12

-.051.99

9.093.154.60

.84

.88

4.14(23).66

(-43).30.09

124

3.614.165.66

1.11120

2.54(.33)-.07(.04).06

-.111.88

6236.40929

.901.00

-4.43(-14).58

(.47).12.01

1.14

8.427.018.43

.56

.56

-4.67(.16)1.57(25)35.42

1.09

7363.434.76

.74

.79

.49(.89).96

(.94).98.10

1.16

Notes: For definitions e/c., see notes to Table 2. Current-year data for Europe cover the period 1980-1991.Current-year data for Middle East cover the period 1979-1991. Current-year data for nonfiiel exports cover theperiod 1973-1994. Year-ahead data for Europe and Middle East cover the period 1980-1990.


-42-

to the year ahead forecasts, the suggestion of bias is more widespread: positive growth bias issignificant for Africa, Europe and the Western Hemisphere, and a negative inflation bias isevident for all but two of the regions distinguished. The Q-statistic test for up to third orderserial correlation, on the other hand, reveals rather little evidence that the forecast errors areautocorrelated: the principal exception is the finding of some first order serial correlation inthe year ahead inflation forecast errors for Africa, the Western Hemisphere and the group ofnon-fuel exporters. Tables 22-26 show the average absolute error of the forecasts (andimmediately above, the mean absolute actual value), the RMSE, Theil statistics for twoalternatives (Naive 1 - the random walk alternative; Naive 2~the mean reversion alternative),and the weak efficiency realization-forecast regression. Table 22 concerns the forecasts foroutput growth: it is easy to see that these are little better on average than a random walkassumption would be, as the Theil: Naive 1 statistics are close to (sometimes above) unity.Nor is the evidence from the weak efficiency test reassuring: although for the most part thejoint test for efficiency is satisfied, the p-values recorded are generally not high. The criteriasuggest that the year-ahead forecasts are less accurate than the current-year projections.

For forecasts of inflation, Table 23 affords a not dissimilar picture. The Theil: Naive1 statistics are close to, sometimes above, unity whilst the forecasts barely pass and often failthe weak efficiency test. The year ahead projections appear again to be less satisfactory thanthe current year forecasts. Again, by comparison with the results for the industrial countries(Table 2) the quality of these forecasts is markedly poorer.

The WEO also provides data, for roughly one half of our period, pertaining to themedian forecasts for inflation and output growth. The former are available for the regionalaggregates quoted in the main tables here, the latter only (except for a very short period) forthe developing countries as a bloc. It might be expected that the median figures would beless prone to disturbance by single large country shocks, or by the practice of citing programtargets as forecasts, than the average figures processed in Tables 21-26 here. As far as theinflation forecasts go, this speculation is largely confirmed when the median data aresubstituted for the averages: the results (not shown) involve a large fall in the RMSE andaverage absolute error statistics. Nevertheless, the quality of the forecasts continues to leavea good deal to be desired as regards conformity with weak efficiency desiderata andacceptably low Theil statistics. The errors in median output growth forecasts, which couldonly be examined for the developing country group as a whole and not for the regionalaggregates, showed no improvement at all in relation to the errors revealed for the averageforecasts.

Whilst the balance of payments forecasts for the industrial countries were notablyweaker than the output growth and inflation forecasts for those countries this is not obviouslythe case for the developing countries (Table 24). However, these forecasts are not of highquality either. The export and import volume growth forecasts (Tables 25 and 26) are ofcomparable quality to the output forecasts: again the value of the Theil: Naive 1 statistics areclose to, sometimes above, unity whilst the estimates pass the weak efficiency test with littleto spare.


-43 -

Table 27, finally, examines WEO forecasts of commodity prices (other than fuel).Prices of agricultural raw materials, beverages, food, minerals and metals and total nonfuelexports are separately distinguished. These forecasts generally satisfy the weak efficiencytest on the parameters of the realization-forecast regression and, when compared to the twonaive alternatives, are apparently superior. But their accuracy is not high: the averageabsolute error of forecast is high in relation to the mean absolute actual value (in one case -the year ahead forecasts for minerals and metals - greater), Theil statistic values are quitehigh with a number exceeding 0.70; the R2 of the realization-forecast regression is generallylow.

The overall conclusion must be that the developing country forecasts are distinctlyweaker than those for the developed industrial group; whilst most of them pass the hurdlesset by Theil statistic values less than one and parameter values in the realization-forecastregression which do not reject the standard joint restriction, these are not powerful tests andoverall accuracy is not high. This finding qualitatively repeats the conclusion of the earlierstudy. There are a number of reasons why we might expect it to be the case.

V. Conclusions

The overall conclusions to which this study directs us are not dissimilar from thosethat were derived in the earlier investigation.

Taking the period as a whole, WEO forecasting

passes most conventional tests in forecasting economic developments in theindustrial country group, for the most part, quite easily, but

the balance of payments is much the worst forecast variable and

year-ahead forecasts are less good than near term forecasts; yet

in terms of directional accuracy, the record is less satisfactory, whilst

the record in anticipating the strength of the boom and subsequently the lengthof the recession in the last cycle was rather poor; and,

for the developing countries the record on conventional measures of forecastaccuracy is much less good than it is for the developed group;

comparison with private sector forecasts over a limited period suggests thatthe major forecasting errors were substantially the same


- 44 -

Table 27. World Economic Outlook Forecast Accuracy: Nonfuel Commodity Prices

(In percent)

Agricultural RawMaterials Beverages Food

Mineralsand Metals

NonfuelExports

Current year (1981-1994)Mean absolute actual valueAverage absolute errorRMSEThcil statistic

Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

7.585.718.03

.58

.67

-2.99(-25)1.69(-16)32.49

2.11

16.7912.8820.53

.75

.79

-.41(-95),89

(-78).96.23

1.09

9.246.248.01

.53

.59

-1.98(37)1.33(.30).40.58

1.43

11.0210.2012.47

.75

.79

.49(.89)1.05(.91).9922

1.40

8.076.327.67

.61

.66

-128(.57)1.05(.89).84.42

1.59Year ahead (1981-1994)

Mean absolute actual valueAverage absolute errorRMSEThcil statistic

Naive 1Naive 2

Regressionintercept

slope

joint testR2

D-W

7.476.519.02

.51

.72

-1.54(-61)1.39(.54).80.23

1.48

16.8113.9120.54

.76

.78

-1.16(.85)1.48(.45).74.27

1.32

8.946.969.13

.47

.66

•225(39)1.27(.53).57.40

1.10

11.4213.0416.03

.73

.93

.74(-87).30

(.18).37

-.05124

7.916.097.64

.50

.61

-.98(.65)1.37(.40).64.42

1.28



-45-

At this point we can return to the issue whether any improvement is detectablethrough time in WEO forecasting. There are a number of reasons why there should beimprovement: there is the cumulation of experience, there have been significant advances indata processing which should improve timeliness and there is the competition offered by thelarge increase in economic forecasting practice around the world. At the same time it is clearthat there are changes in the stochastic structure of the world economy. Earlier in the studywe noted that on the basis of summary statistics of forecast accuracy, little could be saidregarding change in forecast accuracy between the first half of our sample period and thesecond. A different type of comment might be prompted by the relative experiences incyclical turning point prediction. In the last study it could be seen that the forecasting recordfor events following the second oil price rise was better than that for the period following thefirst; and it also seemed reasonable to excuse the forecasters for not having foreseen the priceincreases themselves. Thus the two major cycles associated with the oil price shocks were ofa character such that the forecast record actuaJJy emerged rather we]]. The period since themid-1980s has been rather different. The prevalence of supply shocks is not so obvious(although there is the 1986 fall in oil prices and the 'mixed' but substantial shock of Germanunification); the major world boom towards the end of the decade and the following deeprecession appear to be endogenous to the development of the economy in a way that providesfewer obvious 'excuses' to forecasters. The greatest weakness of the subsequent forecastrecord lies therefore in the failure to anticipate in timely fashion this cycle. If the judgementis right that this cycle was largely endogenous to the natural momentum of a world economynow more integrated then ever before and substantially less regulated than in earlier decades,the forecasters will do well to learn from it: it is too early to say how successful this learningprocess will be.


- 46 - APPENDIX A

Table Al. Classification and Calibration of Forecasts in The World Economic Outlook (WHO)

Current year

Publication

May 27, 1971

April 13, 1972

June 14, 1973

May 23, 1974

May, 23, 1975

July 9, 1976

July 5, 1977

April 4, 1978

February 15 or

May 19801

June 198 1!

April 19821

May 19831

April 19841

April 19851

April 19861

April 19871

April 19881

April 19891

May 1990'

May 199V

May 19921

May 1993'

May 19941

May 1995'

forecast and first available

Forecast

1971

1972

1973

1974

1975

1976

1977

1978

June 11, 1979 1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

--

out-turn

Out-turn

--1971

1972

1973

1974

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

Year ahead forecast and

Publication

January 31, 1973

December 21, 1973

December 24 & 31, 1974

December 12, 15 & 16, 1975

March 3, 1977

December 27, 1977or April 4, 1978

December 1, 1978or February 15, 1979

August 30, 1979or May 19801

August 22, 1980or June 198 11

August 24, 1981or April 1982'

August 2, 1982or May 19831

August 19, 1983or April 19841

September 19841

October 1985'

October 1986'

October 19871

October 1988'

October 19891

October 19901

October 199 1 1

October 19921

October 19931

October 19941

October 19951

first settled

Forecast

1973

1974

1975

1976

1977

19781978

19791979

19801980

19811981

19821982

19831983

19841984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

--

~

estimate

Estimate

----

1973

1974

1975

19761976, 1977

1977

19781978, 1979

19791980

19801981

19811982

19821983

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

Notes:The table shows, for each of the documents listed, the forecast and out-turn information to be found in it. Forexample, the April 1984 issue of The World Economic Outlook contains current year forecast data for 1984 and,as well, first available estimates for 1983. Occasionally the complete data set is to be found only in two or moredocuments (more details may be found in Tables B1 -B12 and C1 -C12). The superscript (1) indicates a publisheddocument; as shown by the absence of the superscript, many of the earlier documents were not published, beingcirculated only within the Fund.


