The Economist - USA (2020-08-08)

TheEconomistAugust 8th 2020 Finance & economics 59

M

ostpeople, whenpresentedwithbad

news,tendtoplayitdown.Evenpro-

fessionaleconomicforecastersarenotim-

munetothetemptationsofhope.InFebru-

arymorethan500mpeopleinChinawere

experiencingsomeformoflockdown,and

covid-19hadspreadtoItaly.Yettheimfsaid

thatinitsbase-caseforecastglobalgdp

growththisyearwouldbeonly0.1 percent-

agepointslowerthanpreviouslyexpected.

ByAprilithadcutitsforecastby6.2per-

centagepoints,to-3%.ByJuneit hadsawn

off another1.9 percentagepoints. Justa

weeklateraninformalpollofabout 40 imf

stafffoundthattwo-thirdsexpectedanoth-

erdownwardrevisioninOctober.

Byandlarge,economicforecastersarea

sunnybunch.Theyrarelypredicta down-

turn.Humannature,incentivesandpoliti-

calpressuregetintheway.Yetrosyfore-

castsbytheimfandtheWorldBankcan

haveseriousconsequences.Thatisespe-

cially the case inpoor countries today,

wherecovid-19isravagingeconomies,and

governments,internationalorganisations

and investors are using forecasts to guide

their decisions.

imfand World Bank projections can be

very influential in some countries. They

can affect governments’ spending and bor-

rowing plans. Investors may lend more

cheaply to countries expected to grow rap-

idly. And the forecasts determine whether

the fund and the bank think a country’s

debt is sustainable, which in turn deter-

mines whether it qualifies for a bail-out.

The fund tends to be optimistic. Its one-

year-ahead growth forecasts for develop-

ing countries in 1990-2016 were, on aver-

age, 0.42 percentage points above subse-

quently published gdpfigures. Most of the

optimism comes from failing to predict

downturns. Even once a recession has be-

gun, forecasters are still slow to accept the

news (see chart).

Such errors can drastically change debt

dynamics. Take a country expected to have

public debt of 50% of gdp in 20 years’ time.

If annual economic growth is 0.5 percent-

age points less than predicted, and nothing

else changes, then the debt ratio could in-

stead be 90% of gdp. In a recent study Paul

Beaudry of the University of British Colum-

bia and Tim Willems of the imfeven link

over-optimism to future fiscal crises. They

find that overestimating average annual

growth by a percentage point for the next

three years, as the imfdoes about 40% of

thetime,reducesgrowththreeyearslater

by a full percentage point. Governments

and firms seem to celebrate good forecasts

by racking up debt. Trouble sets in.

Predicting growth, and especially

downturns, is fiendishly hard. Getting it

right is not helped by forecasters having lit-

tle incentive to spot clouds on the horizon.

Analysts fear that gloom could become

self-fulfilling. Standing out from the crowd

and wrongly calling a recession damages a

forecaster’s reputation more than failing to

predict one along with everyone else. Then

there is “pushback from governments”,

says Maurice Obstfeld, who was the fund’s

chief economist in 2015-18.

Internal pressure to nudge up forecasts

in order to justify a lending package is also

“definitely an issue ”, says Mr Obstfeld. A

paper by Giang Ho and Paolo Mauro of the

fund in 2014 found that forecasts were es-

pecially optimistic when countries were

just about to enter a programme. The fund’s

Independent Evaluation Office (ieo) ac-

knowledges that forecasts tend to be rosy

in high-profile bail-outs, but notes that

these are usually corrected at the pro-

gramme’s first review three months later.

(By then, of course, the agreement has al-

ready been signed.)

Some economists at the fund are more

optimistic than others, find Messrs Beau-

dry and Willems. In poor countries, less ex-

perienced economists tend to be less accu-

rate. Although oil and mineral discoveries

do not boost growth immediately, imf

forecasters consistently predict that they

will, according to research in 2017 by James

Cust of the World Bank and David Mihalyi

of the Natural Resource Governance Insti-

tute, a think-tank.

The fund’s lack of consistency attracts

criticism, too. Since the coronavirus pan-

demic began it has revised down growth in

rich countries in 2020 by three percentage

points more than that in developing ones.

That is odd, argued Justin Sandefur of the

Centre for Global Development, another

think-tank, and Arvind Subramanian of

Ashoka University, in June. Lockdowns and

social distancing are at least as severe in

poor countries as in rich ones. But fiscal re-

sponses have been much weaker, and, as

the fund itself has argued, capital outflows

and currency pressure are bigger threats.

Perhaps, the authors speculate, the fund

has been rosier about poorer countries this

time in order to avoid having to provide

support. Theimf says that China’s success

in containing the virus explains why its

forecasts are more positive for developing

countries and strongly denies lending in-

fluences its growth forecasts.

The other big official forecaster in poor

countries is the World Bank. (Private firms,

including The Economist’s sister organisa-

tion, the Economist Intelligence Unit, also

publish forecasts.) Over the past decade,

the bank has produced more accurate fore-

casts for Africa and the Middle East than

the fund, but done a worse job for Latin

America (though differences are small).

Overall, a comparison of the forecasts pub-

lished by both institutions every January

by Prakash Loungani of theieosuggests

that the fund is still better than the bank at

predicting downturns in poor countries.

Perhaps people should simply expect

less of forecasts, says Mr Obstfeld. They

may represent an expectation of the most

likely outcome, but the chance of them be-

ing bang-on is slim. “You are getting some-

thing that is useful,” he argues, but “in gen-

eral, you are not getting high accuracy.”^7

TheBrettonWoodsinstitutions’economicforecastsforpoorcountriesare

influentialbuttoooptimistic.Thatcanbecostly

TheIMFandtheWorldBank

Abandoninghope

Springing eternal

Average IMF forecast for GDP in poor countries

in recession years*, % change on a year earlier

Source:

IEOoftheIMF

*Emergingmarketsanddeveloping countries.

Averageof 360 recessions, 1990-2016

4

2

0

-2

-4

-6

-8

Apr Oct Apr Oct Actual

year of

recession

year before

recession

Forecast made in:

Looking on the bright side