- 47 - APPENDIX A

Table A2. Basic Sample Distribution Statistics: Current Year Forecasts and RealizationsUnitedStates Japan Germany France Italy


GDP GrowthMean:

Standarddeviation:

Skewness:

Kurtosis:

InflationMean:

Standarddeviation:

Skewness:

Kurtosis:

RealizationForecastRealizationForecastRealizationForecastRealizationForecast

RealizationForecast

RealizationForecastRealizationForecastRealizationForecast

2.602.682.562.57

-.32-.60-.44.85

5.115.13

2.482,38.83.78

-.60-.73

4.274.372.792.60

.131.66**.91

3.91**

3.774.234.624.392.61**2.42**8.03**6.91**

2.162.342.191.95-.83.07.76.76

4.084.121.801.54.73.82.05.27

.2.452.542.011.72-.21.51.56

-.45

6.756.493.613.51

.08

.33-1.43-1.10

2.092.222.241.54-.62-.86.73

1.15

11.1510.615.495.61

.24

.44-1.50-1.43

1.531.682.262.15

.09-.32-.26.60

9.128.665.935.301.52**1.29*2.78*1.29

2.853.072.552.18

-1.23*-.192.47*-.16

5.835.693.622.99.38.35

-.99-1.08

Balances of Payments on Current AccountMean:

Standarddeviation:

Skewness:

Kurtosis:


-57.29-56.2561.5757.83

-.41-.42

-1.55-1.64

41.8139.9846.4145.61

.71

.82-.76-.36

7.528.87

23.8821.99

.60

.90-.54.66

-1.70-1.625.674.84

.511.08.02.96

-3.51-3.379.279.42-.431.04.43

3.94**

-5.47-6.1811.9511.29

-.83-.81.58

-.29

-7.56-6.828.006.85-.88-.61-.40-.40

Growth of Export VolumesMean:

Standarddeviation:

Skewness:

Kurtosis:

Growth ofMean:

Standarddeviation:

Skewness:

Kurtosis:


Import VolumesRealizationForecastRealizationForecastRealizationForecast

RealizationForecast

6.475.868.356.79.02

-.01.93.84

6.725.13

8.695.98-.16.65.20

1.58

5.264.606.795.601.011.24*.33

4.36**

5.916.538.685.97.16

LOO1.123.81**

4.754.806.443.59-.70.40

1.55.83

5.315.516.203,80-.891.022.47*1.09

4.854.934.813.22-.38.91.00

2.81*

4.764.706.504.45-.18.91.91

1.61

5.345.474.533.30-.721.38*2.52*1.17

4345.007.544.15-.67-.50-.211.17

4.083.934.003.20.10.73

-.202.52*

4.754.105.593.61-.61-.41

-.10.05

5.843.956.804.26

.48

.181.57.07

6.334.27

7.964.46-.62.50

1.00.24

Note:'**' indicates skewness or kurtosis at 1% significance level; '*' at 5% significance level.


- 48 - APPENDIX A

Table A3. Basic Sample Distribution Statistics: Year Ahead Forecasts and Realizations

GDPGrowthMean: Realization

ForecastStandard Realizationdeviation: Forecast

Skewness: RealizationForecast

Kurtosis: RealizationForecast

InflationMean: Realization




Balances of PaymentsMean: Realization




UnitedStates

2.522.792.612.06-.43-.49-.60.40

5.305.252.612.02.68.89

-.91-.11

on Current-53.79-54.6161.4061.28

-.40-.30

-1.58-1.77

Japan

3.934.692.612.01.04

1.96**1.244.42**

3.654.104.803.862.61**1.95**7.66**3.88**

Account42.1738.8046.1442.04

.71

.80-.76-.20

Germany

2.162.772.25.98

-.72.80.32.27

3.823.851.621.45LOO.72

1.10.33

10.9212.4823.4318.64

.42

.36-.471.15

France

2.212.731.761.18.07

1.21*.22

2.11

7.156.654.163.61.13.26

-1.48-1.40

-1.16-1.385.423.94

.61

.16-.16-.66

Italy

2.112.442.311.36-.61.39.60

1.78

11.4610.655.495.31.10.37

-1.60-1.15

-2.96-4.359.948.77-.48

-1.85**.64

4.78**

Unitedkingdom

1.692.132.301.18-.33-.21-.591.73

9.148.136.094.551.60**1.13*3.08*.66

-3.82-4.3212.6010.99

-.69-.87.07

-.62

Canada

2.573.352.501.39

-1.09-.031.78-.38

6.025.643.812.54

.38

.45-.98-.26

-7.75-6.448.636.79-.93-.81-.32-.53

Growth of Export VolumesMean: Realization




6.445.198.365.94.00

-.76.93.53

5.634.756.864.491.02

-1.17*.30

1.89

4.714.935.952.38-.30.67

1.42.24

4.655.434.412.39-.511.21*-.471.58

5.195.254.423.20-.982.43**2.56**7.19**

4.164.043.822.78-.12-.12-.252.51*

6.624.445.693.04.64.06

1.54-.13

Growth of Import VolumesMean: Realization




6.625.148.784.67

.16

.62

.21-.07

5.136.438.374.93

.431.032.032.81*

5.025.195.563.00-.17,44

-.02.17

4.925.005.923.38

.25

.73

.671.76

4.104.967.693.91-.73-.20-.12.82

4.394.175.632.80-.50-.10-.271.11

6.004.107.493.58-.96.74

1.491.55

Note:indicates skewness or kurtosis at 1% significance level; ** at 5% significance level.


Table Bl. Industrial Countries: Real GDP: Current Year Forecasts and First Available Out-Turns (F.A.)(Annual Percent Change)

United StatesForecast F.A.

197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

3.06.07.00.0

•4.06.85.14.53.0

•1.02.4

-1.02.45.03.42.92.32.93.11.70.21.63.23.9

2.76.45.9

-2.1-2.06.14.94.02.3

-0.22.0

-1.73.36.82.22.52.93.93.01.0

-0.72.13.04.1

JapanForecast F.A.

9.56.0

12.72.02.36.66.15.75.04.04.03.52.83.94.33.02.74.14.54.43.62.21.30.7

6.19.2

11.0-1.82.06.35.15.66.04.22.93.03.15.84.62.54.25.74.95.64.51,3O.I0.6

Germany8

Forecast F.A.3.32.05.82.00.56.84.53.14.02.5

-0.91.00.52.62.73.71.91.72.43.52.81.3

-2.00.5

2.92.95.30.4

-3.45.72.53.44.41.8

-0.3-1.11.32.62.42.41.73.44.04.53.11.5

-1.92.3

FranceForecast F.A.

5J4.76.14.51.85.53.43.13.51.80.42.10.00.61.42.41.81.62.83.12.11.80.01.2

5.45.76.13.9

-2.55.22.73.33.31.20.81.60.61.81.12.22.13.43.42.81.21.8

-0.72.5

ItalyForecast F.A.

4.03.04.54.0

•2.02.21.52.64.53.00.22.3

-0.11.92.82.32.92.53.43.01.71.60.31.1

1.43.25.93.4

-3.75.61.72.65.04.0

-0.2-0.3-1.42.62.22.82.83.83.21.91,00.9

-0.72.5

United KingdomForecast F.A.

7,54.76.5

•1.00.82.90.92.92.5

•2.2-2.80.81.52.63.02.83.03.03.31.1

-2.10.81.42.5

0.92.36.60.3

-1.71.30.03.31.1

-1.7-2.10.72.92.43.32.54.54.42.30.6

-2.2-0.61.93.8

CanadaForecast F.A.

5,26.67.35.2

•0.75.33.34.53.50.22.5

•0.51.95.03.23.32.03.42.91.6

• 1.12.33.23.5

5.45.87.13.70.64.92.63.42.90.13.0

-4.83.04,74.53.13.94.52.90.9

-1.50.92.44.5

All Industrial6

Forecast F.A,4.05.17.11.5

-1.15.74.44.03.51.01.50.81.63.63.13.02.32.83.32.71.31.81.72.4

3.25,66.4

-0.2-1.65.43.73.83.41.21.2

-0.32.34.92.82.43.14.13.52.50.81.51,23.0

Summit SevenForecast F.A.

4.05.37.5'/.3'

-1.5'6.14.54.23.50.81.60.71.83.93.23.02.42.93.42.71.31.71.92.5

3,25.84

6.74

-0.54

-1.6'5.63.94.03.51.21.4

-0.42.55.12.72,53.14.23.52.60.81.61.43.1

Europe6'7

Forecast F.A.J.62

3.4s

5.62.607473.12.83.51.5

-0.41.50.61.92.42.92.22.02.8271.41.80.11.3

2.85

3.75.62.0

-2.64.42.02.93.41.4

-0.30,21.12.42.32.42,63.63.42.60.81.1

-0.32.8

1, Constructed from country detail using 1971 GNP weights, Main Economic Indicators (MEI), December 1972.2 & 3. Constructed as residual from detail shown using GNP weights from MEI December 1972, November 1971, the "first available11.4. Constructed from country detail and 1971 GNP weights, MEI, December 1972.5. Constructed as residual from figures shown and 1970 GNP weights, MEI, November 1971.6. Definitions of "All Industrial" and "European" were changed with effect from 1980 on; these figures are reconstructed from country detail on old definitions.7. Definition of "European" was changed from "European countries" to "European Community" in 1992 and subsequently to "European Union" in 1994.8. Germany is West Germany throughout.


Table B2. Industrial Countries: Inflation (GDP Deflator): Current Year Forecasts and First Available Out-Turns (F.A.)(Annual Percent Change)


197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

4.63.34.28,59.35.55.86.29.09.09.67.04.14.13.8332.83.24.74.13.72.42.62.2

4.63.05.4

10.38.85.15,57.38.89,09.26.04.23.83.32.63.03,44.24.23.72.62,62.1

JapanForecast F.A.

5.04.86.4

20.012.66.85.83.94.07.74.423170.71.41.11.11.21.41.92.62.11.51.2

4.34.8

12.220.97.36.46,34.82.03.12.92.00,70.51,71.8

-0.10.51,51,91.91.81.00.2

Germany7

Forecast F.A.635.56.58.06.33.83.73.54.05.05.04.54.03.02.22.62.52.02.52.93.9433.93.0

7.76.15.96.68.23.13.63.93.85.04.34.83.21.92.13.32.11.52.53.44,54,53.92.0

FranceForecast F.A.

4.55.25.8

10.59.09.48.78.19.5

10.411.913.79.87.26.14.33.02.53.2333.32.52.01.9

5.55.67.29.6

12,79.79.39.99.6

11,61 1.012.19.27.05,95.33.22.82.73.02.92.72.11.4

ItalyForecast F.A.

5.55.08.0

10.219.319.019.512.514.518,518.017,015.112.29.08.65.75.06.16.56.15.24.73.5

6.75.9

11.216.317.417.819,513.315.120.317.617.515.010.79.09.15,85.46.37,67.54.74.43.5

United Kingdom CanadaForecast F.A, Forecast F.A.

7.59.07.0

15.923315.513.09.5

10.018.711.010.06.05.05.03.74.64.86.65.16.64.42.53.1

11,16.58.0

12.627.615.215.410.113,918.812.58.05.!4.26.13.64.46.06.76.16.94.63.42,0

3.63.65.09.0

11.19.07.06.07.5

1039.89.8834.93.03.93.54.04.34.0432.41.11.1

3.44.67.1

13.110.79.56.56.89.9

10.510,010.66.23.03.22.84,64.24.93.12,71,00.80.6

All Industrial5

Forecast F.A.5.04.45.4

10.810.8

7.66.96.27.59.68.97.65.64.53.93.42.93.03.83.94.13.22.823

5.64.47.1

1 1.710.67.27.17.07,69.08.47.25.14.13.93.42.93.13,94.13.93.12.61.9

SummitForecast

4.9435.4*

10.92

10.8*7.57.06.18.09.68.97.55.14.23.73.22.72.93.83,63.83.02.72.2

SevenF.A.5.54.24

7.14

1K94

10.54

7.17.17.27.69,08.36.74.73.63.63,12.72,93,63,83.83.02.51.7

Europe5'6

Forecast F.A.5.7'4.6'6.9

10311.5

9.68.67,07.5

1039.69.88.06.15.1433.6334.14.65.14.53.63.1

7.7>6.47.79,7

13.69.28.97.68,5

10.79.69.57.25.75.35.03.63.6g4,65.25.54.63.62.6

1. Constructed as residual of figures shown, using 1970, 1971 GNP weights (Main Economic Indicators (MEI), November 1981, December 1982).2. Constructed from country detail using GNP weights.3. Calculated as residual from figures shown using 1970 GNP weights (MEI November 1971),4. Constructed from country detail using first available GNP weights.5. Definitions of "All Industrial" and "European" were changed with effect from 1980 on; these figures are reconstructed from country detail on old definitions.6. Definition of "European" was changed from "European countries" to "European Community" in 1992 and subsequently to "European Union" in 1994.7. Germany is West Germany throughout.


Table B3, Industrial Countries: Balances of Payments on Current Account1: Current Year Forecasts and(in billions of U.S. dollars)


197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

n.a.n.a.

•5.71.00.5

-2.2-70.5-19.1•10.0

-3.09.00.0

-25.0-73.0

-124.0•110.5-138.9-141.1-139.3•113.3

-37.8-53.4

-101.0-140.3

n.a.n.a.3,1

-4.011.9-0.6

-20.2-16.0-0.33.76.6

-8,1-39,3

-101.6-117.7-140.6•160.7-135.3-106.0

-99.3-8.6

-62.4-109.2-155.7

JapanForecast F.A,

n.a.n.a.5.2

-6.4-0.92.65.1

11.45.0

^16.5-2.012.017.027.539.472.083.077.884.057.442.0933

137.2133.4

n.a.n.a.

-0.1-4.7-0.73,7

11.016.6-8.6

-10.74.86.9

21.035.049.786.086.779.557.235.772.6

117.6131.4129.3

Germany2

Forecast F.A.n.a.n.a.033.68.82.73.04.05.5

-15.0-14.0

4.07.54.0

11.025.434.941.449.762.3

9.8-14.1-26.7•13.0

n.a.n.a.4.79.33.92.93.58.7

-5.7-16.0

-7.63.33.36.3

13.236.044.248.552.844,5

-20.7-25.1-21.9-22.6

FranceForecast F.A.

n.a.n.a.0.9

-6.3-4.1-333.8-1.92.0

-4.0-8.0-7.5-5.00,00.59.03.6

-2.9-2.0-3.9-6.8-4.52.0

10.3

n.a.n.a.0.2

-6.20.3

-6.0-3.14.02.4

-7.4-7.9

-12.0-3.80.00.33.7

-4.4-3.8-3.3-7.5-5.72.8

10.59.6

ItalyForecast F.A.

n.a.n.a.1.1

•7.9•3.6-2,1-0.91.93.5

-4.0•9.5-7.0-1.50.5

-5.52.13.8

-1.0-53

-11.5-17.5•20.2-15.726.2

n.a.n.a,

-3.2-7.9-0.5-2.92.26.35.2

-9.8-8.0-5.50.3

-3.2-3.74.70.0

-4.2-10.9-15.7-19.3-25.2

10.613.5

First Available Out-Turns


n.a.n.a.

•1.6•10.4

-5.5-2.80.21.0L5

•2.58.59.03.02.01.83.9

•4.9-7.3

•30.3•25.7-15.6•14.9-26.0•19.4

n.a.n.a.

-3.6-9.0-3.8-2.50.00.5

-5.26.0

16.26.93.10.33.8

-1.6-2.8

-26.2-34.2-22.8-7.8

-21.0-16.0-0.6

CanadaForecast F.A.

n.a.n.a.

-0.7-0.8•4.8-4.6-3.7•3.0-4.0•6.0-2.5-5,04.5

•1.52.1

•4.2•8.8

-10.7•13.8-20.2-10.2-18.4-19.2•14.5

n.a.n.a.

-0.3-1.9-4.9-4.2-4.0-4.6-4.4-1.3-5.52.20.61.5

-1.9-6.3-7.2-9.2

-16.6-13,7-23.4-23.7-19.5-18.1

(F.A.)


n.a.n.a.

•0.5•27.2

•9.6•9.7

• 10.6•5.73.5

-5LO• 18.5

5.51.0

•40.5-74.7

-2.4-273•43.7•57.0•54.9•36.2•32.1-49.2• 173

n.a.n.a.0,8

-24.46.2

-9.6-10.6

15.5-16.6-35.5

•1.5-6.3

-14.9-61.3-56.2-18.1-44.2-50,7-61,1-78.8-12.9-37.0-14.2-44.6

1. Figures include official transfers.2. Data through June 1990 apply to West Germany only.


Table B4. Industrial Countries: Growth of Export Volumes: Current Year Forecasts and First Available Out-Turns (F.A.)(Annual Percent Change)


1971197219731974197519761977197819791

1980)9811982198319841985198619871988198919901991199219931994

n.a.8.5

17.07.50.13.33.46.39.27.9

-L4•8,5-7.21.33.83.3

12.721.311.312.05.84.74.77.8

n.a.9.3

23.88.2

-2.53.40.57.79.68.3

-3.3-12.0

-6.88.6

-1.27.8

12.124.112.09.06.66.95.2

11.4

JapanForecast F.A.

n.a.6.99.05.43.6

22.95,93.0

•1.49.07.04.00.96.06.4

-1.8-4.0•4.25.94.85.53.95.2

-1.1

n.a.6.65.0

15.82.0

21.85.3

-1,1-1.317.610.5-3.08.7

14.84.8

-1.4-2.04.33.95.03.00.7

-1.01.0

Germany2

Forecast F.A.n.a.3.9

12.89.0

•3.012.1

7.44.25.94.73.56.51.64.58.44.00.52.05.65.82.82.71.24.4

n.a.8.5

18.012.3

-10.412.45.65.57.44.05.31.90.49.26.41.22.47.67.74.50.8

-0.1-10.3

8.9

FranceForecast F.A.

n.a.7.4

14.46.4

-1.79.98.26.45.54.32.23.04.04.94.03.43.03.95.46.74.44.30.62.8

n.a.14.310.910.2-3.89.26.66.38.83.25.2

-3.94.06.01.90.21.36.99.45.93.65.5

-4.54,4

ItalyForecast F.A.

n.a.7.5

13.711.14.0

12.88.93.66.02.02.05.52.53.44.13.52.53.84.96.74.93.25.33.8

n.a.14.24.46,91.7

12.26.06.57.1

-7.94.70.35.15.66.52.23.16.49.24.70.53.78.8

11.0


n.a.4.58.32.31.98.86.73.35.42.9

•3.44.01.74.34.93.02.42.71.8

12.91.13.95.31.7

n.a.-1.112.26.8

-4.08.38.23.83.61.8

-1.40.81.47.15.83.75.70.25.76.81.62.83.0

11. 1

CanadaForecast F.A.

n.a.6.5

13.00.3

-4.810.311.05.14.1

•0.52.5

-1.53.17.86.44.23.32.50.42.2

•1.45.14.46.9

n.a.9.89.2

-5.9^7.111.39.97.92.60.53.2

-0.27.4

24.44.22,66.49.3

-0.73.81.69,0

10.115.1

All IndustrialForecast F.A.

n.a.6.0

12.76.3

-0.510.46.94.75.34.81.72.00.64.45.32.73.44.66.17.24.0423.54.0

n.a,8.9

13.47.7-4.510.54.45.76.54,52.6

-2.52.09.93,92.94.38.77.05.42.73.20.48.6


n.a.6.4

13.46.4

^0.310.1

7.14.95.95.21.61.00.04.35.62.53.35.16.37.83.94.03.73.9

n.a.8.5

14.88.3

-4.110.05.45.86.05.13.0

-3,61.1

10,33.52.54.49.47.36.12.43.50.08.6

1. From February 1979 WEO (not June).2. Data through June 1990 apply to West Germany only.


Table B5. Industrial Countries: Growth of Import Volumes:(Annual Percent Change)

Unites StatesForecast F.A.

1971197219731974197519761977197819791

198019811982198319841985198619871988198919901991199219931994

n.a.5.97.80.5

•7.321.210.63.42.3

-2.50.41.03.7

15.29.84.11.85.95.09.61.51.47,79.1

n.a.13,74.8

-0.4-13.122.213.47.32.1

-6.00,6

-4.410.023.26.5

14.65.17.15.33.7

-0.411.612.715.0

JapanForecast F.A.

n.a.11.425.3

5.6-4.81L38.98.27.9

^3.51.42.02.33.86.65.58.5

12.78.24.39.43.05.07.2

n,a.10,827.8-0,2

-13,412.03.99.5

11.6-7.4-2.0-0.11.3

10.7-0,413.58.3

16.77.85,83.0

-0.74.0

13.5

Germany2

Forecast F.A.n.a.6.9

14.75.03.2

14.48.46.08.94.20.02.00.94.04.66.63.65.25.08.08.73.12.4LQ

n.a.9.47.8

-1.32.6

15.35.78.49.32.1

-3.50.75.45.44.66.15,06.98.0

13.016.21.3

-12.96.7

Current Year Forecasts

FranceForecast F.A.

n.a.8,4

15.26.2

-3.215.72.76.06.54.00.05.5

-0.70.02.76.54.83.65.57.64.34.00.32.5

n.a.13.513.23.4

-6.919.1

1.46.0

11.75,5

-1,23.1

-1.62,74.37.96.28.09,05.72,10.9

-10.45.9

ItalyForecast F.A.

n.a.6.1

14.06.5

-4.38.24.64.6

10.76.5

-4.05.51.54.07.08.57.56.16.06.74.43.40.01.5

n.a.12.014.2-5.8

-11.614.00.03.2

14.02.0

-10.02.01.09.47.56.0

10.07.28.34.04.73.2

-7.712.2

and First Available Out-Turns (F.A.)


n.a.9.3

10.31.3

-3.23.74.15.44.3

-0,8-2.210.04.26.25.53.16.06.24.55.7

-2.03.43.55.9

n.a.7.6

13.21.2

-7.16.21.46.5

11.6-5,2-4.75.07.1

10.44.66.27.4

13,29.12.2

-2.85.94.26.0

CanadaForecast F.A.

n.a.10 .113.75.2

-2.09.85.4

-0.21.52.03.5

-3.01.5

13.53.42.04.96.04,93.9

-2.74.43.76.6

n.a.16.313.83.0

-4.67.7I . I3.49.2

-4.12.4

-14.714.721.3

8.26.09.1

13.97,6

-0.61.67.0

10.313.1

All IndustrialForecast F.A.

n.a.7.4

13.04.1

•2.612.46.64.25.6LS

•0.22.02.16.66.15.04.36.05.96.73.63.53.95.4

n.a,10.611.60.4

-7.514.54.85,78.6

-0.8-2.3-0.54,1

12.25.28.96,19.58.05.12.54.0

-0.210.5

SummitForecast

n.a.7.8

13.63.9

•2.813.97.14.75.90.7

-0.22.52.57.96.55.14.46.65.97.23.53,34.65.7

SevenF.A.

n.a.11.811.60.0

-6.915.45.46.78.5

-1.8•2.4-1.15.4

13.85,09.66.2

10.07.95.62,54.80.7

11.4

1. From February 79 WHO (not June).2. Data through June 1990 apply to West Germany only.


Table B6. World Trade Volumes and Terms of Trade: Current Year Forecasts and First Available Out-Turns(Annual Percent Change)

197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

WorldForecast

n.a.6.81

12.0'6.0

-1.9s

10.5*7.05.05,53.01.52.01.05.55.43.33.35.55.86.62.45.05.25.8

TradeOut-Turn

n.a.8.3'

12.55.1'

-4.12

11.55.05.06.51.50.0

-2.52.08.82.94.94.99.37,23.93.34.22.49.4

Industrial Countries' Terms of TradeForecast Out-Turn

n.a.n.a.n.a.

-8.6I.I

•0.7•0.21.4

•Ltf-6.0•0.51.51.50.00.06.70.62.00.00.21.0

-0.3-0.21.0

n.a.n.a.

-2,2-11 .5

2.8-0.70.03.03

-3.0-6.5-0.62.52.1

-0.21.89.40.91.70.0

-0.51,61.81.41.2

1. Computed as simple average of import and export growth figures.2. Import growth.3. Computed as the difference between export unit value increases and import unit value increases.


Table B7. Developing Countries:(Annual Percent Change)

Real GDP: Current Year Forecasts and First Available Out-Turns

AfricaForecast Out-Turn

197719781979!98019811982198319841985198619871988198919901991199219931994

4.23.7373.8422.62.33.52.92.82.53.1232.72.02.72.73.4

2.92,52.64.92.61.10.12.21.61.60.91.72.91.91.40.91.12.7

AsiaForecast Out-Turn

6.95.76.35.85.34.85.06.15.85.55.16.96.45.45.05.56.77.5

6.06.93.24.94.74.26.56.46,15.76.69.05.15.35,87.98.48,6

EuropeForecast Out-Turn

n.a.n.a.n.a.2.12.92.81.51.83.23.52.52.12.30.1

•3.5-13.5

n.a.n.a.

n.a.n.a.3.52.02.72.00.62,52.53.22.62.50.3

-2.9-16.0

n.a.n.a.n.a.

Middle EastForecast Out-Turn

4.76,65.45.65.76.83.03.62.90.2

•1.01.52.33.2

-3.315.0n.a.n.a.

5.17.27.95.14.72,74.22.3

-1.60.0

•1.03.93.9

-1.50.4n.a.n.a.n.a.

Western HemisphereForecast Out-Turn

n.a.n.a.n.a.5.65J3.5

-0.21.33.21.63.32.10.81.41.02.72.32.8

n.a.n.a.6.35.8

-0.1-1.5-2.32.43.84.02.30.90.9

-1.02.82.33.44.6

NonfuclForecast

5.7'5.0'5.4*4.?4.9*3.82

2.32

3.5*4.4

4.64.24.63.73.01.4

-3.15.36.3

ExportsOut-Turn

4.7'5.22

4.6'4.82

2.5'1.42

1.63

4.4

4.85.44.55.22.90.6

-4.75.87.07.2

1. Less developed countries.2. Non-oil developing countries.


Table B8. Developing Countries: Consumer Prices:(AmiuaJ Pert-cm Change)

Current Year Forecasts and First Available Out-Turns

Africa

Forecast Out-Turn

1977

1978

19791980

1981

1982

19831984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

13,4

19.4

20.019.7

17.4

24.3

12.8

15.7

16.1

13.1

11.6

12.7

15.113.4

21.9

25.8

16.8

30.6

27,4

21.4

20.1

19,4

22,715.8

19.5

17.8

13.6

14,8

16.2

18,8

20,2

15.6

22,5

40,2

31,7

33.6

Asia

Forecast Out-Turn

6,94.99.4

11.4

7.17.74.34.95.55.45.47.6

10.0

7.49.18.67.47.9

8,46.8

/0.3

12.2

9.96.85.96,96.75,98.7

14.6

11.97.99.47.49.5

13.5

Europe

Forecast Out-Turn

n.a.

n.a.

n.a.

27.0

24.521.2

20.2

23.022.4

25.2

22.7

31.3

50.3119.9

60.8

636.8

n.a.

n.a.

n.a,

n.a.

29.5

40.3

25.9

23.8

23.328.0

27.9

27.4

30.3

49.3

169.8166.4

90.9

n.a,

n.a.

n.a.

Middle East

Forecast Out-tum

77.9

15.4

19.835.0

33.8

31.8

35.9

47,2

17.3

11.9

11.1

14.1

14.614.2

13.7

14.8

n.a.

n.a.

17.5

17.5

26.844.4

32.8

37.6

40.3

16,5

13.8

1 1 . 1

16.3

18.8

13.9

13.3

16.0

n.a.

n.a.

n.a.

West Hemisphere

Forecast Out-Turn

n.a.

n.a.

n.a.47,661.4

61.1

90.2

91.6113.7

76.0

97.7

177.5

154.9312.2

122.9

140.0

150.9

213.9

n.a.

n.a.

48.7

60.2

65.7

78.0

122.7

119.8

144.0

86.9

130.8

277,6

531.0

768.0

162.5

169.9

236.5

225.8

Nonfucl ExportsForecast Out-Turn

27.6'

20.91

25.4*

29.<?

28.12

29 A2

34.02

34.5*

45.0

30.6

33.8

54.0

58.1

95.9

49.7

187.6

40.0

49.1

32.2'

25.6'

29.4*

32.52

31.4'

32,82

44.12

47.1

53,0

33.1

43.5

82.5

148.6

167,8

69.0

46,3

55.1

57.4

1. Less developed countries,2. Non-oil developing countries.


Table B9. Developing Countries: Balances of Payments on Current Account: Current Year Forecasts and First Available Out-Turns(in billions of U.S. dollars)


19711972197319741975197619771978I9794

198019811982198319841985198619871988198919901991199219931994

n,a.n.a,n.a.n.a,n.a.n.a.

-4.9-7.6•8.5

-10.1•12.9• 13.0•13.4

-9.4-9.8

-11.6•67•6.7-7.8-7.7-6.3-7.5-6.8-4.1

n.a.n.a.n.a.n.a.n.a.

-5.2-6.0-8.4-8.5

•11.2-13.3-13,2-10.8-10.9

-1.4-7.0-6.6-9.5-8.9-4.0-5.8-7.8-8.3

-12.6

AsiaForecast Out-Tum

n.a.n.a.n.a.n.a.n.a.n.a.

-3.3-4.0•8.0

•21.1•25.5•26.8•20.8

-8.2-8.2

-15.41.7

19.08.8

•0.8-19.2-17.4-27.3•28.3


-2.7-1.4-5.2

-J3.8-23.7-24.7-20.8-10.7

•7.9-15.4

2.123,010.0-0.2-3.7

-13.1-21.2-25.1-11.2


n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.

-10.8-9.2•6.2-4.2-2.5•271.3

-2.3-1.91.4

.1.2-77

-23.9n.a.n.a.

n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.

-9.3-10.3

-7.9-7,1-5.5-3.3-2.3-1.70.23.53.8

-2.3-10.0

n,a.n.a.n.a.

Middle EaslForecast Out-Tum


-8.4-7.3•8,5

-10.1-7.9

-12.3-12.2-11.9-16.1-367-18.1

-7.9-10.4

27•49.3-22.2

n.a.n.a.

n.a.n.a.n.a.n,a.n.a.

-6.6-6.1-5.8-8,1•7.7-9.0

-12.9-12.0-16.3-(0.7-24.9•2.9

-11.72.6

13.8-45.7

n.a.n.a.n.a.

Western HemisphereForecast Out-Tum


-7.9-11.6-13.0•21.6-39.9-35.7-217-18.5

-7.5•6.9

-75.9-70.5-77.5-72.5-12.3-22.4-33.4•46.6


-11.2-8.6

-11.6-18.6-33.1-41.5-34.9-18.5

-5.5-4.3

-16.1-9.3

-11.5-11.2-12.0-19.4-33.5-43.3-47.9

Nonfucl ExportsForecast Out-Turn

n.a.n.a.

-77.0'-20.02

-357*-J2.02

•25.02

•30.CP•38.0'-68.0*•97.5'-97.0'-67.8*-50.03

-35.3•20.3-20.5

0.9-6.3

•17.4-37.5-54.8-39.0-58,8

n.a.-6.8'-8.8*

-27.22

-37.02

-25.82

-22, 13

-31.05

-55.03

-82.13

-99.03

-86.83

-56.43

-38.2-27.6-11.9

5.41.3

-9.3-16.8-34.3-32.6-56.2-47.2

1. Less developed countries.2. Other developing countries.3. Non-oil developing countries.4. From February 1979 WHO (not June).


Table BIO. Developing Countries: Growth of Export Volumes: Current Year Forecasts and(Annual Percent Change)

AfricaForecast Out-Tum

1973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.3.86.84.96.64.33.11.63,63.85.62.82.74.0

•0.1

n.a.n.a.n.a.n.a.n.a.n.a.n.a.4.1

-3.80.43.16.26.12.2

-1.7

2.46.18.33,61.43.50.1

AsianForecast Out-Turn

n.a,n.a.n.a.n.a.n.a.n.a.n.a.n.a.6.89.14,28.67.46.26.18.69.89.43.2

10.011.810.1

n.a.n.a.n.a.n.a.n.a.n.a.n.a.9.07.81.19.5

14.02.5

17. i13.813.410.15.8

12.611.19.8

13.4


n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.6.25.03,94.57.24.13.24,03.21.42.6

-7.9n.a.n.a.


13.01.85.2

13.14.62.46.76.40.8

-9.0-13.9

n.a.n.a.n.a.

First Available Out-Turns

Middle EastForecast Out-Tum

n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.3.5

-1.08.85.93.02.9

•3.24.5

-2.03.1

-7.513.6n.a.n.a.

n.a,n.a.n.a.n.a.n.a.n.a.n.a.9.8

-1.3-4.7-5.3-0.4-5.916.6-0.812.27.00.6

-3,8n.a.n.a.n.a.

Western HemisphereForecast Out-Turn

n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.

12.18.25.46.94.8

-0.20.15.20.83.87.15.06.07.3

n.a.n.a.n.a.n.a.n.a.n.a.n.a.9.08.40.22.57.7

-1.2-8.85.2

11.23.03.23.24.38.09.4

NonfuelForecast

6.2'3.01

-l.O1

8.3*7.2}

7.2}

8.51

5.7*6.f6.4*4.7*7.1*7.64.33.96.97.07.23.55.4

10.39.6

ExportsOut-Turn

7.0'2.0'0,0'

12.73

6.7'8.01

8.3'7.9'3.92

0.82

5.32

12,03,47.7

10.010.87.33.54.49.59.7

12.2

1. Other developing countries.2. Non-oil developing countries.3. Less developed countries.


1. Less developed countries.2. Other developing countries.3. Non-oil developing countries.

Table Bl I, Developing Countries: Growth(Annual Percent Change)


19721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

n.a.n.a.n.a.n.a.n.a.n.a.n.a.

•0.30.71.20.72.8-0.3•0.6-02-4.03.30.52.4

•1.44.02.2

-6.5


-1.11.4

-4.5-4,1-7.80.8

-5.2-11.0

-4.81.32.30.1L93.30.4

-2,9

of Import

AsiaForecast Out-Turn


13.15.25.17.16.77.46.92.84.9

12.98.89.16.09.9

11.99.4


12,76.24.4

-0.98.87.16.34.39.4

17.813.05.5

14.712.011.213.1

Volumes: Current Year Forecasts and First Available Out-Turns

EuropeForecast Out-Tum

n.a.n.a.n.a.n.a.n.a.n.a.n.a.n.a.2.51.06.10.7

-0.95.05.32.88.03.49.8

14.14.7n.a.n.a.

n.a.n.a.n.a.n.a.n.a.n.a.n.a.7.60.71.5

-6.70.26.84.15.64,05.68,91,4

-15.1n.a.n.a.n.a.

Middle EastForecast Out-Turn

n.a.n.a.n.a.n.a.n.a.n.a.n.a.7.80.63.38.64.94.81.2

•13.3•12.5

-3.16.44.1

12.913.6

n.a.n.a.


10.1

1.37.0

-0.95.8

-5.5-11.4-18.3

-9.5-1.63.4

-7.32.5n.a.n.a.n.a.

West HemisphereForecast Out-Tum


•1.48.5

-4.7-5.59.38.232

-0.81.7

-2.6333.6

10.54.47.3

n.a.n.a.n.a.n,a,n.a.n.a.n.a.9.99.81.2

-21.7-19.0

1.5-1.3-4.82.67.32.87.3

15.618.08.0

13.7

Nonfuel ExportsForecast Out-Turn

n.a.7.3'4.5*

-7.0*3.0'

10.51

7.5'6.52

4.3}

5.2s

3.2}

2.33

5.5}

5.65J3.3

10.16.27.36.68.4

10.38.4

1.7'9.42

12.02

-4. 13

1.4'4.9'9.02

8.23

5.8'2,23

-7,73

-0.63

5.93.33.97,3

12.39.63.15.8

10.811.411.2


Table B12. Developing Countries: Nonfuel Commodity Prices: Current Year Forecasts and First Available Out-Turns(Annual Percent Change)

AgriculturalForecast

19811982198319841985198619871988198919901991199219931994

-4.6-7.14.98.60.93.76.48.12.3

-2.00.74.55.66.7

Raw MaterialsOut-Tum

•9.7-13.7

9.44.0

-12.2-1.029.48.2

-0.7-3.4-0,6

1.80.5

11.5

BeveragesForecast Out-Turn-22.6

7.82.36.1

•7.036.6

-21.68.81.3

-18.43.13.0

11.07.6

-22.32.57.6

14.7-H.616.2

-28.70.2

-17.0-13.3

-6.8-12.7

6.974.6

FoodForecast Out-Turn

-7.2-16.3

4.28.6

-4.53.4

-1.511.80.3

^4.6-1.40.9

-0.39.2

-13.9-21.0

8.8-1.5

-18.7-11.6

7.325.3-0.2-6.6

1.6-0.8-2.59.6

Minerals and MetalsForecast Out-Turn

-11.2'0.8'9.6'4.4*3.6'].!'3.79.22.0

•18.3^6.0-4.6-1.3•0.9

-14.0'-9,2'0.0'

-6.3'-2.8'

-10,019.940.35.4

-8.3-9.3-2.4

-15.610.8

NonfuelForecast

-10.2*-5.7*6.3*7J3

-2.32

12.3-5.410.00.3

-9.3-2.70.82.55.6

ExportsOut-Turn

-14.92

-13.32

7.92

2.72

-12.3-1.42.2

18.7-0.8-7.0-4.1-2.5-4.320,9

1. Metals only.2. Non-oil developing countries.


Table Cl. Industrial Countries:(Annual Percent Change)

Real GDP: Year Ahead Forecasts and First Settled Estimates (F,S.)

United States Japan Germany3

Forecast F.S. Forecast F,S. Forecast F.S,1973 6.3 5.9 11.0 10.5 5,0 5,31974 2.8 -2.1 9.0 -1.8 3.0 0.41975 -7.5 -1,8 3.3 2.1 2.2 -3,21976 6.2 6.0 5.4 6.1 J.5 5,71977 5.2 4.9 6.6 5.4 4.8 2.61978 4.5 4.4 4.5 5.6 3.4 3.41979 5.4 2.3 4.9 6.0 4.0 4,41980 0.0 -0.2 4.5 4,3 J./ 1.81981 -7.5 1.9 4.0 2.9 2.0 -0,31982 0.7 -1.9 5.7 3.0 2.0 -1.11983 2.1 3.7 JJ 3,0 2.0 1.31984 4.3 6.8 4.0 5,8 2,0 2.61985 4.0 2.1 4.1 4.5 2.4 2.51986 3.3 2.9 4.0 2.4 3.1 2.51987 J.5 3.4 2,9 4.2 3.0 1.81988 2.7 4.4 3.4 5.7 2.1 3.61989 2.5 2.5 4.2 4.9 1.9 3.91990 2.1 1.0 4.7 5.6 3.0 4.51991 1.7 -1.2 3.7 4.4 3,3 3.11992 3.0 2,6 3.4 1.3 2,0 1.61993 3.1 3,1 3.8 0.1 1.9 -1.71994 2.6 4.1 2.0 0.5 7.2' 2.94

FranceForecast

6.05,23,53.43.43.43,22,22,02.22.51.01.71.72.21.82.43.03.02.42.71.1

F.S.6.03,8

-1.25.23,0333.21.20.41,60,71.31,32.22.23.53.62.81.21.4

-1,02,9

ItalyForecast F.S.

4.5 5,96.0 3.40,9 -3.71.4 5.6

-0.5 1.73.0 2.64,0 5.02.8 4.00.4 -0.27.5 -0,33.3 -1.22.4 2.62.5 2.32.6 2,72.5 3.12.3 3.92.4 3,22.9 2.02.7 1.42.5 0,97.5 -0.77.7 2.2

United Kingdom CanadaForecast F.S.

5,0 6.03.7 0.12.7 -1.97.4 2.12.3 1.72.9 3.42.7 1,7

-0.3 -1.7-0.4 -2.20.6 1.11.6 3,22.7 2.62.4 3.32,2 3.02.5 4.42.3 4.22.5 2.22.7 0.87.3 -2.22.4 -0.52.7 2.02.8 3.8

Forecast6,05.02.55.74.04.54.03.00.87.42.74.23.72.43.63,23,22.07.73.84.43.8

F.S.6.82,80.64.92.73.42.70.03.1

-4.43.35.04.03.34.05.03.00,5

-1.70.72,24,6

AH Industrial SummitForecast F.S

6.3 6.34.3 -0.20,9 -1.44.6 5.34.6 3,84.0 4.03.7 3.4L9 1.30.7 1.22.0 -0.32.4 2.63.3 4,93.4 3.03.7 2.73,7 3.32.6 4.42.8 3.42.9 2.62,4 0.62.8 1.72.9 1.32.2 3.1

. Forecast6.6'4.3'0.7'4.9'4.74.23.51.80.62.02.43.53,53.23.22.72.93.02.42.93.02.3

SevenF.S.6,6'

-0,5'-1.35.64.04.23.51.21.3

-0,42.75.13.02.83.44.53,32,60.61.81,43.1

Kuropc2

Forecast5,04.32J2.73.23.23.42,21.21.82.31.92.42.52.42.02.32.S2.62.22.31.6

F.S,5,52,0

-2.34.42,22.93.31.5

•0,30,21.32,32,42.62.73.63.42.70.8l . t

-0.32.8

1. Constructed from country detail using GNP weights.2. Definition of "European" was changed from "European countries" to "European Community" in 1992 and subsequently to "European Union" in 1994.3. Germany is West Germany unless otherwise stated.4. For whole Germany.


Table C2. Industrial Countries: Inflation (GDP Deflator): Year Ahead Forecasts and First Settled Estimates (F.S.)(Annual Percent Change)

United StatesForecast F.S.

1973197419751976197719781979198019811982198319841985198619871988198919901991199219931994

JJ5.69.85,95.46.07,89.18.17.25.64.24.43.73.43,84.14.64.23.72.92.7

5.610.39.35.35.97.38.99,09.46.03,83.83,42.63.33.34,14.14.12.92.22.1

JapanForecast F.S.

6.012.216.16.27.04.02.27.54.53.63.02.21.20.91.21.71.41.32.12.61.91.3

12.320,8

7.16.35.74.82.03,12.92.10.70.61.71.8

-0.20.41.51.91.91.81.00.2

Germany3

Forecast F.S.5.57.07.04.53.74.03.75.24.54.54.53.02.62.31.62.22.22.53.63,73.73.14

5.96.88.13.33.63.93.95.14.34,83,21.92.23.12,11.52.63,44.54.43.22.34

FranceForecast F.S.

5.07.2

11.810.88.28.78.99.7

10.513.011.8

7.65.45.23.43.02.22.83.22.92.82.2

7.311.613.99.78.79.9

10,411.513.612.59,77.35.84.62.83.13.52.82.82.32.31.3

ItalyForecast F.S.

8.09.0

16.515.021.012.513.314.113.819.517.014.811J7.95.45.35.05.15.75.54.74.2

!0,316.617.517.818.313.315.120.317.617.515,010.78.88,05.66.06.37.57.34.74.43.6

United Kingdom CanadaForecast F.S. Forecast F.S.

7.07.4

20.115.212.510.610.214.313.79.07.66.05.04.23.74.84.75.75.74.23.33.9

7.512.227.915.213.110.214.418,912.58.05.44.26.13.54.86.76,96.86,94.43.42.1

4.86.5

11.39.07.26.36.97.48.19.87.04.74.74.13.53.53.64.45.12.72.01.5

7.613.910.89.56.96.3

10.310.610.110,15.42.83.43.04,34.14.93.02.71 , 11.10.6

Ail IndustrialForecast F.S.

4.77.2

11.67.66.96.37.08.68.27.96.85.04.43.73.13.33.33.74.03.73.12.7

7.211.911.07.26.97.07.88.98.87.35.04.13.83.32.93.23.94.14.13.12.51.9

Summit SevenForecast F.S.

4.6'7.2'

11.6'7.6'7.0'6.37.18.98.17.86.54.84.13.53.03.33.23.63.83.53.02.6

7.21

12.1'10.81

7,2'6.97,17,89.08.86,84.63.73.63.02.72.93.73.83.93.02.31.7

Europe2

Forecast F.S.

6.27.4

12.09.68.57.46.88.79.19.79.06.85.44.63.43.53.33.94.74.13.93.4

7.510.514.09.28.57.78.6

10.910.19.67.35.75.24.73.54.04.75.25.54.53.72.7

1. Constructed from country detail using GNP weights.2. Definition of "European" was changed from "European countries" to "European Community" in 1992 and subsequently to "European Union" in J994.3. Germany is West Germany unless otherwise stated,4. For whole Germany.


Table C3. Industrial Countries: Balances of Payments on Current Account:1'2 Year Ahead Forecasts and First Settled Estimates (F.S.)(in billions of U.S. dollars)


197319741975197619771978I9794

198019811982198319841985198619871988198919901991199219931994

-4.18.7

-3.56.1

-3.0^13.5-4.03.0

2L5•2,00.5

-54.0•109.0-141.0-123.0-140.6-128.7-138.7

-99.7-92.0-54.5

-130.0

2.52.1

14.61.7

-12.5-10.7

2.78.49.0

-5.8-35.5-93.0

-117.7-141.4-154.0-126.5-110.0

-92,1-3.7

-66.4-103.9-151.2

JapanForecast F.S.

6.3-0.90.3

-1.52.0

11.011.6-0.5

-11.09.5

17.026.041.557.074.183.280.989.755.859.4

100.9141.3

O.i-4.5-0.43.9

10.916.8-8.0-9.56.28.1

22.236.449.285.887.079.657.235.872.9

117.6131.4129.1

Germany3

Forecast F.S,0.93.6

12.37.48.07.0

10.55.0

-5.5-1.011.513.012.024.325.532.54L556.838.4

9.4•8.6

-29.9

7.012.37.57.27.9

13.10.6

-9.0-1.09.99.8

13.213.337.245.04S.655.447.9

-19.8-25.1-20.1-20.6

FranceForecast F.S,

1.9-0.2-5.3-1.7-3.3-2.03.41.0

•4.5-3.0-8.00.02.64.26.5

•4.0-2.8-4.9•4.9-8.0-0.43.1

0,3-4.8

1,1-5.1-2.15.33.0

-6.3-6.6-9.4-1.8

1.1-0.13.4

-4.1-4,3-3,9-8.4-5.92.8

10.59,7

ItalyForecast F.S.

3.4-2.0-4.52.00.62.55.95.0•1.0-6.5•1.5-3.0-1.8-7.63.0

-1.9-4.2

•10.2-11.7-16.1•33.4•12.8

-1.3-6.60.9

-1.74.08.96.1

-9.4-7.5-4,91.0

-2.8-4.24.4

-1.0-5,2

-10.5-14.5-21.1-26.6

11.415.5

United Kingdom CanadaForecast F.S, Forecast F.S.

0.1-3.5-6.6•2.9-0.76.05.06.04.58.52.04.54.24.9

•1.3-3.5

-19.0-26.0-21.0-12.3•19.2-24.8

-2.1-7.8-2.8-0,82.74.1

-0.610.617.710.47.53,74.9

-1,4-2.6

-26.0-31.3-25,7-11,2-15,2-15.5

-2.6

-0.9-0.4-3.9-3.9-3.8•4.0•3.4-6.0-5.0-2.5-0.53.50.0

-1.1-2.9-9.2

-11.9•16.5-16.9•14.6•20.8-16.9

-0.5-1.8-4.8-4.3-3.9-4.7-4.4-1.9-4.82.21.32.2

-0.4-6.7-8.0-8.4

-14.1-18.9-25.5-22.9-23.8-16.3

All IndustrialForecast F.S,

9.47.3

•12.57.11.84,0

25.48.0

-25.0-23.013.0

•11.5•46.2•51.1-11.6-44.2-51.4-69.6

-100.8-96.5-35.6-51.9

9,6-10.717.8-1.4-0.731.1

-14.1-46.3

-3.7-3.62.8

-35.2-53.5-18.1-42.9-54,4-82,4-97.5-23.5-39.019.3-6.3


7.65.3

-11.25.5

-0.27.0

29.013.5-1.03.0

21.0-100.0

-50.5-59.3-18.0-43.4•44.2-50.6-60.5•74.4-36.1-70.1

6.0-11.1

16.10.97.0

32.8-0.6

-17.113,010.54.5

-39,2-55.1-18,6-37.6-42.2-57,2-75,8-14,3-35.8

-9,9-36,4

1. The forecasts through 1985 and the first settled estimates through 1983 are those which exclude official transfers.2. The forecasts after 1985 and the first settled estimates after 1983 are those which include official transfers.3. Data through June 1990 apply to West Germany only.4. From February 15, 1979.


Table C4. Industrial Countries: Growth of Export Volumes: Year Ahead Forecasts and First Settled Estimates (F.S.)(Annual Percent Change)


197319741975197619772

1978*19791980'1981s

19826

19837

19848

1985198619871988198919901991199219931994

13.610.0-1.45.56.6639.27.9

-1.4•8.5•7.21.33.63.3

11.514.610.8

7.16.04.57.43.5

23.77.9

-2.53.60.5

10.29.68.3

-3.3-12.0

-6.88,30.88.0

13,223.510.98.87.26.95.89.0

JapanForecast F.S,

(0.010.98.97.69.53.0

-1.49.07.04.00.96.05.55.5

-8.2-0.33.17.46.05.34.9

-0.1

5.616.12.3

22.15.3

-0.9-1.317.610.5-3.08.7

15.74.4

-1.30.44.33.85,52.51.6

-1.15.1

Germany1

Forecast F.S.

10.08.13.65.1

W.O4.25,94.73.56.51.64.57.05.33.12.13.66.02.85.54.11.2

17.612.6

-10.612.55.64.17.44.05.31,90,49.66.41.22.97.47.74.50.2

-2.7-1.87.5

FranceForecast F.S.

12.09.66.64.89.06.45.5432.23.04.04.94.53.72.53.44.56.96.85.15.84.0

10.710.8-3.59.36.66.18.83.25,2

-3.94.05.21,7

-0.21.28.7

10.24.94.74.7

-2.55.9

ItalyForecast F.S.

11.316.56.06.1633.66.02.02.05.52.53.44.04.03.03.54.05.56.05.14.44.8

4.37.42.3

12.66.0

11.17,1

-7,94.70.35.16.57.51.63.45.79.03.50.73.98.5

10.9

United KingdomForecast F.S.

8.010.84.83.97.43.35.42.9

•3.44.01.74.34.01.82.02.21.54.05.74.85.64.2

11.66.7

-4.09.78.22.63.61.8

-1.40.81.47.06.83.75.7

-0,65.56.61.73.32.68.2

CanadaForecast F.S.

10.05.41.3

10.27.15.14.1

•0.52.5

•1.53.17.87.70.85.12.53.54.03.05.25.35.9

9.0-4.27.0

11.39.99.52.60.53.2

-0.27.4

22.25.94.36.7

10.01.34.5

1.08.8

10.714.2

All Industrial Summit SevenForecast F.S. Forecast F.S.

10.69.53.95.88.14.75.35.0V

3.5'°4.5"0.5"4.0"5.14.13.14.05.16.05.24.65.72.8

13.87.9

-4.610.74,45.59

6.5 '°4,4"2.9"

-1.8"2.09.74.32.65.38.97.05.92.93.52.48.1

n.a.n.a.n.a.n.an.an.a.n.a.5.21.61.00.04.35.44.02.64.45.46.35.25.25.92.7

n.a.n.a.n.a.n.a.n.a.n.a.n.a.5.13.0

-3.61.1

10.24.02.65.19.67.35.82.73.12.28.2

1. Data through June 1990 apply to West Germany only.2. From March 3, 1977.3. From April 4, 1978.4. From May 1980.

5. From June 1981.6. From April 1982.7. From May 1983.8. From April 1984.

9. From August 30, 1979.10. From August 22, 1980.11. From August 24t 1981.12. From August 2, 1982.13. From August 19. 1983.


Table C5. Industrial Countries: Growth of Import Volumes: Year Ahead Forecasts and First Settled Estimates (F.S.)(Annual Percent Change)


197319741975197619772

19783

197919804

1981s

19826

19837

19848

1985198619871988198919901991199219931994

7.11.0

•0.314.59.63.42.3

-2.50.41.03.7

15.211.15.83.91.28.57.21.4726.44.9

12.6-0.8

-13.222.613.47.42.1

-6.00.6

-4.410.022.8

4.514.85.57.05.83.30.6

10,912.813.4

JapanForecast F.S.

20.814.04.5

10.012.48.27.9

-3.51.42.0233.85.53.77.95.94.47.74.66.65.16.2

27.70.6

-13.212.03.96.3

11.6-7.4-2.0-0,1

1.310.40,47.29.1

16.77,85,82.8

•0.74.28.4

Germany1

Forecast F.S.11.76.06.96.5

11.06.08.94.20.02.00.94.04.75.44.84.54.06.48.23.83.01.2

7,4-1 .12.4

15.45.77.59.32.1

-3,50.75.45.54.55,95.46.77.1

12.814.4-4.3-5.97.1

FranceForecast F.S.

14.111.15.58.24.06.06.54.00.05.5

•0.70.01.53.65.74.84.65.76.64.25.73.3

13.64.4

-5.819.1

1.46,8

11.75.5

-1.23.1

-1.63.64.96.96.29.0

10.75.22.81.0

-5.96,8

ItalyForecast F.S.

13.011.0-1.04.0

•1.04.6

10.76.5

•4.05.51.54.06.55.58.05.44.96.55.64.74.03.3

13.6-4.8

-11.415.20.07.9

14.02.0

-10.02.01.09.28.94.4

10.87.08.34.54,83.5

-10.49.8

UnitedForecast

9.57.13.5L34.25.443

-0.8•2.210.04.26.24,74.05.54,52.91.72.53,84.55.0

KingdomF.S.

13.01.0

-7.25.31.44.4

11.6-5.2-4,75.07.1

10.35,56.47,4

12.87.81.2

-2.86.73.46.1

CunaduForecast F.S.

10.55.44.57.84.4

•0.21.52.03.5

•3.01,5

13.57.73.90.23.05.83.82.63.25.33.3

12.610.0-4.67.7I . I4.59.2

-4.12.4

-14.714.718.49.38.08.2

14.64.8

-0.42.36.5

11,010.5

All IndustrialForecast F.S,

11.27.63.6737.14.25.61.59

•0.2'°2.0"2.1n

6.6"6.54.74.93.65.45.64.54.55.13.4

11.71.0

-7,514.64.85,49

8.610

-0.8"-2.312

-0.5 »4.1

11.74.88.46.99.58.15.22.43.71.89.2



-0.22.52.57.97.04.85.03.75.76.04.1535.64.0

n.a.n.a.n,a.n.a.n.a.n.a.n.a.

-1.8-2.4-1.15.4

13.44.78.96.79.97.85.32.83.53.39.6

1. Data through June 1990 apply to West Germany only,2. From March 3, 1977.3. From April 4, 1978.4. From May 1980.

5. From June 1981.6. From April 1982.7. From May 1983.8. From April 1984.

9. From August 30, 1979,10. From August 22, 1980.11. From August 24, 1981,12. From August 2, 1982.13. From August 19, 1983.


- 66 - APPENDIX C

Table C6. World Trade Volumes and Terms of Trade:(Annual Percent Change)

Year Ahead Forecasts and First Sealed Estimates

World TradeForecast Estimate

197319741975W619771978197919801981198219S3198419S519861987198819S919901991199219931994

10.0*8.9'4.06.57.4>5.05.(f4.03.05.55.04.55.54.33.84.4

5.65.75.35.06.75.0

12.56.0

-4.6-n.o5.0:

5.07.02.00.5

-2.52.08.53.14.85.89.07.34.32.34.64.08.7

Industrial Countries' Terms of Trade-Forecast Estimate

0.4J

-2.9>r

L8-0.5-0.41.0s

03-M'-0.50.00.00.00.41.01.7

-0.60.10.0-0.2-0J-0.503

-2.5-11.2

2.9-0.51

-0.73.0'

-2.5-6.7-O.S2.42.2

-0.4

0.99.00.51.6

-0.2-0.4

1.61.51.60.4

1. Figures are computed as the simple average of export and import growth figures.2. World imports in volume.3. Figures are computed from those given on export and import unit values.


Table C7. Developing Countries: Real GDP; Year Ahead Forecasts and First Settled Estimates (F.S.)(Annual Percent Change)

AfricaForecast F.S.

1979198019811982198319841985198619871988198919901991199219931994

4.03.34.33.72.03.03.63.11.33.32.82.83.23.33.32.6

2.94.82.70.60.82.62.0O.H1.32.23.22,11.50.41.02.6

AsiaForecast F.S.

6.56.25.86.45.65.25.86.04.65.96.36.15.45.26.67.1

3.24.84.85.27.08.16.06.36.89.25.05.55.77.88.58.5

EuropeForecast F.S.

n.a.3.23.93.12.62.32.83.32.42.92.83.02.1

-3.3n.a.n.a.

4.51.72,32.30.93.52.23.22.51.2

-1.4-2,7n.a.n.a.n.a.n.a.

Middle EastForecast F.S.

6.75.36.58.35.54.24.62.6OJ2.71.83.03.7

IL2n.a.n.a.

1.66.64.93.53.81.7

-1.21.2

-0.53.53.70.7n.a.n.a,n.a.n.a.

Western HemisphereForecast F.S,

5.55.95.24,43.22.83.43.33.54.73.42.53.62,23.93.5

6.55.90.5

-1.8-2.83.13.74.4

2.50.71.6

-0.92.92,53.44.6

NonfuclForecast5.8'5.5'5.22

5.72

4,tf3.73

4.f4.63,94.94.64.24,21.75.35.8

ExportersF.S.

4.82

4.82

2.82

1.72

1,82

5.64.85.84.64.92.90.74.05,77.16.8



Table C8. Developing Countries:(Annual Percent Change)

Consumer Prices: Year Ahead Forecasts and First Settled Estimates (F.S.)

AfricaForecast F.S,

1979198019811982198319841985198619871988198919901991199219931994

16.019.116.316.124.613.614.024.213.810.812.013.713.213.618.622.6

19,819.022.716.426.519.813.113.715.819.720.015.427.141.332,632.9

AsiaForecast F.S.

5.48.18.67.16.14.85.36.05.65.75.99.47.68.78.17.8

10.012.010,66,15.87.17.47.88.8

14.611.78.09,07.59.7

13.5

EuropeForecast F.S.

n.a.17.922.222.118.317.220.418.821.522.825.343.319.023.6

n.a.n.a.

27.340.426.223.623.228.128.624.930.362.5

171.135.8

n.a.n.a.n.a.n.a.


18.024.934.931.229.833.261.413.0ll.l11312.114,014.112.2n.a.n.a.

26.744.134.737.139.213.911.711.714.717,813.712.9n.a.n.a.n.a.n.a.

West HemisphereForecast F.S.

35.636.944.555.356.857.461.586.532.264.989.4

108.430.155.087.6

162.9

48.860.265.677.9

112.7117.0145.786.5

131,2286.4538.2769,8163.2165.9236.4226.7

Nonfuel ExportersForecast F.S.

27.5'24.7*26.4*26.5*27.J2

25.1*27.1*37.015.125.934.845.417.025.538.442.3

28,52

32.42

31.82

32.92

41.42

46,054.433.543.788,6

149.5116.654,745,955,750.0



Table C9. Developing Countries:(in billions of U.S. dollars)

Balance of Payments on Current Account: Year Ahead Forecasts and First Settled Estimates (F.S.)

AfricaForecast F.S.

19781979198019811982198319841985198619871988198919901991199219931994

-7.5-5.5-8.0

-10.5-12.5-72.5-12.0

•9.6•7,7-6.4•6.1-7.8•7.5•83•7.2-7.4•7.0

-7.9-9.1

-12.1-13.4-12.4-10.2-11.6

-0.1-9.4-5.0-9.4-8.9-2.7-3.6-7.6-7.9

-11.2

AsiaForecast F.S.

•3.5•8.0

•15.5•25.0-25.5-25.0-17.5

-6.5-14.6

•9.110.08.22.8

-4.6^10.0-14.2-28.0

-4.8-13,4-23.7-21.7-20,3-11,5

-7.9-12.7

4.920.912.3-1,9-1.7-4,3-4.7

-24.64,0

EuropeForecast F.S.

n.a.n.a.n.a.

•11.0•9.0-4.0•4.0-3.0-2.1•1.8-2.5-0.42.10.4

-13.1n.a.n.a.

n.a.n.a.

-10.9-8.4-7.0-5.4-2.7-2.0-1,70.97.03.6

-23.6n.a.n.a.n.a.n.a.


•7.0-8.5

-13.0-10.0

•8.5•16.0-13.5-10.6-16.0•16.7

-7,0"13.0

•0.613.0

-18.8n.a.n.a.

-7.3•8.1-7.7

-10.5-11.2-10.2-14.9

0.5-23.3

-5.2-8.30.3

10.3n.a.n.a.n.a.n.a.


-77.5+13.0-7(5,5•27.5-47.5•J0.5"21.5-15.5

-8.0• 72.6-13.8-13.2-13.0-11.4-15.2-27.9-37.0

-12.0-19.8-32.5-42.4-36.0-15.5

-5.0-4.6

-17.6-11.3-10.7

-8,9-6.9

-19.9-34.4-45.8-46.8

NonfuclForecast

-mo'-38.0*-53.0'-80.0'

•100.0'-90.0'-67.5'-45.0'-39.2-22.5-14.6

-5.5•12.9-21.9-37.4-27.4-48.2

ExportsF.S.

-32.0'-56.11

-83.7'-101.01

-84,0'-52.61

-38.9-23.3

-9.24.3

10.0-9.8

-35.3-11.7-17.7-56.8-57.5

1. Non-oil developing countries.2, Less developed countries.


Table CIO. Developing Countries:(Annual Percent Change)

Growth of Export Volumes: Year Ahead Forecasts and First Settled Estimates (F.S.)

AfricaForecast F.S,

1977197819791980

19811982

19831984

19851986

1987

1988

19891990

1991199219931994

n.a.n.a.n.a.n.a.

3.81

6.82

4.9s

6.64

3.84.22.93.54.74.00.02.05.50,0

n.a.n.a.

n.a.n.a.

-3.82

0.43

3.14

7.23.75.2

-2.9

3.42.76,12.22.13.6

-0.4

AsiaForecast F.S,

n.a.n.a.n.a.n.a.

6.8'9.124.23

8.64

6.66.96.36.89.77.87.09.0

11.711.5

n.a.n.a.

n.a.n.a.

7.81

M3

9.5'13.93.8

17.216.6

13.1

9.27.1

12.311.210.113.9

EuropeForecast F.S.

n.a.n.a.

n.a.n.a.

6.2'5,tf3.9s

4.54

5.26.05.94.55.72.43.0

-2.0n.a.n.a.

n.a.n.a.

n,a.n.a,

13.02

1.8'5.24

13.0

4.90,17.85.80.8

-19.8n.a.n.a.n.a,n.a.


n.a.n.a.

n.a.n.a.

3.5'•l.tf8.8*5.9<7.73.33.28,62.93.13.77.3n.a.n.a.

n.a.n,a.n.a.n.a.

-I.31

-4,73

-5.34

0.4-5.4

15.2-1.3

14.6

6.2-1.0n,a.n.a.n.a.n.a.


n.a.n.a.

n.a.n.a.

12. /'8.2*

5.4*6#6.93.15.77.32.75.65.35.76.46.7

n.a.n.a.

n.a.n.a.8.43

0.23

2,54

8.2-1.7-7.0

6.58.54.84.74.08.57.98.3

NonfuelForecast

7.57

7.5*7.2'

6Sf8.07

8.07

5.57

5.57

6.27

6.15.66.47.66.55.65.8

10.69.7

ExportsF.S.

5.0s

8.06

9.57

8.07

4.67

0.47

5,87

12.03,47.7

12.1

10.7

6.4-1.89.59.8

10. 112.7

1. From June 1981.2. From April 1982.3. From May 1983.4. From April 1984.5. Less developed countries.6. Other developing countries.7. Non-oil developing countries.


Table Cl 1. Developing Countries:(Annual Percent Change)

Growth of Import Volumes: Year Ahead Forecasts and First Settled Estimates (F.S.)

AfricaForecast F.S.

1977

1978

1979198019811982198319841985198619871988198919901991199219931994

n.a.n.a.n.a.0.71.21

0.7*2.81

-0.34

1.9•03-4.32.92.51.3

-yj2.42.6

-2.5

n.a.n.a.n.a.1,4'

-4.5'-4. 13

-7,84

2,5-7.0-6.7

-5.36.1

-0.5-0.2-3,04.8

-0.92.7

AsiaForecast F.S.

n.a.n.a.n.a.5.25.1'7.1*6.73

7A4

6,54.61,78.2

11.27.26.18.7

11.912.3

n.a.n.a.n.a.6.2'4.42

•0.93

8.84

5,95.64.8

11.817.313.67.3

12.5

12,212.912.1

EuropeForecast F.S.

n.a.n.a.n.a.2.51.0'6.13

0.7*-0.9*4.35.66.58.86.87.14,4

-3.8n.a.n.a.

n.a.n.a.

n.a.0.7'1.52

-6.73

0.2<5.94.81.55.62.09,8

-1.0n.a.n.a.n.a.n.a.


n.a.n.a.n.a.0.63.3'8.6*4.9*4.8*6.2

-2.9-8.30.32.64.6

11.912.4n.a.n.a.

n.a.n.a.n.a.1.31

7.02

-0.95

5.8<-5.9

-14.0-19.3

-8.70,72.3

-2.6n.a.n.a.n.a.n.a.

West HemisphereForecast F,S.

n.a.n.a.n.a.

-1.48.5'

.4.7*-5.5'9.34

6.95.07,94.76.66.38.75.28.15.8

n,a,n,a.n.a.9.8'1.2'

-21,7*-19.0'

2.20.6

-1.5

3.64.51.56.3

16.9

16.98,9

11.7

Nonfuel ExportsForecast F.S.

6.47

7.0s

4.8?4.(f4.57

6.07

4.07

5.07

5.77

3.73.27.38.96.74.65.7

10.710.2

4.96

8,03

10.07

6,07

2.97

-8,27

-1.87

5,03.44.08.9

11.99.33.2

10.011.213.010.7

1. From June 1981.2. From April 1982.3. From May 1983,4. From April 1984.5. Less developed countries.6. Other developing countries.7. Non-oil developing countries.


Table C12. Developing Countries:(Annual Percent Change)

Nonfuel Commodity Prices: Year Ahead Forecasts and First Settled Estimates (F.S.)

Agricultural Raw MaterialsForecast

198I1

19822

19833

19844

1985198619871988198919901991199219931994

1. From June 1981.2. From April 1982.3. From May 1983.4. From April 1984.5. Metals only.

-4.6-7.14.98.63.62.14.30.4

^0.62.12.72.65.96,5

F.S.

-9.7-13.7

9.44.0

-12.0-1.029.4

8.2-0,7-3.5-0.61.80.5

10,1

BeveragesForecast

-22.67.82.36.1

•3.80.4

-10.77.90.4

-8.58.78.8

I O.I11.3

F.S.

-22.32.57.6

14.7-11.6

16.2-28.7

0.2-17.0-13.3-6.8

-12.76.9

74.9

FoodForecast

-7.2-16.3

4.28.62.7

-0.30.57.7

-0.4-5.63.72.70.71.6

F.S.

-13.9-21.0

8.8-1.4

-18.6-11.6

7.325,30.4

-6.91,6

•0.8-2.55,1

Minerals and MetalsForecast

-11.25

0.8s

9.6*4.4*7.8'5.2*3.tfa/5

-14.3s

-19.5•10.8

-0.92.73.8

F.S.

-14.05

-9.25

O.O5

-6.35

-3.5s

-9.95

19.9*40.3

5,4-8.1-8.6-2.5

-15.616.6

NonfuelForecast

•I 0.2*•5.7*6.3*7.16

i.r1.1

-L64.4

-3.3-7.0-0.51.43.35.0

ExportsF.S.

-14.9*-13.36

7.96

2.9-12.4

-1.32.2

19.3-0.7

-6.9-3.9-2.5-4.318.3

6. Non-oil developing countries.


- 73 -

References

Artis MJ. (1982), "Why Do Forecasts Differ?" Papers presented to the Panel of AcademicConsultants, 17 (Bank of England).

Artis MJ. (1988), "How accurate is the World Economic Outlook? A Post Mortem onShort-Term Forecasting at the International Monetary Fund," Staff Studies for theWorld Economic Outlook (Washington, D.C.: International Monetary Fund, July), pp.1-49.

Artis MJ. and W. Zhang (1990), "BVAR Forecasts for the G-7," International Journal ofForecasting, pp. 349-362.

Ashley R. (1988), "On the Relative Worth of Recent Macroeconomic Forecasting,"InternationalJournal of Forecasting^ 4, pp. 363-376.

Ashley R., Granger C.WJ. and R. Schmalensee (1980), "Advertising and AggregateConsumption: An Analysis of Causality," Econometrica, 48 (July), pp. 1149-1167.

Barrionuevo J.M. (1993), "How Accurate are the World Economic Outlook Projections?"Staff Studies for the World Economic Outlook (Washington D.C.: InternationalMonetary Fund: December).

Daniel W.W. (1978), Applied Non-Parametric Statistics (London: Houghton-Mifflin).

Diebold, F.X., and J.A. Lopez (1996), "Forecast Evaluation and Combination," NBERTechnical Working Papers, 1992.

Granger C. and P. Newbold (1977), Forecasting Economic Time Series (New York:Academic Press).

Holden K. and D.A. Peel (1990), "On Testing for Unbiasedness and Efficiency of Forecasts,"Manchester School of Economic and Social Studies, 58 (June), pp. 120-127.

Kenen P.B. and A.B. Schwarz (1986), "An Assessment of Macroeconomic Forecasts in theInternational Monetary Fund's World Economic Outlook," Working Papers inInternational Economics (Princeton University, International Finance Section).

Kolb R.A. and H.O. Stekler (1993), "Are Economic Forecasts Significantly Better ThanNaive Predictions? An Appropriate Test," InternationalJournal of Forecasting., 9,pp. 117-120.


untitled - imf elibrary

Documents